id
stringlengths 9
16
| submitter
stringlengths 1
64
⌀ | authors
stringlengths 3
62.8k
| title
stringlengths 1
382
| comments
stringlengths 1
1.15k
⌀ | journal-ref
stringlengths 1
557
⌀ | doi
stringlengths 7
153
⌀ | report-no
stringlengths 1
509
⌀ | categories
stringlengths 5
125
| license
stringclasses 9
values | abstract
stringlengths 6
6.09k
| versions
listlengths 1
187
| update_date
timestamp[s]date 2007-05-23 00:00:00
2025-04-25 00:00:00
| authors_parsed
sequencelengths 1
2.83k
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2504.14300 | Hao Wang | Xinyu Liang and Hao Wang | Learning and Generating Diverse Residential Load Patterns Using GAN with
Weakly-Supervised Training and Weight Selection | 12 pages | IEEE Transactions on Consumer Electronics, 2025 | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | The scarcity of high-quality residential load data can pose obstacles for
decarbonizing the residential sector as well as effective grid planning and
operation. The above challenges have motivated research into generating
synthetic load data, but existing methods faced limitations in terms of
scalability, diversity, and similarity. This paper proposes a Generative
Adversarial Network-based Synthetic Residential Load Pattern (RLP-GAN)
generation model, a novel weakly-supervised GAN framework, leveraging an
over-complete autoencoder to capture dependencies within complex and diverse
load patterns and learn household-level data distribution at scale. We
incorporate a model weight selection method to address the mode collapse
problem and generate load patterns with high diversity. We develop a holistic
evaluation method to validate the effectiveness of RLP-GAN using real-world
data of 417 households. The results demonstrate that RLP-GAN outperforms
state-of-the-art models in capturing temporal dependencies and generating load
patterns with higher similarity to real data. Furthermore, we have publicly
released the RLP-GAN generated synthetic dataset, which comprises one million
synthetic residential load pattern profiles.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 13:50:49 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Liang",
"Xinyu",
""
],
[
"Wang",
"Hao",
""
]
] |
2504.14301 | Nazia Aslam | Nazia Aslam and Kamal Nasrollahi | Balancing Privacy and Action Performance: A Penalty-Driven Approach to
Image Anonymization | Accepted to CVPRW 2025 | null | null | null | cs.CV cs.AI cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The rapid development of video surveillance systems for object detection,
tracking, activity recognition, and anomaly detection has revolutionized our
day-to-day lives while setting alarms for privacy concerns. It isn't easy to
strike a balance between visual privacy and action recognition performance in
most computer vision models. Is it possible to safeguard privacy without
sacrificing performance? It poses a formidable challenge, as even minor privacy
enhancements can lead to substantial performance degradation. To address this
challenge, we propose a privacy-preserving image anonymization technique that
optimizes the anonymizer using penalties from the utility branch, ensuring
improved action recognition performance while minimally affecting privacy
leakage. This approach addresses the trade-off between minimizing privacy
leakage and maintaining high action performance. The proposed approach is
primarily designed to align with the regulatory standards of the EU AI Act and
GDPR, ensuring the protection of personally identifiable information while
maintaining action performance. To the best of our knowledge, we are the first
to introduce a feature-based penalty scheme that exclusively controls the
action features, allowing freedom to anonymize private attributes. Extensive
experiments were conducted to validate the effectiveness of the proposed
method. The results demonstrate that applying a penalty to anonymizer from
utility branch enhances action performance while maintaining nearly consistent
privacy leakage across different penalty settings.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 13:52:33 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Aslam",
"Nazia",
""
],
[
"Nasrollahi",
"Kamal",
""
]
] |
2504.14302 | Yogev Kriger | Yogev Kriger and Shai Fine | Learning to Score | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Common machine learning settings range from supervised tasks, where
accurately labeled data is accessible, through semi-supervised and
weakly-supervised tasks, where target labels are scant or noisy, to
unsupervised tasks where labels are unobtainable. In this paper we study a
scenario where the target labels are not available but additional related
information is at hand. This information, referred to as Side Information, is
either correlated with the unknown labels or imposes constraints on the feature
space. We formulate the problem as an ensemble of three semantic components:
representation learning, side information and metric learning. The proposed
scoring model is advantageous for multiple use-cases. For example, in the
healthcare domain it can be used to create a severity score for diseases where
the symptoms are known but the criteria for the disease progression are not
well defined. We demonstrate the utility of the suggested scoring system on
well-known benchmark data-sets and bio-medical patient records.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 13:53:38 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Kriger",
"Yogev",
""
],
[
"Fine",
"Shai",
""
]
] |
2504.14303 | Emil Abasov | E. Abasov, L. Dudko, E. Iudin, A. Markina, P. Volkov, G. Vorotnikov,
M. Perfilov and A. Zaborenko | Reconstruction of angular correlations in the associated top quark and
the dark matter mediator production | 14 pages, 10 figures | null | null | null | hep-ph | http://creativecommons.org/licenses/by/4.0/ | For the process of single top quark production within the "simplified model"
with a scalar dark matter mediator, a new variable based on angular
correlations was presented, for the proper reconstruction of which it is
necessary to separate the contributions of two undetectable particles: the
neutrino and the mediator. In this work, various machine learning approaches
for reconstructing the momenta of these particles are analyzed. A comparison is
made between the results obtained using a multilayer perceptron and the
Normalizing Flows architectures. The neural networks based on Normalizing
Flows, presented in this work, demonstrate a high quality of reconstruction of
the target variable and can be used for collider data analysis.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 13:55:46 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Abasov",
"E.",
""
],
[
"Dudko",
"L.",
""
],
[
"Iudin",
"E.",
""
],
[
"Markina",
"A.",
""
],
[
"Volkov",
"P.",
""
],
[
"Vorotnikov",
"G.",
""
],
[
"Perfilov",
"M.",
""
],
[
"Zaborenko",
"A.",
""
]
] |
2504.14304 | Alexandru Ionescu | Yu Deng, Alexandru Ionescu, and Fabio Pusateri | On the wave turbulence theory of 2D gravity waves, II: propagation of
randomness | 74 pages, 5 figures | null | null | null | math.AP math-ph math.MP | http://creativecommons.org/publicdomain/zero/1.0/ | This is the second part of our work initiating the rigorous study of wave
turbulence for water waves equations. We combine energy estimates, normal
forms, and probabilistic and combinatorial arguments to complete the
construction of long-time solutions with random initial data for the 2d (1d
interface) gravity water waves system on large tori.
This is the first long-time regularity result for solutions of water waves
systems with large energy (but small local energy), which is the correct setup
for applications to wave turbulence. Such a result is only possible in the
presence of randomness.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:02:26 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Deng",
"Yu",
""
],
[
"Ionescu",
"Alexandru",
""
],
[
"Pusateri",
"Fabio",
""
]
] |
2504.14305 | Chenjia Bai | Jiyuan Shi, Xinzhe Liu, Dewei Wang, Ouyang Lu, S\"oren Schwertfeger,
Fuchun Sun, Chenjia Bai, Xuelong Li | Adversarial Locomotion and Motion Imitation for Humanoid Policy Learning | null | null | null | null | cs.RO | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Humans exhibit diverse and expressive whole-body movements. However,
attaining human-like whole-body coordination in humanoid robots remains
challenging, as conventional approaches that mimic whole-body motions often
neglect the distinct roles of upper and lower body. This oversight leads to
computationally intensive policy learning and frequently causes robot
instability and falls during real-world execution. To address these issues, we
propose Adversarial Locomotion and Motion Imitation (ALMI), a novel framework
that enables adversarial policy learning between upper and lower body.
Specifically, the lower body aims to provide robust locomotion capabilities to
follow velocity commands while the upper body tracks various motions.
Conversely, the upper-body policy ensures effective motion tracking when the
robot executes velocity-based movements. Through iterative updates, these
policies achieve coordinated whole-body control, which can be extended to
loco-manipulation tasks with teleoperation systems. Extensive experiments
demonstrate that our method achieves robust locomotion and precise motion
tracking in both simulation and on the full-size Unitree H1 robot.
Additionally, we release a large-scale whole-body motion control dataset
featuring high-quality episodic trajectories from MuJoCo simulations deployable
on real robots. The project page is https://almi-humanoid.github.io.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:03:57 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Shi",
"Jiyuan",
""
],
[
"Liu",
"Xinzhe",
""
],
[
"Wang",
"Dewei",
""
],
[
"Lu",
"Ouyang",
""
],
[
"Schwertfeger",
"Sören",
""
],
[
"Sun",
"Fuchun",
""
],
[
"Bai",
"Chenjia",
""
],
[
"Li",
"Xuelong",
""
]
] |
2504.14306 | Yitao Zhao | Yitao Zhao, Sen Lei, Nanqing Liu, Heng-Chao Li, Turgay Celik, and Qing
Zhu | Exploring Generalizable Pre-training for Real-world Change Detection via
Geometric Estimation | Submitted to IEEE TGRS | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As an essential procedure in earth observation system, change detection (CD)
aims to reveal the spatial-temporal evolution of the observation regions. A key
prerequisite for existing change detection algorithms is aligned geo-references
between multi-temporal images by fine-grained registration. However, in the
majority of real-world scenarios, a prior manual registration is required
between the original images, which significantly increases the complexity of
the CD workflow. In this paper, we proposed a self-supervision motivated CD
framework with geometric estimation, called "MatchCD". Specifically, the
proposed MatchCD framework utilizes the zero-shot capability to optimize the
encoder with self-supervised contrastive representation, which is reused in the
downstream image registration and change detection to simultaneously handle the
bi-temporal unalignment and object change issues. Moreover, unlike the
conventional change detection requiring segmenting the full-frame image into
small patches, our MatchCD framework can directly process the original
large-scale image (e.g., 6K*4K resolutions) with promising performance. The
performance in multiple complex scenarios with significant geometric distortion
demonstrates the effectiveness of our proposed framework.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:05:39 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Zhao",
"Yitao",
""
],
[
"Lei",
"Sen",
""
],
[
"Liu",
"Nanqing",
""
],
[
"Li",
"Heng-Chao",
""
],
[
"Celik",
"Turgay",
""
],
[
"Zhu",
"Qing",
""
]
] |
2504.14307 | Muhammad Haseeb Aslam | Muhammad Haseeb Aslam, Clara Martinez, Marco Pedersoli, Alessandro
Koerich, Ali Etemad, and Eric Granger | Learning from Stochastic Teacher Representations Using Student-Guided
Knowledge Distillation | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Advances in self-distillation have shown that when knowledge is distilled
from a teacher to a student using the same deep learning (DL) architecture, the
student performance can surpass the teacher particularly when the network is
overparameterized and the teacher is trained with early stopping.
Alternatively, ensemble learning also improves performance, although training,
storing, and deploying multiple models becomes impractical as the number of
models grows. Even distilling an ensemble to a single student model or weight
averaging methods first requires training of multiple teacher models and does
not fully leverage the inherent stochasticity for generating and distilling
diversity in DL models. These constraints are particularly prohibitive in
resource-constrained or latency-sensitive applications such as wearable
devices. This paper proposes to train only one model and generate multiple
diverse teacher representations using distillation-time dropout. However,
generating these representations stochastically leads to noisy representations
that are misaligned with the learned task. To overcome this problem, a novel
stochastic self-distillation (SSD) training strategy is introduced for
filtering and weighting teacher representation to distill from task-relevant
representations only, using student-guided knowledge distillation (SGKD). The
student representation at each distillation step is used as authority to guide
the distillation process. Experimental results on real-world affective
computing, wearable/biosignal datasets from the UCR Archive, the HAR dataset,
and image classification datasets show that the proposed SSD method can
outperform state-of-the-art methods without increasing the model size at both
training and testing time, and incurs negligible computational complexity
compared to state-of-the-art ensemble learning and weight averaging methods.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:08:56 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Aslam",
"Muhammad Haseeb",
""
],
[
"Martinez",
"Clara",
""
],
[
"Pedersoli",
"Marco",
""
],
[
"Koerich",
"Alessandro",
""
],
[
"Etemad",
"Ali",
""
],
[
"Granger",
"Eric",
""
]
] |
2504.14308 | Yang Hu | Yang Hu and Jianzhou Liu and Wenlong Zeng | The Schur complements for $SDD_{1}$ matrices and their application to
linear complementarity problems | 26pages | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by/4.0/ | In this paper we propose a new scaling method to study the Schur complements
of $SDD_{1}$ matrices. Its core is related to the non-negative property of the
inverse $M$-matrix, while numerically improving the Quotient formula. Based on
the Schur complement and a novel norm splitting manner, we establish an upper
bound for the infinity norm of the inverse of $SDD_{1}$ matrices, which depends
solely on the original matrix entries. We apply the new bound to derive an
error bound for linear complementarity problems of $B_{1}$-matrices.
Additionally, new lower and upper bounds for the determinant of $SDD_{1}$
matrices are presented. Numerical experiments validate the effectiveness and
superiority of our results.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:13:06 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Hu",
"Yang",
""
],
[
"Liu",
"Jianzhou",
""
],
[
"Zeng",
"Wenlong",
""
]
] |
2504.14309 | Ruoyan Xiong | Ruoyan Xiong, Huanbin Zhang, Shentao Wang, Hui He, Yuke Hou, Yue
Zhang, Yujie Cui, Huipan Guan, and Shang Zhang | FGSGT: Saliency-Guided Siamese Network Tracker Based on Key Fine-Grained
Feature Information for Thermal Infrared Target Tracking | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Thermal infrared (TIR) images typically lack detailed features and have low
contrast, making it challenging for conventional feature extraction models to
capture discriminative target characteristics. As a result, trackers are often
affected by interference from visually similar objects and are susceptible to
tracking drift. To address these challenges, we propose a novel saliency-guided
Siamese network tracker based on key fine-grained feature infor-mation. First,
we introduce a fine-grained feature parallel learning convolu-tional block with
a dual-stream architecture and convolutional kernels of varying sizes. This
design captures essential global features from shallow layers, enhances feature
diversity, and minimizes the loss of fine-grained in-formation typically
encountered in residual connections. In addition, we propose a multi-layer
fine-grained feature fusion module that uses bilinear matrix multiplication to
effectively integrate features across both deep and shallow layers. Next, we
introduce a Siamese residual refinement block that corrects saliency map
prediction errors using residual learning. Combined with deep supervision, this
mechanism progressively refines predictions, ap-plying supervision at each
recursive step to ensure consistent improvements in accuracy. Finally, we
present a saliency loss function to constrain the sali-ency predictions,
directing the network to focus on highly discriminative fi-ne-grained features.
Extensive experiment results demonstrate that the pro-posed tracker achieves
the highest precision and success rates on the PTB-TIR and LSOTB-TIR
benchmarks. It also achieves a top accuracy of 0.78 on the VOT-TIR 2015
benchmark and 0.75 on the VOT-TIR 2017 benchmark.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:13:15 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Xiong",
"Ruoyan",
""
],
[
"Zhang",
"Huanbin",
""
],
[
"Wang",
"Shentao",
""
],
[
"He",
"Hui",
""
],
[
"Hou",
"Yuke",
""
],
[
"Zhang",
"Yue",
""
],
[
"Cui",
"Yujie",
""
],
[
"Guan",
"Huipan",
""
],
[
"Zhang",
"Shang",
""
]
] |
2504.14310 | Dailin Yang | Dailin Yang, Shuhang Zhang, Hongliang Zhang, Lingyang Song | End-Edge Model Collaboration: Bandwidth Allocation for Data Upload and
Model Transmission | null | null | null | null | cs.ET | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The widespread adoption of large artificial intelligence (AI) models has
enabled numerous applications of the Internet of Things (IoT). However, large
AI models require substantial computational and memory resources, which exceed
the capabilities of resource-constrained IoT devices. End-edge collaboration
paradigm is developed to address this issue, where a small model on the end
device performs inference tasks, while a large model on the edge server assists
with model updates. To improve the accuracy of the inference tasks, the data
generated on the end devices will be periodically uploaded to edge serve to
update model, and a distilled model of the updated one will be transmitted back
to the end device. Subjected to the limited bandwidth for the communication
link between the end device and the edge server, it is important to investigate
the system should allocate more bandwidth for data upload or model
transmission. In this paper, we characterize the impact of data upload and
model transmission on inference accuracy. Subsequently, we formulate a
bandwidth allocation problem. By solving this problem, we derive an efficient
optimization framework for the end-edge collaboration system. The simulation
results demonstrate our framework significantly enhances mean average precision
(mAP) under various bandwidths and datasizes.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:23:30 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Yang",
"Dailin",
""
],
[
"Zhang",
"Shuhang",
""
],
[
"Zhang",
"Hongliang",
""
],
[
"Song",
"Lingyang",
""
]
] |
2504.14311 | Ruoyan Xiong | Ruoyan Xiong, Yuke Hou, Princess Retor Torboh, Hui He, Huanbin Zhang,
Yue Zhang, Yanpin Wang, Huipan Guan, and Shang Zhang | DCFG: Diverse Cross-Channel Fine-Grained Feature Learning and
Progressive Fusion Siamese Tracker for Thermal Infrared Target Tracking | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | To address the challenge of capturing highly discriminative features in
ther-mal infrared (TIR) tracking, we propose a novel Siamese tracker based on
cross-channel fine-grained feature learning and progressive fusion. First, we
introduce a cross-channel fine-grained feature learning network that employs
masks and suppression coefficients to suppress dominant target features,
en-abling the tracker to capture more detailed and subtle information. The
net-work employs a channel rearrangement mechanism to enhance efficient
in-formation flow, coupled with channel equalization to reduce parameter count.
Additionally, we incorporate layer-by-layer combination units for ef-fective
feature extraction and fusion, thereby minimizing parameter redun-dancy and
computational complexity. The network further employs feature redirection and
channel shuffling strategies to better integrate fine-grained details. Second,
we propose a specialized cross-channel fine-grained loss function designed to
guide feature groups toward distinct discriminative re-gions of the target,
thus improving overall target representation. This loss function includes an
inter-channel loss term that promotes orthogonality be-tween channels,
maximizing feature diversity and facilitating finer detail capture. Extensive
experiments demonstrate that our proposed tracker achieves the highest
accuracy, scoring 0.81 on the VOT-TIR 2015 and 0.78 on the VOT-TIR 2017
benchmark, while also outperforming other methods across all evaluation metrics
on the LSOTB-TIR and PTB-TIR benchmarks.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:24:37 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Xiong",
"Ruoyan",
""
],
[
"Hou",
"Yuke",
""
],
[
"Torboh",
"Princess Retor",
""
],
[
"He",
"Hui",
""
],
[
"Zhang",
"Huanbin",
""
],
[
"Zhang",
"Yue",
""
],
[
"Wang",
"Yanpin",
""
],
[
"Guan",
"Huipan",
""
],
[
"Zhang",
"Shang",
""
]
] |
2504.14312 | Alexander Teretenkov | N. M. Gerasimov, A. E. Teretenkov | Repeated temperature measurements in quantum thermodynamics | null | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we model the temperature measurement as a transformation of the
arbitrary state into the Gibbs state. We start with a general formalism of
ansatz-posteriors, which includes many usual models of posterior states due to
measurement or state preparation. On the one hand, it contains models of
selective and non-selective measurements posteriors as special cases, which
allows us to consider it as a generalization of usual measurement models. On
the other hand, it contains the abovementioned transformation into the Gibbs
state as a special case as well. We derive an analog of the master equation in
the stroboscopic limit of repeated measurements. Then we apply our general
approach to temperature measurement.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:28:53 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Gerasimov",
"N. M.",
""
],
[
"Teretenkov",
"A. E.",
""
]
] |
2504.14313 | Gaurav Katoch | Gaurav Katoch, Debajyoti Sarkar and Bhim Sen | Holographic timelike entanglement in AdS$_{3}$ Vaidya | Version 1. 27 pages, 17 figures | null | null | null | hep-th gr-qc | http://creativecommons.org/licenses/by/4.0/ | Based on the studies of pseudo-entropy in de Sitter, there have been recent
proposals for a timelike entanglement in AdS/CFT. In this work, we explore this
proposal in the context of a holographic CFT undergoing a global quench. We
study various cases in which the timelike intervals are anchored at various
boundary times, sometimes straddling the infalling shell. Our results seem to
be consistent with how the quenched CFT evolve towards thermalization.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:29:47 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Katoch",
"Gaurav",
""
],
[
"Sarkar",
"Debajyoti",
""
],
[
"Sen",
"Bhim",
""
]
] |
2504.14314 | Antony Seabra Medeiros | Antony Seabra, Sergio Lifschitz | Towards Polyglot Data Processing in Social Networks using the
Hadoop-Spark ecosystem | 6th International Conference on Artificial Intelligence and Big Data
(AIBD 2025), January 18 ~ 19, 2025, Zurich, Switzerland Volume Editors :
David C. Wyld, Dhinaharan Nagamalai (Eds) ISBN : 978-1-923107-48-9 | null | null | null | cs.DC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This article explores the use of the Hadoop-Spark ecosystem for social media
data processing, adopting a polyglot approach with the integration of various
computation and storage technologies, such as Hive, HBase and GraphX. We
discuss specific tasks involved in processing social network data, such as
calculating user influence, counting the most frequent terms in messages and
identifying social relationships among users and groups. We conducted a series
of empirical performance assessments, focusing on executing selected tasks and
measuring their execution time within the Hadoop-Spark cluster. These insights
offer a detailed quantitative analysis of the performance efficiency of the
ecosystem tools. We conclude by highlighting the potential of the Hadoop-Spark
ecosystem tools for advancing research in social networks and related fields.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:33:55 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Seabra",
"Antony",
""
],
[
"Lifschitz",
"Sergio",
""
]
] |
2504.14315 | Naoto Shiraishi | Naoto Shiraishi, Mizuki Yamaguchi | Dichotomy theorem distinguishing non-integrability and the lowest-order
Yang-Baxter equation for isotropic spin chains | 7 pages + 14 pages, 1 figure | null | null | null | cond-mat.stat-mech cond-mat.str-el math-ph math.MP nlin.SI quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the integrability and non-integrability of isotropic spin
chains with nearest-neighbor interaction with general spin $S$. We prove a
dichotomy theorem that a single relation sharply separates two scenarios: (i)
this system is non-integrable, or (ii) the lowest-order Yang-Baxter equation is
satisfied. This result solves in the affirmative the Grabowski-Mathieu
conjecture stating that a model is integrable only if this model has a 3-local
conserved quantity. This theorem also serves as a complete classification of
integrability and non-integrability for $S\leq 13.5$, suggesting that all the
integrable models are in the scope of the Yang-Baxter equation.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:35:49 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Shiraishi",
"Naoto",
""
],
[
"Yamaguchi",
"Mizuki",
""
]
] |
2504.14316 | Shayan Alahyari | Shayan Alahyari, Mike Domaratzki | Local distribution-based adaptive oversampling for imbalanced regression | null | null | null | null | cs.LG stat.ML | http://creativecommons.org/licenses/by/4.0/ | Imbalanced regression occurs when continuous target variables have skewed
distributions, creating sparse regions that are difficult for machine learning
models to predict accurately. This issue particularly affects neural networks,
which often struggle with imbalanced data. While class imbalance in
classification has been extensively studied, imbalanced regression remains
relatively unexplored, with few effective solutions. Existing approaches often
rely on arbitrary thresholds to categorize samples as rare or frequent,
ignoring the continuous nature of target distributions. These methods can
produce synthetic samples that fail to improve model performance and may
discard valuable information through undersampling. To address these
limitations, we propose LDAO (Local Distribution-based Adaptive Oversampling),
a novel data-level approach that avoids categorizing individual samples as rare
or frequent. Instead, LDAO learns the global distribution structure by
decomposing the dataset into a mixture of local distributions, each preserving
its statistical characteristics. LDAO then models and samples from each local
distribution independently before merging them into a balanced training set.
LDAO achieves a balanced representation across the entire target range while
preserving the inherent statistical structure within each local distribution.
In extensive evaluations on 45 imbalanced datasets, LDAO outperforms
state-of-the-art oversampling methods on both frequent and rare target values,
demonstrating its effectiveness for addressing the challenge of imbalanced
regression.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:36:41 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Alahyari",
"Shayan",
""
],
[
"Domaratzki",
"Mike",
""
]
] |
2504.14318 | Santiago Orcajo | Santiago Orcajo, Lucas A. Cieza, Octavio Guilera, Sebastian Perez,
Fernando R. Rannou, Camilo Gonzalez-Ruilova, Grace Batalla-Falcon, Trisha
Bhowmik, Prachi Chavan, Simon Casassus, Anuroop Dasgupta, Kevin Diaz, Jose L.
Gomez, Antonio S. Hales, J.M. Miley, Marcelo M. Miller Bertolami, P.H.
Nogueira, Maria Paula Ronco, Dary Ruiz-Rodriguez, Anibal Sierra, Julia
Venturini, Philipp Weber, Jonathan P. Williams, and Alice Zurlo | The Ophiuchus DIsk Survey Employing ALMA (ODISEA): A Unified
Evolutionary Sequence of Planet-Driven Substructures Explaining the Diversity
of Disk Morphologies | 18 pages, 5 figures, 1 video in mp4 | null | 10.3847/2041-8213/adcd58 | null | astro-ph.EP astro-ph.SR | http://creativecommons.org/licenses/by/4.0/ | Understanding the origin of substructures in protoplanetary disks and their
connection to planet formation is currently one of the main challenges in
astrophysics. While some disks appear smooth, most exhibit diverse
substructures such as gaps, rings, or inner cavities, with varying brightness
and depth. As part of the Ophiuchus Disk Survey Employing ALMA (ODISEA), we
previously proposed an evolutionary sequence to unify this diversity, driven by
the formation of giant planets through core accretion and subsequent
planet-disk interactions. By combining the disk evolution and planet formation
code PLANETALP with the radiative transfer code RADMC-3D, we have now
reproduced the key aspects of the proposed evolutionary sequence. Starting with
a smooth disk (like e.g., WLY 2-63), we modeled the evolution of a fiducial
disk with a 1 Jupiter-mass planet at 57 au. Within a few hundreds of orbits, a
narrow gap forms, resembling ISO-Oph 17. By $\sim$0.1 Myr, the gap widens, and
dust accumulates at the cavity edge, producing a structure similar to Elias
2-24. At $\sim$0.4 Myr, the disk evolves further into a morphology akin to DoAr
44, characterized by a smaller inner disk and a brighter inner rim. By $\sim$1
Myr, the system transitions to a single narrow ring, resembling RXJ1633.9-2442.
This line of work strongly supports the planetary origin of substructures and
enables the possibility of identifying a population of planets that is
currently beyond the reach of more direct detection techniques.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:46:33 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Orcajo",
"Santiago",
""
],
[
"Cieza",
"Lucas A.",
""
],
[
"Guilera",
"Octavio",
""
],
[
"Perez",
"Sebastian",
""
],
[
"Rannou",
"Fernando R.",
""
],
[
"Gonzalez-Ruilova",
"Camilo",
""
],
[
"Batalla-Falcon",
"Grace",
""
],
[
"Bhowmik",
"Trisha",
""
],
[
"Chavan",
"Prachi",
""
],
[
"Casassus",
"Simon",
""
],
[
"Dasgupta",
"Anuroop",
""
],
[
"Diaz",
"Kevin",
""
],
[
"Gomez",
"Jose L.",
""
],
[
"Hales",
"Antonio S.",
""
],
[
"Miley",
"J. M.",
""
],
[
"Bertolami",
"Marcelo M. Miller",
""
],
[
"Nogueira",
"P. H.",
""
],
[
"Ronco",
"Maria Paula",
""
],
[
"Ruiz-Rodriguez",
"Dary",
""
],
[
"Sierra",
"Anibal",
""
],
[
"Venturini",
"Julia",
""
],
[
"Weber",
"Philipp",
""
],
[
"Williams",
"Jonathan P.",
""
],
[
"Zurlo",
"Alice",
""
]
] |
2504.14319 | Maciej Nieszporski | Jan L. Cie\'sli\'nski, Alexander V. Mikhailov, Maciej Nieszporski and
Frank W. Nijhoff | Discrete integrable principal chiral field model and its involutive
reduction | 16 pages | null | null | null | nlin.SI | http://creativecommons.org/licenses/by/4.0/ | We discuss an integrable discretization of the principal chiral field models
equations and its involutive reduction. We present a Darboux transformation and
general construction of solution solutions for these discrete equations.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 14:53:03 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Cieśliński",
"Jan L.",
""
],
[
"Mikhailov",
"Alexander V.",
""
],
[
"Nieszporski",
"Maciej",
""
],
[
"Nijhoff",
"Frank W.",
""
]
] |
2504.14321 | Xingyu Li | Xingyu Li, Chen Gong and Guohong Fu | Multimodal Coreference Resolution for Chinese Social Media Dialogues:
Dataset and Benchmark Approach | null | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multimodal coreference resolution (MCR) aims to identify mentions referring
to the same entity across different modalities, such as text and visuals, and
is essential for understanding multimodal content. In the era of rapidly
growing mutimodal content and social media, MCR is particularly crucial for
interpreting user interactions and bridging text-visual references to improve
communication and personalization. However, MCR research for real-world
dialogues remains unexplored due to the lack of sufficient data resources.To
address this gap, we introduce TikTalkCoref, the first Chinese multimodal
coreference dataset for social media in real-world scenarios, derived from the
popular Douyin short-video platform. This dataset pairs short videos with
corresponding textual dialogues from user comments and includes manually
annotated coreference clusters for both person mentions in the text and the
coreferential person head regions in the corresponding video frames. We also
present an effective benchmark approach for MCR, focusing on the celebrity
domain, and conduct extensive experiments on our dataset, providing reliable
benchmark results for this newly constructed dataset. We will release the
TikTalkCoref dataset to facilitate future research on MCR for real-world social
media dialogues.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 15:15:59 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Li",
"Xingyu",
""
],
[
"Gong",
"Chen",
""
],
[
"Fu",
"Guohong",
""
]
] |
2504.14322 | Antony Seabra Medeiros | Antony Seabra, Sergio Lifschitz | Advancing Polyglot Big Data Processing using the Hadoop ecosystem | null | International Journal of Data Mining & Knowledge Management
Process 2025 (IJDKP) | null | null | cs.DC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This article explores the utilization of the Hadoop ecosystem as a polyglot
big data processing platform, focusing on the integration of diverse
computation and storage technologies and their potential advantages in certain
computational contexts. It delves into the potential of this ecosystem as a
unified platform highlighting its architectural foundations and their
complementary strengths in distributed storage, processing efficiency and
real-time analytics. The article explores potential use cases within domains
such as Smart Cities and Social Networks, illustrating how the platform's
diverse components can be orchestrated in a polyglot manner and how these
fields can benefit from the ecosystem's capabilities. Finally, the article
concludes by showcasing alternatives for future research, including specialized
architectural aspects of the ecosystem to advance the polyglot paradigm.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 15:20:48 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Seabra",
"Antony",
""
],
[
"Lifschitz",
"Sergio",
""
]
] |
2504.14323 | Peter Gerdes | Leo A. Harrington and Peter M. Gerdes | The Tree Pulldown Method: McLaughlin's Conjecture and Beyond | null | null | null | null | math.LO | http://creativecommons.org/licenses/by/4.0/ | This paper finally fully elaborates the tree pulldown method used by one of
us (Harrington) to settle McLaughlin's conjecture. This method enables the
construction of a computable tree $T_0$ whose paths are incomparable over
$0^{(\alpha)}$ and resemble $\alpha$-generics while leaving us almost
completely free to specify the homeomorphism class of $[T_0]$. While a version
of this method for $\alpha = \omega$ previously appeared in print we give the
general construction for an arbitrary ordinal notation $\alpha$. We also
demonstrate this method can be applied to a `non-standard' ordinal notation to
establish the existence of a computable tree whose paths are
hyperarithmetically incomparable and resemble $\alpha$-generics for all $\alpha
< \omega_1^{CK}$. Finally, we verify a number of corollaries including
solutions to problems 57$^{*}$ , 62, 63 (McLaughlin's conjecture), 65 and 71
from Friedman's famous "One Hundred and Two Problems in Mathematical Logic."
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 15:25:53 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Harrington",
"Leo A.",
""
],
[
"Gerdes",
"Peter M.",
""
]
] |
2504.14324 | Florentiana Yuwono | Acacia Chong Xiao Xuan, Florentiana Yuwono, Melissa Anastasia
Harijanto, Xu Yi | Meltdown: Bridging the Perception Gap in Sustainable Food Behaviors
Through Immersive VR | null | null | null | null | cs.CY | http://creativecommons.org/licenses/by/4.0/ | Climate change education often struggles to bridge the perception gap between
everyday actions and their long-term environmental consequences. In response,
we developed Meltdown, an immersive virtual reality (VR) escape room that
simulates a grocery shopping and food waste management experience to educate
university students in Singapore about sustainable consumption. The game
emphasizes sustainable food choices and disposal practices, combining
interactive elements and narrative feedback to promote behavioral change.
Through a user study with 36 university students, we observed statistically
significant improvements in participants objective knowledge, perceived
confidence, and intention to adopt sustainable behaviors. Our results suggest
that experiential VR environments can enhance climate education by making
abstract environmental concepts more immediate and personally relevant.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 15:26:48 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Xuan",
"Acacia Chong Xiao",
""
],
[
"Yuwono",
"Florentiana",
""
],
[
"Harijanto",
"Melissa Anastasia",
""
],
[
"Yi",
"Xu",
""
]
] |
2504.14326 | Jinbo Wen | Jinbo Wen, Jiawen Kang, Yang Zhang, Yue Zhong, Dusit Niyato, Jie Xu,
Jianhang Tang, and Chau Yuen | Diffusion-based Dynamic Contract for Federated AI Agent Construction in
Mobile Metaverses | null | null | null | null | cs.NI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Mobile metaverses have attracted significant attention from both academia and
industry, which are envisioned as the next-generation Internet, providing users
with immersive and ubiquitous metaverse services through mobile devices. Driven
by Large Language Models (LLMs) and Vision-Language Models (VLMs), Artificial
Intelligence (AI) agents hold the potential to empower the creation,
maintenance, and evolution of mobile metaverses. Currently, AI agents are
primarily constructed using cloud-based LLMs and VLMs. However, several
challenges hinder their effective implementation, including high service
latency and potential sensitive data leakage during perception and processing.
In this paper, we develop an edge-cloud collaboration-based federated AI agent
construction framework in mobile metaverses. Specifically, Edge Servers (ESs),
acting as agent infrastructures, collaboratively create agent modules in a
distributed manner. The cloud server then integrates these modules into AI
agents and deploys them at the edge, thereby enabling low-latency AI agent
services for users. Considering that ESs may exhibit dynamic levels of
willingness to participate in federated AI agent construction, we design a
two-period dynamic contract model to continuously motivate ESs to participate
in agent module creation, effectively addressing the dynamic information
asymmetry between the cloud server and the ESs. Furthermore, we propose an
Enhanced Diffusion Model-based Soft Actor-Critic (EDMSAC) algorithm to
efficiently generate optimal dynamic contracts, in which dynamic structured
pruning is applied to DM-based actor networks to enhance denoising efficiency
and policy learning performance. Extensive simulations demonstrate the
effectiveness and superiority of the EDMSAC algorithm and the proposed contract
model.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 15:34:00 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Wen",
"Jinbo",
""
],
[
"Kang",
"Jiawen",
""
],
[
"Zhang",
"Yang",
""
],
[
"Zhong",
"Yue",
""
],
[
"Niyato",
"Dusit",
""
],
[
"Xu",
"Jie",
""
],
[
"Tang",
"Jianhang",
""
],
[
"Yuen",
"Chau",
""
]
] |
2504.14327 | Yang Zhang Dr. | Yang Zhang and Ling-Fang Lin and Adriana Moreo and Satoshi Okamoto and
Thomas A. Maier and Elbio Dagotto | General trends of electronic structures, superconducting pairing, and
magnetic correlations in the Ruddlesden-Popper nickelate $m$-layered
superconductors La$_{m+1}$Ni$_{m}$O$_{3m+1}$ | null | null | null | null | cond-mat.supr-con cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We report a comprehensive theoretical analysis of the Ruddlesden-Popper
layered nickelates La$_{m+1}$Ni$_m$O$_{3m+1}$ ($m = 1$ to 6) under pressure.
Our results suggest that, while these Ruddlesden-Popper layered nickelates
display many similarities, they also show noticeable differences. The Ni
$d_{3z^2-r^2}$ orbitals display bonding-antibonding, or
bonding-antibonding-nonbonding, characteristic splittings, depending on the
even or odd number of stacking layers $m$. In addition, the ratio of the
in-plane interorbital hopping between $d_{3z^2-r^2}$ and $d_{x^2-y^2}$ orbitals
and in-plane intraorbital hopping between $d_{x^2-y^2}$ orbitals was found to
be large in La$_{m+1}$Ni$_m$O$_{3m+1}$ ($m = 1$ to 6), and this ratio increases
from $m = 1$ to $m = 6$, suggesting that the in-plane hybridization will
increase as the layer number $m$ increases. In contrast to the dominant
$s^\pm$--wave state driven by spin fluctuations in the bilayer
La$_3$Ni$_2$O$_7$ and trilayer La$_4$Ni$_3$O$_{10}$, two nearly degenerate
$d_{x^2-y^2}$-wave and $s^\pm$-wave leading states were obtained in the
four-layer stacking La$_5$Ni$_4$O$_{13}$ and five-layer stacking
La$_6$Ni$_5$O$_{16}$. The leading $s^\pm$-wave state was recovered in the
six-layer material La$_7$Ni$_6$O$_{19}$. In general, at the level of the random
phase approximation treatment, the superconducting transition temperature $T_c$
decreases in stoichiometric bulk systems from the bilayer La$_3$Ni$_2$O$_7$ to
the six-layer La$_7$Ni$_6$O$_{19}$, despite the $m$ dependent dominant pairing.
Both in-plane and out-of-plane magnetic correlations are found to be quite
complex. Within the in-plane direction, we obtained the peak of the magnetic
susceptibility at ${\bf q} = (0.6 \pi, 0.6 \pi)$ for La$_5$Ni$_4$O$_{13}$ and
La$_7$Ni$_6$O$_{19}$, and at ${\bf q} = (0.7 \pi, 0.7 \pi)$ for
La$_6$Ni$_5$O$_{16}$.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 15:41:14 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Zhang",
"Yang",
""
],
[
"Lin",
"Ling-Fang",
""
],
[
"Moreo",
"Adriana",
""
],
[
"Okamoto",
"Satoshi",
""
],
[
"Maier",
"Thomas A.",
""
],
[
"Dagotto",
"Elbio",
""
]
] |
2504.14328 | Subhra Mazumdar | Diptendu Chatterjee, Avishek Majumder and Subhra Mazumdar | ScaloWork: Useful Proof-of-Work with Distributed Pool Mining | 17 pages | null | null | null | cs.CR cs.ET | http://creativecommons.org/licenses/by/4.0/ | Bitcoin blockchain uses hash-based Proof-of-Work (PoW) that prevents unwanted
participants from hogging the network resources. Anyone entering the mining
game has to prove that they have expended a specific amount of computational
power. However, the most popular Bitcoin blockchain consumes 175.87 TWh of
electrical energy annually, and most of this energy is wasted on hash
calculations, which serve no additional purpose. Several studies have explored
re-purposing the wasted energy by replacing the hash function with meaningful
computational problems that have practical applications. Minimum Dominating Set
(MDS) in networks has numerous real-life applications. Building on this
concept, Chrisimos [TrustCom '23] was proposed to replace hash-based PoW with
the computation of a dominating set on real-life graph instances. However,
Chrisimos has several drawbacks regarding efficiency and solution quality. This
work presents a new framework for Useful PoW, ScaloWork, that decides the block
proposer for the Bitcoin blockchain based on the solution for the dominating
set problem. ScaloWork relies on the property of graph isomorphism and
guarantees solution extractability. We also propose a distributed approach for
calculating the dominating set, allowing miners to collaborate in a pool. This
enables ScaloWork to handle larger graphs relevant to real-life applications,
thereby enhancing scalability. Our framework also eliminates the problem of
free-riders, ensuring fairness in the distribution of block rewards. We perform
a detailed security analysis of our framework and prove our scheme as secure as
hash-based PoW. We implement a prototype of our framework, and the results show
that our system outperforms Chrisimos in all aspects.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 15:43:27 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Chatterjee",
"Diptendu",
""
],
[
"Majumder",
"Avishek",
""
],
[
"Mazumdar",
"Subhra",
""
]
] |
2504.14329 | Riccardo Capelli | Bruno Stegani and Riccardo Capelli | Kinetic rates calculation via non-equilibrium dynamics | null | null | null | null | physics.chem-ph | http://creativecommons.org/licenses/by-nc-sa/4.0/ | This study introduces a novel computational approach based on
ratchet-and-pawl molecular dynamics (rMD) for accurately estimating ligand
dissociation kinetics in protein-ligand complexes. By integrating Kramers'
theory with Bell's equation, our method systematically investigates the
relationship between the effective biasing force applied during simulations and
the ligand residence times. The proposed technique is demonstrated through
extensive simulations of the benzamidine-trypsin complex, employing first an
implicit solvent model (multi-eGO) to set up the approach parameters and thus
an explicit solvent model. Our results illustrate the method's reliability,
accuracy, and computational efficiency, with calculated kinetic rates closely
matching experimental values. Overall, this study highlights rMD as a versatile
and efficient non-equilibrium methodology, broadly applicable to kinetic
analyses in chemical and biological systems.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 15:43:29 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Stegani",
"Bruno",
""
],
[
"Capelli",
"Riccardo",
""
]
] |
2504.14330 | Zhengqiu Zhu | Xiaoran Zhang, Yatai Ji, Yong Zhao, Chuan Ai, Bin Chen, and Zhengqiu
Zhu | DLW-CI: A Dynamic Likelihood-Weighted Cooperative Infotaxis Approach for
Multi-Source Search in Urban Environments Using Consumer Drone Networks | null | null | null | null | cs.IT cs.RO math.IT | http://creativecommons.org/licenses/by/4.0/ | Consumer-grade drones equipped with low-cost sensors have emerged as a
cornerstone of Autonomous Intelligent Systems (AISs) for environmental
monitoring and hazardous substance detection in urban environments. However,
existing research primarily addresses single-source search problems,
overlooking the complexities of real-world urban scenarios where both the
location and quantity of hazardous sources remain unknown. To address this
issue, we propose the Dynamic Likelihood-Weighted Cooperative Infotaxis
(DLW-CI) approach for consumer drone networks. Our approach enhances
multi-drone collaboration in AISs by combining infotaxis (a cognitive search
strategy) with optimized source term estimation and an innovative cooperative
mechanism. Specifically, we introduce a novel source term estimation method
that utilizes multiple parallel particle filters, with each filter dedicated to
estimating the parameters of a potentially unknown source within the search
scene. Furthermore, we develop a cooperative mechanism based on dynamic
likelihood weights to prevent multiple drones from simultaneously estimating
and searching for the same source, thus optimizing the energy efficiency and
search coverage of the consumer AIS. Experimental results demonstrate that the
DLW-CI approach significantly outperforms baseline methods regarding success
rate, accuracy, and root mean square error, particularly in scenarios with
relatively few sources, regardless of the presence of obstacles. Also, the
effectiveness of the proposed approach is verified in a diffusion scenario
generated by the computational fluid dynamics (CFD) model. Research findings
indicate that our approach could improve source estimation accuracy and search
efficiency by consumer drone-based AISs, making a valuable contribution to
environmental safety monitoring applications within smart city infrastructure.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 15:44:09 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Zhang",
"Xiaoran",
""
],
[
"Ji",
"Yatai",
""
],
[
"Zhao",
"Yong",
""
],
[
"Ai",
"Chuan",
""
],
[
"Chen",
"Bin",
""
],
[
"Zhu",
"Zhengqiu",
""
]
] |
2504.14331 | Zhipeng Gao | Yubo Mai and Zhipeng Gao and Xing Hu and Lingfeng Bao and Jingyuan
Chen and Jianling Sun | Code2API: A Tool for Generating Reusable APIs from Stack Overflow Code
Snippets | null | null | null | null | cs.SE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Nowadays, developers often turn to Stack Overflow for solutions to daily
problems, however, these code snippets are partial code that cannot be tested
and verified properly. One way to test these code snippets is to transform them
into APIs (Application Program Interface) that developers can be directly
invoked and executed. However, it is often costly and error-prone for
developers to manually perform this transformation (referred to as AIPzation
task) due to different actions to be taken (e.g., summarizing proper method
names, inferring input parameters list and return statements). To help
developers quickly reuse code snippets in Stack Overflow, in this paper, we
propose Code2API, a Google Chrome extension that uses Large Language Models
(LLMs) to automatically perform APIzation of code snippets on Stack Overflow.
\toolname guides LLMs through well-designed prompts to generate reusable APIs,
using Chain-of-Thought reasoning and few-shot in-context learning to help LLMs
understand and solve the APIzation task in a developer-like manner. The
evaluation results show that Code2API significantly outperforms the rule-based
approach by a large margin.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 15:49:03 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Mai",
"Yubo",
""
],
[
"Gao",
"Zhipeng",
""
],
[
"Hu",
"Xing",
""
],
[
"Bao",
"Lingfeng",
""
],
[
"Chen",
"Jingyuan",
""
],
[
"Sun",
"Jianling",
""
]
] |
2504.14332 | Atefeh Javadi | Sepideh Ghaziasgar, Mahdi Abdollahi, Atefeh Javadi, Jacco van Loon,
Iain McDonald, Joana Oliveira, Amirhossein Masoudnezhad, Habib Khosroshahi,
Bernard Foing, Fatemeh Fazel | Dusty stellar sources classification by implementing machine learning
methods based on spectroscopic observations in the Magellanic Clouds | Accepted for publication in ApJ | null | null | null | astro-ph.GA astro-ph.IM astro-ph.SR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Dusty stellar point sources are a significant stage in stellar evolution and
contribute to the metal enrichment of galaxies. These objects can be classified
using photometric and spectroscopic observations with color-magnitude diagrams
(CMD) and infrared excesses in spectral energy distributions (SED). We employed
supervised machine learning spectral classification to categorize dusty stellar
sources, including young stellar objects (YSOs) and evolved stars (oxygen- and
carbon-rich asymptotic giant branch stars, AGBs), red supergiants (RSGs), and
post-AGB (PAGB) stars in the Large and Small Magellanic Clouds, based on
spectroscopic labeled data from the Surveying the Agents of Galaxy Evolution
(SAGE) project, which used 12 multiwavelength filters and 618 stellar objects.
Despite missing values and uncertainties in the SAGE spectral datasets, we
achieved accurate classifications. To address small and imbalanced spectral
catalogs, we used the Synthetic Minority Oversampling Technique (SMOTE) to
generate synthetic data points. Among models applied before and after data
augmentation, the Probabilistic Random Forest (PRF), a tuned Random Forest
(RF), achieved the highest total accuracy, reaching $\mathbf{89\%}$ based on
recall in categorizing dusty stellar sources. Using SMOTE does not improve the
best model's accuracy for the CAGB, PAGB, and RSG classes; it remains
$\mathbf{100\%}$, $\mathbf{100\%}$, and $\mathbf{88\%}$, respectively, but
shows variations for OAGB and YSO classes. We also collected photometric
labeled data similar to the training dataset, classifying them using the top
four PRF models with over $\mathbf{87\%}$ accuracy. Multiwavelength data from
several studies were classified using a consensus model integrating four top
models to present common labels as final predictions.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 15:53:01 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Ghaziasgar",
"Sepideh",
""
],
[
"Abdollahi",
"Mahdi",
""
],
[
"Javadi",
"Atefeh",
""
],
[
"van Loon",
"Jacco",
""
],
[
"McDonald",
"Iain",
""
],
[
"Oliveira",
"Joana",
""
],
[
"Masoudnezhad",
"Amirhossein",
""
],
[
"Khosroshahi",
"Habib",
""
],
[
"Foing",
"Bernard",
""
],
[
"Fazel",
"Fatemeh",
""
]
] |
2504.14334 | Zeyad Manaa | Zeyad M. Manaa, Ayman M. Abdallah, Mohamed Ismail, and Samil El Ferik | Koopman-Based Event-Triggered Control from Data | null | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | Event-triggered Control (ETC) presents a promising paradigm for efficient
resource usage in networked and embedded control systems by reducing
communication instances compared to traditional time-triggered strategies. This
paper introduces a novel approach to ETC for discrete-time nonlinear systems
using a data-driven framework. By leveraging Koopman operator theory, the
nonlinear system dynamics are globally linearized (approximately in practical
settings) in a higher-dimensional space. We design a state-feedback controller
and an event-triggering policy directly from data, ensuring exponential
stability in Lyapunov sense. The proposed method is validated through extensive
simulation experiments, demonstrating significant resource savings.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 15:57:59 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Manaa",
"Zeyad M.",
""
],
[
"Abdallah",
"Ayman M.",
""
],
[
"Ismail",
"Mohamed",
""
],
[
"Ferik",
"Samil El",
""
]
] |
2504.14335 | Yuxi Zhou | Zhengbo Zhang, Yuxi Zhou, Duo Peng, Joo-Hwee Lim, Zhigang Tu, De Wen
Soh, Lin Geng Foo | Visual Prompting for One-shot Controllable Video Editing without
Inversion | accepted by cvpr2025 | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | One-shot controllable video editing (OCVE) is an important yet challenging
task, aiming to propagate user edits that are made -- using any image editing
tool -- on the first frame of a video to all subsequent frames, while ensuring
content consistency between edited frames and source frames. To achieve this,
prior methods employ DDIM inversion to transform source frames into latent
noise, which is then fed into a pre-trained diffusion model, conditioned on the
user-edited first frame, to generate the edited video. However, the DDIM
inversion process accumulates errors, which hinder the latent noise from
accurately reconstructing the source frames, ultimately compromising content
consistency in the generated edited frames. To overcome it, our method
eliminates the need for DDIM inversion by performing OCVE through a novel
perspective based on visual prompting. Furthermore, inspired by consistency
models that can perform multi-step consistency sampling to generate a sequence
of content-consistent images, we propose a content consistency sampling (CCS)
to ensure content consistency between the generated edited frames and the
source frames. Moreover, we introduce a temporal-content consistency sampling
(TCS) based on Stein Variational Gradient Descent to ensure temporal
consistency across the edited frames. Extensive experiments validate the
effectiveness of our approach.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:00:47 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Zhang",
"Zhengbo",
""
],
[
"Zhou",
"Yuxi",
""
],
[
"Peng",
"Duo",
""
],
[
"Lim",
"Joo-Hwee",
""
],
[
"Tu",
"Zhigang",
""
],
[
"Soh",
"De Wen",
""
],
[
"Foo",
"Lin Geng",
""
]
] |
2504.14336 | Vu Nguyen | Duy Cao, Phu Nguyen, Vy Le, Tien N. Nguyen, and Vu Nguyen | Toward Generation of Test Cases from Task Descriptions via History-aware
Planning | Under review | null | null | null | cs.SE | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In automated web testing, generating test scripts from natural language task
descriptions is crucial for enhancing the test generation process. This
activity involves creating the correct sequences of actions to form test
scripts for future testing activities. Current state-of-the-art approaches are
limited in generating these action sequences, as they either demand substantial
manual effort for human demonstrations or fail to consider the history of
previous web content and actions to decide the next action. In this paper, we
introduce HxAgent, an iterative large language model agent planning approach
that determines the next action based on: 1) observations of the current
contents and feasible actions, 2) short-term memory of previous web states and
actions, and 3) long-term experience with (in)correct action sequences. The
agent generates a sequence of actions to perform a given task, which is
effectively an automated test case to verify the task. We conducted an
extensive empirical evaluation of HxAgent using two datasets. On the MiniWoB++
dataset, our approach achieves 97% exact-match accuracy that is comparable to
the best baselines while eliminating the need for human demonstrations required
by those methods. For complex tasks requiring navigation through multiple
actions and screens, HxAgent achieves an average 82% exact-match. On the second
dataset, comprising 350 task instances across seven popular websites, including
YouTube, LinkedIn, Facebook, and Google, HxAgent achieves high performance,
with 87% of the action sequences exactly matching the ground truth and a
prefix-match of 93%, outperforming the baseline by 59%.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:03:03 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Cao",
"Duy",
""
],
[
"Nguyen",
"Phu",
""
],
[
"Le",
"Vy",
""
],
[
"Nguyen",
"Tien N.",
""
],
[
"Nguyen",
"Vu",
""
]
] |
2504.14337 | Josef Taher | Josef Taher, Eric Hyypp\"a, Matti Hyypp\"a, Klaara Salolahti, Xiaowei
Yu, Leena Matikainen, Antero Kukko, Matti Lehtom\"aki, Harri Kaartinen,
Sopitta Thurachen, Paula Litkey, Ville Luoma, Markus Holopainen, Gefei Kong,
Hongchao Fan, Petri R\"onnholm, Antti Polvivaara, Samuli Junttila, Mikko
Vastaranta, Stefano Puliti, Rasmus Astrup, Joel Kostensalo, Mari Myllym\"aki,
Maksymilian Kulicki, Krzysztof Stere\'nczak, Raul de Paula Pires, Ruben
Valbuena, Juan Pedro Carbonell-Rivera, Jes\'us Torralba, Yi-Chen Chen, Lukas
Winiwarter, Markus Hollaus, Gottfried Mandlburger, Narges Takhtkeshha, Fabio
Remondino, Maciej Lisiewicz, Bart{\l}omiej Kraszewski, Xinlian Liang,
Jianchang Chen, Eero Ahokas, Kirsi Karila, Eugeniu Vezeteu, Petri Manninen,
Roope N\"asi, Heikki Hyyti, Siiri Pyykk\"onen, Peilun Hu, Juha Hyypp\"a | Multispectral airborne laser scanning for tree species classification: a
benchmark of machine learning and deep learning algorithms | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Climate-smart and biodiversity-preserving forestry demands precise
information on forest resources, extending to the individual tree level.
Multispectral airborne laser scanning (ALS) has shown promise in automated
point cloud processing and tree segmentation, but challenges remain in
identifying rare tree species and leveraging deep learning techniques. This
study addresses these gaps by conducting a comprehensive benchmark of machine
learning and deep learning methods for tree species classification. For the
study, we collected high-density multispectral ALS data (>1000 pts/m$^2$) at
three wavelengths using the FGI-developed HeliALS system, complemented by
existing Optech Titan data (35 pts/m$^2$), to evaluate the species
classification accuracy of various algorithms in a test site located in
Southern Finland. Based on 5261 test segments, our findings demonstrate that
point-based deep learning methods, particularly a point transformer model,
outperformed traditional machine learning and image-based deep learning
approaches on high-density multispectral point clouds. For the high-density ALS
dataset, a point transformer model provided the best performance reaching an
overall (macro-average) accuracy of 87.9% (74.5%) with a training set of 1065
segments and 92.0% (85.1%) with 5000 training segments. The best image-based
deep learning method, DetailView, reached an overall (macro-average) accuracy
of 84.3% (63.9%), whereas a random forest (RF) classifier achieved an overall
(macro-average) accuracy of 83.2% (61.3%). Importantly, the overall
classification accuracy of the point transformer model on the HeliALS data
increased from 73.0% with no spectral information to 84.7% with single-channel
reflectance, and to 87.9% with spectral information of all the three channels.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:03:49 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Taher",
"Josef",
""
],
[
"Hyyppä",
"Eric",
""
],
[
"Hyyppä",
"Matti",
""
],
[
"Salolahti",
"Klaara",
""
],
[
"Yu",
"Xiaowei",
""
],
[
"Matikainen",
"Leena",
""
],
[
"Kukko",
"Antero",
""
],
[
"Lehtomäki",
"Matti",
""
],
[
"Kaartinen",
"Harri",
""
],
[
"Thurachen",
"Sopitta",
""
],
[
"Litkey",
"Paula",
""
],
[
"Luoma",
"Ville",
""
],
[
"Holopainen",
"Markus",
""
],
[
"Kong",
"Gefei",
""
],
[
"Fan",
"Hongchao",
""
],
[
"Rönnholm",
"Petri",
""
],
[
"Polvivaara",
"Antti",
""
],
[
"Junttila",
"Samuli",
""
],
[
"Vastaranta",
"Mikko",
""
],
[
"Puliti",
"Stefano",
""
],
[
"Astrup",
"Rasmus",
""
],
[
"Kostensalo",
"Joel",
""
],
[
"Myllymäki",
"Mari",
""
],
[
"Kulicki",
"Maksymilian",
""
],
[
"Stereńczak",
"Krzysztof",
""
],
[
"Pires",
"Raul de Paula",
""
],
[
"Valbuena",
"Ruben",
""
],
[
"Carbonell-Rivera",
"Juan Pedro",
""
],
[
"Torralba",
"Jesús",
""
],
[
"Chen",
"Yi-Chen",
""
],
[
"Winiwarter",
"Lukas",
""
],
[
"Hollaus",
"Markus",
""
],
[
"Mandlburger",
"Gottfried",
""
],
[
"Takhtkeshha",
"Narges",
""
],
[
"Remondino",
"Fabio",
""
],
[
"Lisiewicz",
"Maciej",
""
],
[
"Kraszewski",
"Bartłomiej",
""
],
[
"Liang",
"Xinlian",
""
],
[
"Chen",
"Jianchang",
""
],
[
"Ahokas",
"Eero",
""
],
[
"Karila",
"Kirsi",
""
],
[
"Vezeteu",
"Eugeniu",
""
],
[
"Manninen",
"Petri",
""
],
[
"Näsi",
"Roope",
""
],
[
"Hyyti",
"Heikki",
""
],
[
"Pyykkönen",
"Siiri",
""
],
[
"Hu",
"Peilun",
""
],
[
"Hyyppä",
"Juha",
""
]
] |
2504.14338 | Ionut-Gabriel Farcas | Ionut-Gabriel Farcas, Rayomand P. Gundevia, Ramakanth Munipalli, and
Karen E. Willcox | A parallel implementation of reduced-order modeling of large-scale
systems | 19 pages, 4 figures; the corresponding code can be found at
https://github.com/ionutfarcas/distributed_Operator_Inference | null | 10.2514/6.2025-1170 | null | cs.DC cs.CE | http://creativecommons.org/licenses/by/4.0/ | Motivated by the large-scale nature of modern aerospace engineering
simulations, this paper presents a detailed description of distributed Operator
Inference (dOpInf), a recently developed parallel algorithm designed to
efficiently construct physics-based reduced-order models (ROMs) for problems
with large state dimensions. One such example is the simulation of rotating
detonation rocket engines, where snapshot data generated by high-fidelity
large-eddy simulations have many millions of degrees of freedom. dOpInf
enables, via distributed computing, the efficient processing of datasets with
state dimensions that are too large to process on a single computer, and the
learning of structured physics-based ROMs that approximate the dynamical
systems underlying those datasets. All elements of dOpInf are scalable, leading
to a fully parallelized reduced modeling approach that can scale to the
thousands of processors available on leadership high-performance computing
platforms. The resulting ROMs are computationally cheap, making them ideal for
key engineering tasks such as design space exploration, risk assessment, and
uncertainty quantification. To illustrate the practical application of dOpInf,
we provide a step-by-step tutorial using a 2D Navier-Stokes flow over a step
scenario as a case study. This tutorial guides users through the implementation
process, making dOpInf accessible for integration into complex aerospace
engineering simulations.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:09:16 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Farcas",
"Ionut-Gabriel",
""
],
[
"Gundevia",
"Rayomand P.",
""
],
[
"Munipalli",
"Ramakanth",
""
],
[
"Willcox",
"Karen E.",
""
]
] |
2504.14339 | Carsten Dietzel | Carsten Dietzel | Endocabling of involutive solutions to the Yang-Baxter equation, with an
application to solutions whose diagonal is a cyclic permutation | 25 pages, Comments Welcome! | null | null | null | math.QA math.GR math.RA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this article, we introduce endocabling as a technique to deform
involutive, non-degenerate set-theoretic solutions to the Yang-Baxter equation
(``solutions'', for short) by means of $\lambda$-endomorphisms of their
associated permutation brace, thus generalizing the cabling method by Lebed,
Vendramin and Ram\'{i}rez. In the first part of the article, we define
endocabling and investigate the behaviour of solutions and their invariants
under endocabling. In the second part, we apply our findings to solutions of
size $n$ whose diagonal map is an $n$-cycle: we will prove that solutions with
this property whose size is an odd prime power, are of finite multipermutation
level. Furthermore, solutions with this property whose size is a power of $2$,
will be proven either to be of finite multipermutation level or to admit an
iterated retraction onto a unique solution of size $4$.
We formulate our results in the language of cycle sets.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:10:06 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Dietzel",
"Carsten",
""
]
] |
2504.14340 | Philip Zucker | Philip Zucker | Omelets Need Onions: E-graphs Modulo Theories via Bottom-up E-matching | Submitted to EGRAPHS 2025 | null | null | null | cs.PL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | E-graphs are a data structure for equational reasoning and optimization over
ground terms. One of the benefits of e-graph rewriting is that it can
declaratively handle useful but difficult to orient identities like
associativity and commutativity (AC) in a generic way. However, using these
generic mechanisms is more computationally expensive than using bespoke
routines on terms containing sets, multi-sets, linear expressions, polynomials,
and binders. A natural question arises: How can one combine the generic
capabilities of e-graph rewriting with these specialized theories. This paper
discusses a pragmatic approach to this e-graphs modulo theories (EMT) question
using two key ideas: bottom-up e-matching and semantic e-ids.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:15:11 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Zucker",
"Philip",
""
]
] |
2504.14341 | Cheng Cheng | Cheng Cheng, Qiyu Sun, Cong Zheng | Iterative Polynomial Approximation Algorithms for Inverse Graph Filters | This paper has been accepted by SampTA 2025 | null | null | null | eess.SP | http://creativecommons.org/licenses/by/4.0/ | Chebyshev interpolation polynomials exhibit the exponential approximation
property to analytic functions on a cube. Based on the Chebyshev interpolation
polynomial approximation, we propose
iterative polynomial approximation algorithms to implement the inverse filter
with a polynomial graph filter of commutative graph shifts in a distributed
manner. The proposed algorithms exhibit exponential convergence properties, and
they can be implemented on distributed networks in which agents are equipped
with a data processing subsystem for limited data storage and computation
power, and with a one-hop communication subsystem for direct data exchange only
with their adjacent agents. Our simulations show that the proposed polynomial
approximation algorithms may converge faster than the Chebyshev polynomial
approximation algorithm
and the conventional gradient descent algorithm
do.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:18:16 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Cheng",
"Cheng",
""
],
[
"Sun",
"Qiyu",
""
],
[
"Zheng",
"Cong",
""
]
] |
2504.14343 | Maria Olympia Tsianni | Christoph Reisinger, Maria Olympia Tsianni | Numerical analysis of a particle system for the calibrated Heston-type
local stochastic volatility model | null | null | null | null | q-fin.CP cs.NA math.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We analyse a Monte Carlo particle method for the simulation of the calibrated
Heston-type local stochastic volatility (H-LSV) model. The common application
of a kernel estimator for a conditional expectation in the calibration
condition results in a McKean-Vlasov (MV) stochastic differential equation
(SDE) with non-standard coefficients. The primary challenges lie in certain
mean-field terms in the drift and diffusion coefficients and the
$1/2$-H\"{o}lder regularity of the diffusion coefficient. We establish the
well-posedness of this equation for a fixed but arbitrarily small bandwidth of
the kernel estimator. Moreover, we prove a strong propagation of chaos result,
ensuring convergence of the particle system under a condition on the Feller
ratio and up to a critical time. For the numerical simulation, we employ an
Euler-Maruyama scheme for the log-spot process and a full truncation Euler
scheme for the CIR volatility process. Under certain conditions on the inputs
and the Feller ratio, we prove strong convergence of the Euler-Maruyama scheme
with rate $1/2$ in time, up to a logarithmic factor. Numerical experiments
illustrate the convergence of the discretisation scheme and validate the
propagation of chaos in practice.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:21:28 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Reisinger",
"Christoph",
""
],
[
"Tsianni",
"Maria Olympia",
""
]
] |
2504.14344 | Igor Svyatnyy | Igor Svyatnyy | On the action of the cactus group on the set of Gelfand-Tsetlin patterns
for orthogonal Lie algebras | null | null | null | null | math.RT math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The purpose of this work is to define a natural action of the cactus group on
the set of Gelfand-Tsetlin patterns for orthogonal Lie algebras. These
Gelfand-Tsetlin patterns are meant to index the Gelfand-Tsetlin basis in the
irreducible representations of the orthogonal Lie algebra $\mathfrak{o}_N$ with
respect to the chain of nested orthogonal Lie algebras $\mathfrak{o}_N \supset
\mathfrak{o}_{N-1} \supset \ldots \supset \mathfrak{o}_3$. Using the Howe
duality between $O_N$ and $\mathfrak{o}_{2n}$, we realize some representations
of $\mathfrak{o}_N$ as multiplicity spaces inside the tensor power of the
spinor representation $(\Lambda \mathbb{C}^{n})^{\otimes N}$. There is a
natural choice of the basis inside the multiplicity space, which agrees with
the decomposition of $(\Lambda \mathbb{C}^{n})^{\otimes N}$ into simple
$\mathfrak{o}_{2n}$-modules. We call such basis principal. The action of the
cactus group $C_N$ by the crystal commutors on the crystal arising from
$(\Lambda \mathbb{C}^{n})^{\otimes N}$ induces the action of $C_N$ on the set
indexing the principal basis inside the multiplicity space. We call this set
regular cell tables. Regular cell tables are the analog of semi-standard Young
tables. There is a natural bijection between a specific subset of semi-standard
Young tables and regular cell tables. In this paper, we establish a natural
bijection between the principal basis and the Gelfand-Tsetlin basis and,
therefore, define an action of the cactus group on the set Gelfand-Tsetlin
patterns.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:24:29 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Svyatnyy",
"Igor",
""
]
] |
2504.14345 | Youngbin Lee | Youngbin Lee, Yejin Kim, Suin Kim, Yongjae Lee | Integrating LLM-Generated Views into Mean-Variance Optimization Using
the Black-Litterman Model | Presented at the ICLR 2025 Workshop on Financial AI
(https://sites.google.com/view/financialaiiclr25/home) | null | null | null | q-fin.PM cs.AI | http://creativecommons.org/licenses/by/4.0/ | Portfolio optimization faces challenges due to the sensitivity in traditional
mean-variance models. The Black-Litterman model mitigates this by integrating
investor views, but defining these views remains difficult. This study explores
the integration of large language models (LLMs) generated views into portfolio
optimization using the Black-Litterman framework. Our method leverages LLMs to
estimate expected stock returns from historical prices and company metadata,
incorporating uncertainty through the variance in predictions. We conduct a
backtest of the LLM-optimized portfolios from June 2024 to February 2025,
rebalancing biweekly using the previous two weeks of price data. As baselines,
we compare against the S&P 500, an equal-weighted portfolio, and a traditional
mean-variance optimized portfolio constructed using the same set of stocks.
Empirical results suggest that different LLMs exhibit varying levels of
predictive optimism and confidence stability, which impact portfolio
performance. The source code and data are available at
https://github.com/youngandbin/LLM-MVO-BLM.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:26:14 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Lee",
"Youngbin",
""
],
[
"Kim",
"Yejin",
""
],
[
"Kim",
"Suin",
""
],
[
"Lee",
"Yongjae",
""
]
] |
2504.14346 | Milton da Costa Lopes Fo. | David Ambrose, Milton Lopes Filho and Helena Nussenzveig Lopes | Existence and analyticity of solutions of nonlinear parabolic model
equations with singular data | 30 pages | null | null | null | math.AP | http://creativecommons.org/licenses/by/4.0/ | We explore two approaches to proving existence and analyticity of solutions
to nonlinear parabolic differential equations. One of these methods works well
for more general nonlinearities, while the second method gives stronger results
when the nonlinearity is simpler. The first approach uses the exponentially
weighted Wiener algebra, and is related to prior work of Duchon and Robert for
vortex sheets. The second approach uses two norms, one with a supremum in time
and one with an integral in time, with the integral norm representing the
parabolic gain of regularity. As an example of the first approach we prove
analyticity of small solutions of a class of generalized one-dimensional
Kuramoto-Sivashinsky equations, which model the motion of flame fronts and
other phenomena. To illustrate the second approach, we prove existence and
analyticity of solutions of the dissipative Constantin-Lax-Majda equation
(which models vortex stretching), with and without added advection, with two
classes of rough data. The classes of data treated include both data in the
Wiener algebra with negative-power weights, as well as data in pseudomeasure
spaces with negative-power weights.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:26:20 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Ambrose",
"David",
""
],
[
"Filho",
"Milton Lopes",
""
],
[
"Lopes",
"Helena Nussenzveig",
""
]
] |
2504.14347 | Jiakuan Lu | Jiakuan Lu, Xi Huang, Qinwei Lian, Wei Meng | Finite groups with few subgroups not in the Chermak-Delgado lattice | null | null | null | null | math.GR | http://creativecommons.org/licenses/by/4.0/ | For a finite group G, we denote by v(G) the number of conjugacy classes of
subgroups of G not in CD(G). In this paper, we determine the finite groups G
such that v(G)=1,2,3.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:27:24 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Lu",
"Jiakuan",
""
],
[
"Huang",
"Xi",
""
],
[
"Lian",
"Qinwei",
""
],
[
"Meng",
"Wei",
""
]
] |
2504.14349 | Robert Erickson | R. P. Erickson | Application of an upsampling algorithm to quantum state preparation of
continuous and discrete probability distributions | null | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Upsampling of time series data is often associated with classical fast
Fourier transform analysis, but it also can be employed in the preparation of a
quantum state vector. The quantum circuit of preparation derived from
functional data sampled in this way exhibits exponential gate depth, like other
divide-and-conquer algorithms, but is transformable to an equivalent circuit of
polylogarithmic gate depth. In this study, we derive a polylogarthmic
upsampling algorithm for construction of a state vector with amplitudes that
are square roots of probabilities sampled from a continuous probability
distribution having support over the entire real line. As examples, we prepare
state vectors associated with Gaussian and Laplace distributions; state vectors
for other continuous probability distributions such as Cauchy and Student's t
follow in a similar manner. We also extend the algorithm to include preparation
of a state vector whose amplitudes are square roots of an arbitrary
distribution of discrete probabilities. Our analyses focus on univariate
distributions, but can be extended to multivariate forms. The polylogarithmic
upsampling algorithm has financial and scientific application.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:29:31 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Erickson",
"R. P.",
""
]
] |
2504.14351 | Shashank Motepalli | Shashank Motepalli and Hans-Arno Jacobsen | Decentralization in PoS Blockchain Consensus: Quantification and
Advancement | null | IEEE Transactions on Network and Service Management (2025) | 10.1109/TNSM.2025.3561098 | null | cs.DC cs.CY cs.ET | http://creativecommons.org/licenses/by/4.0/ | Decentralization is a foundational principle of permissionless blockchains,
with consensus mechanisms serving a critical role in its realization. This
study quantifies the decentralization of consensus mechanisms in proof-of-stake
(PoS) blockchains using a comprehensive set of metrics, including Nakamoto
coefficients, Gini, Herfindahl Hirschman Index (HHI), Shapley values, and Zipfs
coefficient. Our empirical analysis across ten prominent blockchains reveals
significant concentration of stake among a few validators, posing challenges to
fair consensus. To address this, we introduce two alternative weighting models
for PoS consensus: Square Root Stake Weight (SRSW) and Logarithmic Stake Weight
(LSW), which adjust validator influence through non-linear transformations.
Results demonstrate that SRSW and LSW models improve decentralization metrics
by an average of 51% and 132%, respectively, supporting more equitable and
resilient blockchain systems.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:33:30 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Motepalli",
"Shashank",
""
],
[
"Jacobsen",
"Hans-Arno",
""
]
] |
2504.14352 | Kaizhe Chen | Kaizhe Chen, Shiping Liu, Zhe You | Connectivity versus Lin-Lu-Yau curvature | 22 pages | null | null | null | math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We explore the interaction between connectivity and Lin-Lu-Yau curvature of
graphs systematically. The intuition is that connected graphs with large
Lin-Lu-Yau curvature also have large connectivity, and vice versa. We prove
that the connectivity of a connected graph is lower bounded by the product of
its minimum degree and its Lin-Lu-Yau curvature. On the other hand, if the
connectivity of a graph $G$ on $n$ vertices is at least $\frac{n-1}{2}$, then
$G$ has positive Lin-Lu-Yau curvature. Moreover, the bound $\frac{n-1}{2}$ here
is optimal. Furthermore, we prove that the edge-connectivity is equal to the
minimum vertex degree for any connected graph with positive Lin-Lu-Yau
curvature. As applications, we estimate or determine the connectivity and
edge-connectivity of an amply regular graph with parameters $(d,\alpha,\beta)$
such that $1\neq \beta\geq \alpha$.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:42:11 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Chen",
"Kaizhe",
""
],
[
"Liu",
"Shiping",
""
],
[
"You",
"Zhe",
""
]
] |
2504.14353 | Ameneh Farhadian | Ameneh Farhadian | Goldbach Conjecture: Violation Probability and Generalization to
Prime-like Distributions | null | null | null | null | math.NT math.CO | http://creativecommons.org/licenses/by/4.0/ | Due to the distribution of primes among integers, we establish an upper bound
for the probability $\mathbb{P}_n$ that the Goldbach conjecture fails. Assuming
the conjecture holds true for all even number less than $2N$, we prove this
probability is less than $e^{-N^\alpha}$, where $ \alpha = 1 - \frac{2\ln\ln
N}{\ln N}$.
For large $N$, this probability becomes vanishingly small, effectively
precluding the existence of counterexamples in practice. If $N =4 \times
10^{18}$, the probability of a counterexample is less than $e^{-10^{15}}$. Our
approach fundamentally depends on the distributional properties of primes
rather than their primality per se. This perspective enables a natural
generalization of the conjecture to non-prime subsets of integers that exhibit
similar distributional characteristics. As a concrete example, we construct new
subsets by applying random $\pm 1$ shifts to primes, which preserve the
essential prime-like distributional properties. Computational verification
confirms that this generalized Goldbach conjecture holds for all even integers
up to $2 \times 10^{8}$ within these modified subsets.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:43:41 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Farhadian",
"Ameneh",
""
]
] |
2504.14354 | Pablo Mones | Jushan Bai and Pablo Mones | Global identification of dynamic panel models with interactive effects | null | null | null | null | econ.EM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper examines the problem of global identification in dynamic panel
models with interactive effects, a fundamental issue in econometric theory. We
focus on the setting where the number of cross-sectional units (N) is large,
but the time dimension (T) remains fixed. While local identification based on
the Jacobian matrix is well understood and relatively straightforward to
establish, achieving global identification remains a significant challenge.
Under a set of mild and easily satisfied conditions, we demonstrate that the
parameters of the model are globally identified, ensuring that no two distinct
parameter values generate the same probability distribution of the observed
data. Our findings contribute to the broader literature on identification in
panel data models and have important implications for empirical research that
relies on interactive effects.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:46:26 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Bai",
"Jushan",
""
],
[
"Mones",
"Pablo",
""
]
] |
2504.14355 | Oleksiy Bystrenko | Oleksiy Bystrenko (1 and 3), Tianxing Sun (1 and 2), Zhaohua Luo (1
and 2), Jingxian Zhang (1 and 2), Yusen Duan (1 and 2), Kaiqing Zhang (1 and
2), Hu Ruan (1 and 2), Wenyu Tang (1 and 2), Bohdan Ilkiv (3), Svitlana
Petrovska (3), Tetiana Bystrenko (3) ((1) Shanghai Institute of Ceramics of
Chinese Academy of Sciences, Shanghai, China, (2) University of Chinese
Academy of Sciences, Beijing, China, (3) Frantsevich Institute for Problems
of Materials Science, Kyiv, Ukraine) | Mechanical properties of B12-based orthorhombic metal carboborides. A
first principle study | 14 pages, 10 figures | null | null | null | cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Structural and mechanical properties of B12-based orthorhombic metal
carboborides are studied on the basis of first principle DFT approach. The
simulations predict the existence of a new family of phases of the composition
MeC2B12 (Me=Mg, Ca, Sr, Sc, Y) with similar structure and space symmetry Imma.
It has been found that the predicted phases are thermally (dynamically) stable
and have considerably better mechanical properties as compared to the reference
compound AlMgB14. The respective calculated isotropic elastic moduli and
Vickers hardness are significantly higher (G~ 230-250, E~530-550, and Hv~35-55
GPa). These conclusions were confirmed by direct calculations of shear strength
for the above phases, which demonstrated the increase of 30-50% in different
directions. The enhanced mechanical characteristics of the MgC2B12 -based
phases make them promising for creating novel superhard materials
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:47:50 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Bystrenko",
"Oleksiy",
"",
"1 and 3"
],
[
"Sun",
"Tianxing",
"",
"1 and 2"
],
[
"Luo",
"Zhaohua",
"",
"1\n and 2"
],
[
"Zhang",
"Jingxian",
"",
"1 and 2"
],
[
"Duan",
"Yusen",
"",
"1 and 2"
],
[
"Zhang",
"Kaiqing",
"",
"1 and\n 2"
],
[
"Ruan",
"Hu",
"",
"1 and 2"
],
[
"Tang",
"Wenyu",
"",
"1 and 2"
],
[
"Ilkiv",
"Bohdan",
""
],
[
"Petrovska",
"Svitlana",
""
],
[
"Bystrenko",
"Tetiana",
""
]
] |
2504.14356 | Masoud Ataei | Masoud Ataei, Edrin Hasaj, Jacob Gipp, Sepideh Forouzi | Mathematical Programming Models for Exact and Interpretable Formulation
of Neural Networks | null | null | null | null | cs.AI math.OC | http://creativecommons.org/licenses/by/4.0/ | This paper presents a unified mixed-integer programming framework for
training sparse and interpretable neural networks. We develop exact
formulations for both fully connected and convolutional architectures by
modeling nonlinearities such as ReLU activations through binary variables and
encoding structural sparsity via filter- and layer-level pruning constraints.
The resulting models integrate parameter learning, architecture selection, and
structural regularization within a single optimization problem, yielding
globally optimal solutions with respect to a composite objective that balances
prediction accuracy, weight sparsity, and architectural compactness. The
mixed-integer programming formulation accommodates piecewise-linear operations,
including max pooling and activation gating, and permits precise enforcement of
logic-based or domain-specific constraints. By incorporating considerations of
interpretability, sparsity, and verifiability directly into the training
process, the proposed framework bridges a range of research areas including
explainable artificial intelligence, symbolic reasoning, and formal
verification.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:50:17 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Ataei",
"Masoud",
""
],
[
"Hasaj",
"Edrin",
""
],
[
"Gipp",
"Jacob",
""
],
[
"Forouzi",
"Sepideh",
""
]
] |
2504.14357 | Yang Zhang | Yang Zhang, Xiaoming Shi, Suk Hyun Sung, Cong Li, Houbing Huang, Pu
Yu, Ismail El Baggari | Real Space Visualization of Order-Disorder Transition in BaTiO3 | null | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Ferroelectricity in BaTiO3 was observed nearly eighty years ago, but the
mechanism underlying its ferroelectric-paraelectric phase transition remains
elusive. The order-disorder transition has been recognized as playing a
critical role, however, the precise nature of the order parameter still remains
under scrutiny, including the local dipole direction and the correlations above
and below the Curie temperature. Using in situ scanning transmission electron
microscopy, we directly map polar displacements in BaTiO3 across the
ferroelectric-paraelectric phase transition, providing atomistic insights into
a order-disorder mechanism. Atomic tracking reveals finite polar Ti
displacements in the paraelectric phase where they manifest as random polar
nanoregions. The displacements align along <111> direction in both the
ferroelectric and paraelectric phases. The paraelectric-ferroelectric
transition emerges from real-space correlations of the <111> polar Ti
displacements. Our direct visualizations provides atomic insights into the
order-disorder mechanism in the ferroelectric-paraelectric transition of
BaTiO3.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:50:52 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Zhang",
"Yang",
""
],
[
"Shi",
"Xiaoming",
""
],
[
"Sung",
"Suk Hyun",
""
],
[
"Li",
"Cong",
""
],
[
"Huang",
"Houbing",
""
],
[
"Yu",
"Pu",
""
],
[
"Baggari",
"Ismail El",
""
]
] |
2504.14358 | Carlos Quintero-Noda | C. Quintero Noda, J. C. Trelles Arjona, T. del Pino Alem\'an, C.
M\'endez L\'apido, M. J. Mart\'inez Gonz\'alez, T. Felipe, S. Regalado
Olivares, P. G\'omez Gonz\'alez, J. Bienes, J. Quintero Nehrkorn, A.
Matta-G\'omez, M. Barreto, M. Collados | Multi-wavelength spectropolarimetric observations of AR13724 performed
by GRIS | 13 pages, 14 figures | null | null | null | astro-ph.SR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multi-line spectropolarimetric observations allow for the simultaneous
inference of the magnetic field at different layers of the solar atmosphere and
provide insight into how these layers are magnetically coupled. The new upgrade
of the Gregor Infrared Spectrograph (GRIS) instrument offers such a
possibility, allowing for the simultaneous observation of the Ca II line at
8542 A, the Si I line at 10827 A, and the He I triplet at 10830 A in addition
to some additional weaker spectral lines that can probe deeper in the
photosphere. Because these spectral lines are sensitive to the plasma
properties at different regions of the solar atmosphere, their combined
analysis can help understand the stratification of its thermal and magnetic
properties from the photosphere to the chromosphere. This work showcases recent
observations of the upgraded GRIS at the active region AR13724, which shows the
instrument's potential for unravelling the most minute details of solar
phenomena. In particular, we analyse the spatial distribution of the
polarisation signals as well as the distribution of Stokes profiles for
different regimes of the magnetic field strength. We also conduct a preliminary
data analysis using relatively simple and approximate methods.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 16:57:06 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Noda",
"C. Quintero",
""
],
[
"Arjona",
"J. C. Trelles",
""
],
[
"Alemán",
"T. del Pino",
""
],
[
"Lápido",
"C. Méndez",
""
],
[
"González",
"M. J. Martínez",
""
],
[
"Felipe",
"T.",
""
],
[
"Olivares",
"S. Regalado",
""
],
[
"González",
"P. Gómez",
""
],
[
"Bienes",
"J.",
""
],
[
"Nehrkorn",
"J. Quintero",
""
],
[
"Matta-Gómez",
"A.",
""
],
[
"Barreto",
"M.",
""
],
[
"Collados",
"M.",
""
]
] |
2504.14359 | Kyle Buettner | Kyle Buettner, Jacob Emmerson, and Adriana Kovashka | A Multimodal Recaptioning Framework to Account for Perceptual Diversity
in Multilingual Vision-Language Modeling | null | null | null | null | cs.CV cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | There are many ways to describe, name, and group objects when captioning an
image. Differences are evident when speakers come from diverse cultures due to
the unique experiences that shape perception. Machine translation of captions
has pushed multilingual capabilities in vision-language models (VLMs), but data
comes mainly from English speakers, indicating a perceptual bias and lack of
model flexibility. In this work, we address this challenge and outline a
data-efficient framework to instill multilingual VLMs with greater
understanding of perceptual diversity. We specifically propose an LLM-based,
multimodal recaptioning strategy that alters the object descriptions of English
captions before translation. The greatest benefits are demonstrated in a
targeted multimodal mechanism guided by native speaker data. By adding produced
rewrites as augmentations in training, we improve on German and Japanese
text-image retrieval cases studies (up to +3.5 mean recall overall, +4.7 on
non-native error cases). We further propose a mechanism to analyze the specific
object description differences across datasets, and we offer insights into
cross-dataset and cross-language generalization.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 17:23:12 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Buettner",
"Kyle",
""
],
[
"Emmerson",
"Jacob",
""
],
[
"Kovashka",
"Adriana",
""
]
] |
2504.14360 | Shayan Bafandkar | Shayan Bafandkar, Alireza Talebpour | Charging While Driving Lanes: A Boon to Electric Vehicle Owners or a
Disruption to Traffic Flow | null | null | null | null | cs.ET cs.SY eess.SY | http://creativecommons.org/licenses/by/4.0/ | Large-scale adoption of commercial and personal Electric Vehicles (EVs) is
expected to significantly affect traffic flow dynamics, emissions, and energy
consumption in the transportation sector. Range anxiety and challenges
associated with charging EVs are among the key issues that reduce the adoption
rate of EVs and, in turn, limit their system-level impacts. A promising
solution to address these challenges is the introduction of charging while
driving (CWD) lanes. Although technological advancements have made it possible
to charge vehicles wirelessly while driving, introducing such lanes to the
traffic stream can potentially disturb traffic flow and result in new
congestion patterns. This study puts forward a framework to investigate the
effects of CWD lanes on traffic flow, considering %autonomy, speed
harmonization, and environmental factors for different market penetration rates
(MPRs) of personal and commercial EVs. Different policies have been
investigated to suggest the best design for CWD lanes. Results indicate that
introducing CWD lanes can decrease overall traffic throughput and increase
congestion due to additional lane-changing maneuvers by electric vehicles
aiming to utilize the CWD lane. Although higher MPRs of EVs help stabilize
traffic flow and reduce the number of shockwaves, speed disruption tends to
increase in the CWD lane and propagate to adjacent lanes. Emission analyses
show significant reductions (up to 63\%) in pollution levels with increasing
MPRs of personal and commercial EVs. Our analysis shows that while CWD lanes
can facilitate the adoption of EVs, they can deteriorate traffic efficiency,
emphasizing the importance of careful design and policy considerations.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 17:25:41 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Bafandkar",
"Shayan",
""
],
[
"Talebpour",
"Alireza",
""
]
] |
2504.14361 | Ziteng Li | Till Rossner, Ziteng Li, Jonas Balke, Nikoo Salehfard, Tom Seifert,
and Ming Tang | Integrating Single-Cell Foundation Models with Graph Neural Networks for
Drug Response Prediction | 8 pages, 6 figures | null | null | null | cs.LG cs.CL q-bio.QM | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In this study, we propose an innovative methodology for predicting Cancer
Drug Response (CDR) through the integration of the scGPT foundation model
within the DeepCDR model. Our approach utilizes scGPT to generate embeddings
from gene expression data, which are then used as gene expression input data
for DeepCDR. The experimental findings demonstrate the efficacy of this
scGPT-based method in outperforming previous related works, including the
original DeepCDR model and the scFoundation-based model. This study highlights
the potential of scGPT embeddings to enhance the accuracy of CDR predictions
and offers a promising alternative to existing approaches.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 17:35:54 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Rossner",
"Till",
""
],
[
"Li",
"Ziteng",
""
],
[
"Balke",
"Jonas",
""
],
[
"Salehfard",
"Nikoo",
""
],
[
"Seifert",
"Tom",
""
],
[
"Tang",
"Ming",
""
]
] |
2504.14362 | Amin Faraji Astaneh | Amin Faraji Astaneh and Niloofar Vardian | Average Spread Complexity and the Higher-Order Level Spacing | 23 pages, 12 figures | null | null | null | hep-th cond-mat.stat-mech quant-ph | http://creativecommons.org/publicdomain/zero/1.0/ | We investigate the spread complexity of a generic two-level subsystem of a
larger system to analyze the influence of energy level statistics, comparing
chaotic and integrable systems. Initially focusing on the nearest-neighbor
level spacing, we observe the characteristic slope-dip-ramp-plateau structure.
Further investigation reveals that certain matrix models exhibit additional
iterative peaks, motivating us to generalize the known spacing distributions to
higher-order level spacings. While this structure persists in chaotic systems,
we find that integrable systems can also display similar features, highlighting
limitations in using complexity as a universal diagnostic of quantum chaos.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 17:38:11 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Astaneh",
"Amin Faraji",
""
],
[
"Vardian",
"Niloofar",
""
]
] |
2504.14363 | Shihan Dou | Shihan Dou, Muling Wu, Jingwen Xu, Rui Zheng, Tao Gui, Qi Zhang,
Xuanjing Huang | Improving RL Exploration for LLM Reasoning through Retrospective Replay | 13 pages, 3 figures | null | null | null | cs.LG cs.CL | http://creativecommons.org/licenses/by/4.0/ | Reinforcement learning (RL) has increasingly become a pivotal technique in
the post-training of large language models (LLMs). The effective exploration of
the output space is essential for the success of RL. We observe that for
complex problems, during the early stages of training, the model exhibits
strong exploratory capabilities and can identify promising solution ideas.
However, its limited capability at this stage prevents it from successfully
solving these problems. The early suppression of these potentially valuable
solution ideas by the policy gradient hinders the model's ability to revisit
and re-explore these ideas later. Consequently, although the LLM's capabilities
improve in the later stages of training, it still struggles to effectively
address these complex problems. To address this exploration issue, we propose a
novel algorithm named Retrospective Replay-based Reinforcement Learning (RRL),
which introduces a dynamic replay mechanism throughout the training process.
RRL enables the model to revisit promising states identified in the early
stages, thereby improving its efficiency and effectiveness in exploration. To
evaluate the effectiveness of RRL, we conduct extensive experiments on complex
reasoning tasks, including mathematical reasoning and code generation, and
general dialogue tasks. The results indicate that RRL maintains high
exploration efficiency throughout the training period, significantly enhancing
the effectiveness of RL in optimizing LLMs for complicated reasoning tasks.
Moreover, it also improves the performance of RLHF, making the model both safer
and more helpful.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 17:40:04 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Dou",
"Shihan",
""
],
[
"Wu",
"Muling",
""
],
[
"Xu",
"Jingwen",
""
],
[
"Zheng",
"Rui",
""
],
[
"Gui",
"Tao",
""
],
[
"Zhang",
"Qi",
""
],
[
"Huang",
"Xuanjing",
""
]
] |
2504.14364 | Egor Voronetsky | Egor Voronetsky | The Diophantine problem in isotropic reductive groups | Preliminary version | null | null | null | math.NT math.GR math.LO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We begin to study model-theoretic properties of non-split isotropic reductive
group schemes. In this paper we show that the base ring $K$ is e-interpretable
in the point group $G(K)$ of every sufficiently isotropic reductive group
scheme $G$. In particular, the Diophantine problems in $K$ and $G(K)$ are
equivalent. We also compute the centralizer of the elementary subgroup of
$G(K)$ and the common normalizer of all its root subgroups.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 17:40:22 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Voronetsky",
"Egor",
""
]
] |
2504.14365 | Akshat Ramachandran | Akshat Ramachandran, Souvik Kundu, Arnab Raha, Shamik Kundu, Deepak K.
Mathaikutty, Tushar Krishna | Accelerating LLM Inference with Flexible N:M Sparsity via A Fully
Digital Compute-in-Memory Accelerator | null | null | null | null | cs.LG cs.AI cs.AR | http://creativecommons.org/licenses/by/4.0/ | Large language model (LLM) pruning with fixed N:M structured sparsity
significantly limits the expressivity of the sparse model, yielding sub-optimal
performance. In contrast, supporting multiple N:M patterns to provide sparse
representational freedom introduces costly overhead in hardware. To address
these challenges for LLMs, we first present a flexible layer-wise
outlier-density-aware N:M sparsity (FLOW) selection method. FLOW enables the
identification of optimal layer-wise N and M values (from a given range) by
simultaneously accounting for the presence and distribution of outliers,
allowing a higher degree of representational freedom. To deploy sparse models
with such N:M flexibility, we then introduce a flexible, low-overhead digital
compute-in-memory architecture (FlexCiM). FlexCiM supports diverse sparsity
patterns by partitioning a digital CiM (DCiM) macro into smaller sub-macros,
which are adaptively aggregated and disaggregated through distribution and
merging mechanisms for different N and M values. Extensive experiments on both
transformer-based and recurrence-based state space foundation models (SSMs)
demonstrate that FLOW outperforms existing alternatives with an accuracy
improvement of up to 36%, while FlexCiM achieves up to 1.75x lower inference
latency and 1.5x lower energy consumption compared to existing sparse
accelerators. Code is available at: https://github.com/FLOW-open-project/FLOW
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 17:47:01 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Ramachandran",
"Akshat",
""
],
[
"Kundu",
"Souvik",
""
],
[
"Raha",
"Arnab",
""
],
[
"Kundu",
"Shamik",
""
],
[
"Mathaikutty",
"Deepak K.",
""
],
[
"Krishna",
"Tushar",
""
]
] |
2504.14366 | Patrick Haller | Patrick Haller, Jonas Golde, Alan Akbik | Empirical Evaluation of Knowledge Distillation from Transformers to
Subquadratic Language Models | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Knowledge distillation is a widely used technique for compressing large
language models (LLMs) by training a smaller student model to mimic a larger
teacher model. Typically, both the teacher and student are Transformer-based
architectures, leveraging softmax attention for sequence modeling. However, the
quadratic complexity of self-attention at inference time remains a significant
bottleneck, motivating the exploration of subquadratic alternatives such as
structured state-space models (SSMs), linear attention, and recurrent
architectures. In this work, we systematically evaluate the transferability of
knowledge distillation from a Transformer teacher to nine subquadratic student
architectures. Our study aims to determine which subquadratic model best aligns
with the teacher's learned representations and how different architectural
constraints influence the distillation process. We also investigate the impact
of intelligent initialization strategies, including matrix mixing and
query-key-value (QKV) copying, on the adaptation process. Our empirical results
on multiple NLP benchmarks provide insights into the trade-offs between
efficiency and performance, highlighting key factors for successful knowledge
transfer to subquadratic architectures.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 17:49:52 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Haller",
"Patrick",
""
],
[
"Golde",
"Jonas",
""
],
[
"Akbik",
"Alan",
""
]
] |
2504.14367 | Gabriel Santos | Gabriel Machado Santos and Rita Maria da Silva Julia and Marcelo
Zanchetta do Nascimento | Diverse Prompts: Illuminating the Prompt Space of Large Language Models
with MAP-Elites | 8 pages Accepted for publication in IEEE CEC 2025 | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Prompt engineering is essential for optimizing large language models (LLMs),
yet the link between prompt structures and task performance remains
underexplored. This work introduces an evolutionary approach that combines
context-free grammar (CFG) with the MAP-Elites algorithm to systematically
explore the prompt space. Our method prioritizes quality and diversity,
generating high-performing and structurally varied prompts while analyzing
their alignment with diverse tasks by varying traits such as the number of
examples (shots) and reasoning depth. By systematically mapping the phenotypic
space, we reveal how structural variations influence LLM performance, offering
actionable insights for task-specific and adaptable prompt design. Evaluated on
seven BigBench Lite tasks across multiple LLMs, our results underscore the
critical interplay of quality and diversity, advancing the effectiveness and
versatility of LLMs.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 17:50:34 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Santos",
"Gabriel Machado",
""
],
[
"Julia",
"Rita Maria da Silva",
""
],
[
"Nascimento",
"Marcelo Zanchetta do",
""
]
] |
2504.14368 | Shlomi Hod | Shlomi Hod, Lucas Rosenblatt, Julia Stoyanovich | Do You Really Need Public Data? Surrogate Public Data for Differential
Privacy on Tabular Data | null | null | null | null | cs.LG cs.CR | http://creativecommons.org/licenses/by/4.0/ | Differentially private (DP) machine learning often relies on the availability
of public data for tasks like privacy-utility trade-off estimation,
hyperparameter tuning, and pretraining. While public data assumptions may be
reasonable in text and image domains, they are less likely to hold for tabular
data due to tabular data heterogeneity across domains. We propose leveraging
powerful priors to address this limitation; specifically, we synthesize
realistic tabular data directly from schema-level specifications - such as
variable names, types, and permissible ranges - without ever accessing
sensitive records. To that end, this work introduces the notion of "surrogate"
public data - datasets generated independently of sensitive data, which consume
no privacy loss budget and are constructed solely from publicly available
schema or metadata. Surrogate public data are intended to encode plausible
statistical assumptions (informed by publicly available information) into a
dataset with many downstream uses in private mechanisms. We automate the
process of generating surrogate public data with large language models (LLMs);
in particular, we propose two methods: direct record generation as CSV files,
and automated structural causal model (SCM) construction for sampling records.
Through extensive experiments, we demonstrate that surrogate public tabular
data can effectively replace traditional public data when pretraining
differentially private tabular classifiers. To a lesser extent, surrogate
public data are also useful for hyperparameter tuning of DP synthetic data
generators, and for estimating the privacy-utility tradeoff.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 17:55:10 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Hod",
"Shlomi",
""
],
[
"Rosenblatt",
"Lucas",
""
],
[
"Stoyanovich",
"Julia",
""
]
] |
2504.14369 | Pia Loaiza | K. Alfonso, A. Armatol, C. Augier, F. T. Avignone III, O. Azzolini,
A.S. Barabash, G. Bari, A. Barresi, D. Baudin, F. Bellini, G. Benato, L.
Benussi, V. Berest, M. Beretta, L. Berg\'e, M. Bettelli, M. Biassoni, J.
Billard, F. Boffelli, V. Boldrini, E. D. Brandani, C. Brofferio, C. Bucci, M.
Buchynska, J. Camilleri, A. Campani, J. Cao, C. Capelli, S. Capelli, V.
Caracciolo, L. Cardani, P. Carniti, N. Casali, E. Celi, C. Chang, M.
Chapellier, H. Chen, D. Chiesa, D. Cintas, M. Clemenza, I. Colantoni, S.
Copello, O. Cremonesi, R. J. Creswick, A. D'Addabbo, I. Dafinei, F. A.
Danevich, F. DeDominicis, M. De Jesus, P. de Marcillac, S. Dell'Oro, S. Di
Domizio, S. Di Lorenzo, T. Dixon, A. Drobizhev, L. Dumoulin, M. El Idrissi,
M. Faverzani, E. Ferri, F. Ferri, F. Ferroni, E. Figueroa Feliciano, J.
Formaggio, A. Franceschi, S. Fu, B.K. Fujikawa, J.Gascon, S. Ghislandi, A.
Giachero, M. Girola, L. Gironi, A. Giuliani, P. Gorla, C. Gotti, C. Grant, P.
Gras, P. V. Guillaumon, T. D. Gutierrez, K. Han, E. V. Hansen, K. M. Heeger,
D. L. Helis, H. Z. Huang, M. T. Hurst, L. Imbert, A. Juillard, G. Karapetrov,
G. Keppel, H. Khalife, V. V. Kobychev, Yu. G. Kolomensky, R. Kowalski, H.
Lattaud, M. Lefevre, M. Lisovenko, R. Liu, Y. Liu, P. Loaiza, L. Ma, F.
Mancarella, N. Manenti, A. Mariani, L. Marini, S. Marnieros, M. Martinez, R.
H. Maruyama, Ph. Mas, D. Mayer, G. Mazzitelli, E. Mazzola, Y. Mei, M. N.
Moore, S. Morganti, T. Napolitano, M. Nastasi, J. Nikkel, C. Nones, E. B.
Norman, V. Novosad, I. Nutini, T. O'Donnell, E. Olivieri, M. Olmi, B. T.
Oregui, S. Pagan, M. Pageot, L. Pagnanini, D. Pasciuto, L. Pattavina, M.
Pavan, \"O. Penek, H. Peng, G. Pessina, V. Pettinacci, C. Pira, S. Pirro, O.
Pochon, D. V. Poda, T. Polakovic, O. G. Polischuk, E. G. Pottebaum, S. Pozzi,
E. Previtali, A. Puiu, S. Puranam, S. Quitadamo, A. Rappoldi, G. L. Raselli,
A. Ressa, R. Rizzoli, C. Rosenfeld, P. Rosier, M. Rossella, J.A. Scarpaci, B.
Schmidt, R. Serino, A. Shaikina, K. Shang, V. Sharma, V. N. Shlegel, V.
Singh, M. Sisti, P. Slocum, D. Speller, P. T. Surukuchi, L. Taffarello, S.
Tomassini, C. Tomei, A. Torres, J. A. Torres, D. Tozzi, V. I. Tretyak, D.
Trotta, M. Velazquez, K. J. Vetter, S. L. Wagaarachchi, G. Wang, L. Wang, R.
Wang, B. Welliver, J. Wilson, K. Wilson, L. A. Winslow, F. Xie, M. Xue, J.
Yang, V. Yefremenko, V.I. Umatov, M. M. Zarytskyy, T. Zhu, A. Zolotarova, S.
Zucchelli | Sensitivity of the CUPID experiment to $0\nu\beta\beta$ decay of
$^{100}$Mo | null | null | null | null | nucl-ex | http://creativecommons.org/licenses/by/4.0/ | CUPID is a next-generation bolometric experiment to search for neutrinoless
double-beta decay ($0\nu\beta\beta$) of $^{100}$Mo using Li$_2$MoO$_4$
scintillating crystals. It will operate 1596 crystals at $\sim$10 mK in the
CUORE cryostat at the Laboratori Nazionali del Gran Sasso in Italy. Each
crystal will be facing two Ge-based bolometric light detectors for $\alpha$
rejection. We compute the discovery and the exclusion sensitivity of CUPID to
$0\nu\beta\beta$ in a Frequentist and a Bayesian framework. This computation is
done numerically based on pseudo-experiments. For the CUPID baseline scenario,
with a background and an energy resolution of $1.0 \times 10^{-4}$
counts/keV/kg/yr and 5 keV FWHM at the Q-value, respectively, this results in a
Bayesian exclusion sensitivity (90% c.i.) of $\hat{T}_{1/2} > 1.6^{+0.6}_{-0.5}
\times 10^{27} \ \mathrm{yr}$, corresponding to the effective Majorana neutrino
mass of $\hat{m}_{\beta\beta} < \ 9.6$ -- $16.3 \ \mathrm{meV}$. The
Frequentist discovery sensitivity (3$\sigma$) is $\hat{T}_{1/2}= 1.0 \times
10^{27} \ \mathrm{yr}$, corresponding to $\hat{m}_{\beta\beta}= \ 12.2$ --
$20.6 \ \mathrm{meV}$.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 18:01:41 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Alfonso",
"K.",
""
],
[
"Armatol",
"A.",
""
],
[
"Augier",
"C.",
""
],
[
"Avignone",
"F. T.",
"III"
],
[
"Azzolini",
"O.",
""
],
[
"Barabash",
"A. S.",
""
],
[
"Bari",
"G.",
""
],
[
"Barresi",
"A.",
""
],
[
"Baudin",
"D.",
""
],
[
"Bellini",
"F.",
""
],
[
"Benato",
"G.",
""
],
[
"Benussi",
"L.",
""
],
[
"Berest",
"V.",
""
],
[
"Beretta",
"M.",
""
],
[
"Bergé",
"L.",
""
],
[
"Bettelli",
"M.",
""
],
[
"Biassoni",
"M.",
""
],
[
"Billard",
"J.",
""
],
[
"Boffelli",
"F.",
""
],
[
"Boldrini",
"V.",
""
],
[
"Brandani",
"E. D.",
""
],
[
"Brofferio",
"C.",
""
],
[
"Bucci",
"C.",
""
],
[
"Buchynska",
"M.",
""
],
[
"Camilleri",
"J.",
""
],
[
"Campani",
"A.",
""
],
[
"Cao",
"J.",
""
],
[
"Capelli",
"C.",
""
],
[
"Capelli",
"S.",
""
],
[
"Caracciolo",
"V.",
""
],
[
"Cardani",
"L.",
""
],
[
"Carniti",
"P.",
""
],
[
"Casali",
"N.",
""
],
[
"Celi",
"E.",
""
],
[
"Chang",
"C.",
""
],
[
"Chapellier",
"M.",
""
],
[
"Chen",
"H.",
""
],
[
"Chiesa",
"D.",
""
],
[
"Cintas",
"D.",
""
],
[
"Clemenza",
"M.",
""
],
[
"Colantoni",
"I.",
""
],
[
"Copello",
"S.",
""
],
[
"Cremonesi",
"O.",
""
],
[
"Creswick",
"R. J.",
""
],
[
"D'Addabbo",
"A.",
""
],
[
"Dafinei",
"I.",
""
],
[
"Danevich",
"F. A.",
""
],
[
"DeDominicis",
"F.",
""
],
[
"De Jesus",
"M.",
""
],
[
"de Marcillac",
"P.",
""
],
[
"Dell'Oro",
"S.",
""
],
[
"Di Domizio",
"S.",
""
],
[
"Di Lorenzo",
"S.",
""
],
[
"Dixon",
"T.",
""
],
[
"Drobizhev",
"A.",
""
],
[
"Dumoulin",
"L.",
""
],
[
"Idrissi",
"M. El",
""
],
[
"Faverzani",
"M.",
""
],
[
"Ferri",
"E.",
""
],
[
"Ferri",
"F.",
""
],
[
"Ferroni",
"F.",
""
],
[
"Feliciano",
"E. Figueroa",
""
],
[
"Formaggio",
"J.",
""
],
[
"Franceschi",
"A.",
""
],
[
"Fu",
"S.",
""
],
[
"Fujikawa",
"B. K.",
""
],
[
"Gascon",
"J.",
""
],
[
"Ghislandi",
"S.",
""
],
[
"Giachero",
"A.",
""
],
[
"Girola",
"M.",
""
],
[
"Gironi",
"L.",
""
],
[
"Giuliani",
"A.",
""
],
[
"Gorla",
"P.",
""
],
[
"Gotti",
"C.",
""
],
[
"Grant",
"C.",
""
],
[
"Gras",
"P.",
""
],
[
"Guillaumon",
"P. V.",
""
],
[
"Gutierrez",
"T. D.",
""
],
[
"Han",
"K.",
""
],
[
"Hansen",
"E. V.",
""
],
[
"Heeger",
"K. M.",
""
],
[
"Helis",
"D. L.",
""
],
[
"Huang",
"H. Z.",
""
],
[
"Hurst",
"M. T.",
""
],
[
"Imbert",
"L.",
""
],
[
"Juillard",
"A.",
""
],
[
"Karapetrov",
"G.",
""
],
[
"Keppel",
"G.",
""
],
[
"Khalife",
"H.",
""
],
[
"Kobychev",
"V. V.",
""
],
[
"Kolomensky",
"Yu. G.",
""
],
[
"Kowalski",
"R.",
""
],
[
"Lattaud",
"H.",
""
],
[
"Lefevre",
"M.",
""
],
[
"Lisovenko",
"M.",
""
],
[
"Liu",
"R.",
""
],
[
"Liu",
"Y.",
""
],
[
"Loaiza",
"P.",
""
],
[
"Ma",
"L.",
""
],
[
"Mancarella",
"F.",
""
],
[
"Manenti",
"N.",
""
],
[
"Mariani",
"A.",
""
],
[
"Marini",
"L.",
""
],
[
"Marnieros",
"S.",
""
],
[
"Martinez",
"M.",
""
],
[
"Maruyama",
"R. H.",
""
],
[
"Mas",
"Ph.",
""
],
[
"Mayer",
"D.",
""
],
[
"Mazzitelli",
"G.",
""
],
[
"Mazzola",
"E.",
""
],
[
"Mei",
"Y.",
""
],
[
"Moore",
"M. N.",
""
],
[
"Morganti",
"S.",
""
],
[
"Napolitano",
"T.",
""
],
[
"Nastasi",
"M.",
""
],
[
"Nikkel",
"J.",
""
],
[
"Nones",
"C.",
""
],
[
"Norman",
"E. B.",
""
],
[
"Novosad",
"V.",
""
],
[
"Nutini",
"I.",
""
],
[
"O'Donnell",
"T.",
""
],
[
"Olivieri",
"E.",
""
],
[
"Olmi",
"M.",
""
],
[
"Oregui",
"B. T.",
""
],
[
"Pagan",
"S.",
""
],
[
"Pageot",
"M.",
""
],
[
"Pagnanini",
"L.",
""
],
[
"Pasciuto",
"D.",
""
],
[
"Pattavina",
"L.",
""
],
[
"Pavan",
"M.",
""
],
[
"Penek",
"Ö.",
""
],
[
"Peng",
"H.",
""
],
[
"Pessina",
"G.",
""
],
[
"Pettinacci",
"V.",
""
],
[
"Pira",
"C.",
""
],
[
"Pirro",
"S.",
""
],
[
"Pochon",
"O.",
""
],
[
"Poda",
"D. V.",
""
],
[
"Polakovic",
"T.",
""
],
[
"Polischuk",
"O. G.",
""
],
[
"Pottebaum",
"E. G.",
""
],
[
"Pozzi",
"S.",
""
],
[
"Previtali",
"E.",
""
],
[
"Puiu",
"A.",
""
],
[
"Puranam",
"S.",
""
],
[
"Quitadamo",
"S.",
""
],
[
"Rappoldi",
"A.",
""
],
[
"Raselli",
"G. L.",
""
],
[
"Ressa",
"A.",
""
],
[
"Rizzoli",
"R.",
""
],
[
"Rosenfeld",
"C.",
""
],
[
"Rosier",
"P.",
""
],
[
"Rossella",
"M.",
""
],
[
"Scarpaci",
"J. A.",
""
],
[
"Schmidt",
"B.",
""
],
[
"Serino",
"R.",
""
],
[
"Shaikina",
"A.",
""
],
[
"Shang",
"K.",
""
],
[
"Sharma",
"V.",
""
],
[
"Shlegel",
"V. N.",
""
],
[
"Singh",
"V.",
""
],
[
"Sisti",
"M.",
""
],
[
"Slocum",
"P.",
""
],
[
"Speller",
"D.",
""
],
[
"Surukuchi",
"P. T.",
""
],
[
"Taffarello",
"L.",
""
],
[
"Tomassini",
"S.",
""
],
[
"Tomei",
"C.",
""
],
[
"Torres",
"A.",
""
],
[
"Torres",
"J. A.",
""
],
[
"Tozzi",
"D.",
""
],
[
"Tretyak",
"V. I.",
""
],
[
"Trotta",
"D.",
""
],
[
"Velazquez",
"M.",
""
],
[
"Vetter",
"K. J.",
""
],
[
"Wagaarachchi",
"S. L.",
""
],
[
"Wang",
"G.",
""
],
[
"Wang",
"L.",
""
],
[
"Wang",
"R.",
""
],
[
"Welliver",
"B.",
""
],
[
"Wilson",
"J.",
""
],
[
"Wilson",
"K.",
""
],
[
"Winslow",
"L. A.",
""
],
[
"Xie",
"F.",
""
],
[
"Xue",
"M.",
""
],
[
"Yang",
"J.",
""
],
[
"Yefremenko",
"V.",
""
],
[
"Umatov",
"V. I.",
""
],
[
"Zarytskyy",
"M. M.",
""
],
[
"Zhu",
"T.",
""
],
[
"Zolotarova",
"A.",
""
],
[
"Zucchelli",
"S.",
""
]
] |
2504.14370 | Fan Wei | Jon Kleinberg and Fan Wei | Density Measures for Language Generation | null | null | null | null | math.CO cs.CL cs.DM cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The recent successes of large language models (LLMs) have led to a surge of
theoretical research into language generation. A recent line of work proposes
an abstract view, called language generation in the limit, where generation is
seen as a game between an adversary and an algorithm: the adversary generates
strings from an unknown language $K$, chosen from a countable collection of
candidate languages, and after seeing a finite set of these strings, the
algorithm must generate new strings from $K$ that it has not seen before. This
formalism highlights a key tension: the trade-off between validity (the
algorithm should only produce strings from the language) and breadth (it should
be able to produce many strings from the language). This trade-off is central
in applied language generation as well, where it appears as a balance between
hallucination (generating invalid utterances) and mode collapse (generating
only a restricted set of outputs). Despite its importance, this trade-off has
been challenging to study quantitatively. We develop ways to quantify this
trade-off by formalizing breadth using measures of density. Existing algorithms
for language generation in the limit produce output sets that can have zero
density in the true language, and this important failure of breadth might seem
unavoidable. We show, however, that such a failure is not necessary: we provide
an algorithm for language generation in the limit whose outputs have strictly
positive density in $K$. We also study the internal representations built by
these algorithms, specifically the sequence of hypothesized candidate languages
they consider, and show that achieving the strongest form of breadth may
require oscillating indefinitely between high- and low-density representations.
Our analysis introduces a novel topology on language families, with notions of
convergence and limit points playing a key role.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 18:08:18 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Kleinberg",
"Jon",
""
],
[
"Wei",
"Fan",
""
]
] |
2504.14371 | Bosong Chai | Peixi Wu, Bosong Chai, Menghua Zheng, Wei Li, Zhangchi Hu, Jie Chen,
Zheyu Zhang, Hebei Li, Xiaoyan Sun | Efficient Spiking Point Mamba for Point Cloud Analysis | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Bio-inspired Spiking Neural Networks (SNNs) provide an energy-efficient way
to extract 3D spatio-temporal features. However, existing 3D SNNs have
struggled with long-range dependencies until the recent emergence of Mamba,
which offers superior computational efficiency and sequence modeling
capability. In this work, we propose Spiking Point Mamba (SPM), the first
Mamba-based SNN in the 3D domain. Due to the poor performance of simply
transferring Mamba to 3D SNNs, SPM is designed to utilize both the sequence
modeling capabilities of Mamba and the temporal feature extraction of SNNs.
Specifically, we first introduce Hierarchical Dynamic Encoding (HDE), an
improved direct encoding method that effectively introduces dynamic temporal
mechanism, thereby facilitating temporal interactions. Then, we propose a
Spiking Mamba Block (SMB), which builds upon Mamba while learning
inter-time-step features and minimizing information loss caused by spikes.
Finally, to further enhance model performance, we adopt an asymmetric SNN-ANN
architecture for spike-based pre-training and finetune. Compared with the
previous state-of-the-art SNN models, SPM improves OA by +6.2%, +6.1%, and
+7.4% on three variants of ScanObjectNN, and boosts instance mIOU by +1.9% on
ShapeNetPart. Meanwhile, its energy consumption is at least 3.5x lower than
that of its ANN counterpart. The code will be made publicly available.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 18:14:35 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Wu",
"Peixi",
""
],
[
"Chai",
"Bosong",
""
],
[
"Zheng",
"Menghua",
""
],
[
"Li",
"Wei",
""
],
[
"Hu",
"Zhangchi",
""
],
[
"Chen",
"Jie",
""
],
[
"Zhang",
"Zheyu",
""
],
[
"Li",
"Hebei",
""
],
[
"Sun",
"Xiaoyan",
""
]
] |
2504.14372 | Jose Marie Antonio Minoza | Jose Marie Antonio Minoza | Learning Enhanced Structural Representations with Block-Based
Uncertainties for Ocean Floor Mapping | null | Tackling Climate Change with Machine Learning Workshop, ICLR 2025 | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Accurate ocean modeling and coastal hazard prediction depend on
high-resolution bathymetric data; yet, current worldwide datasets are too
coarse for exact numerical simulations. While recent deep learning advances
have improved earth observation data resolution, existing methods struggle with
the unique challenges of producing detailed ocean floor maps, especially in
maintaining physical structure consistency and quantifying uncertainties. This
work presents a novel uncertainty-aware mechanism using spatial blocks to
efficiently capture local bathymetric complexity based on block-based conformal
prediction. Using the Vector Quantized Variational Autoencoder (VQ-VAE)
architecture, the integration of this uncertainty quantification framework
yields spatially adaptive confidence estimates while preserving topographical
features via discrete latent representations. With smaller uncertainty widths
in well-characterized areas and appropriately larger bounds in areas of complex
seafloor structures, the block-based design adapts uncertainty estimates to
local bathymetric complexity. Compared to conventional techniques, experimental
results over several ocean regions show notable increases in both
reconstruction quality and uncertainty estimation reliability. This framework
increases the reliability of bathymetric reconstructions by preserving
structural integrity while offering spatially adaptive uncertainty estimates,
so opening the path for more solid climate modeling and coastal hazard
assessment.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 18:20:08 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Minoza",
"Jose Marie Antonio",
""
]
] |
2504.14374 | Max L\"ubke | Max L\"ubke, Marco De Lucia, Stefan Petri, Bettina Schnor | A fast MPI-based Distributed Hash-Table as Surrogate Model demonstrated
in a coupled reactive transport HPC simulation | Long version, 15 pages, 6 figures; Short version (8 pages) submitted
to "25th International Conference on Computational Science" (ICCS25) | null | null | null | cs.DC | http://creativecommons.org/licenses/by/4.0/ | Surrogate models can play a pivotal role in enhancing performance in
contemporary High-Performance Computing applications. Cache-based surrogates
use already calculated simulation results to interpolate or extrapolate further
simulation output values. But this approach only pays off if the access time to
retrieve the needed values is much faster than the actual simulation. While the
most existing key-value stores use a Client-Server architecture with dedicated
storage nodes, this is not the most suitable architecture for HPC applications.
Instead, we propose a distributed architecture where the parallel processes
offer a part of their available memory to build a shared distributed hash table
based on MPI. This paper presents three DHT approaches with the special
requirements of HPC applications in mind. The presented lock-free design
outperforms both DHT versions which use explicit synchronization by
coarse-grained resp. fine-grained locking. The lock-free DHT shows very good
scaling regarding read and write performance. The runtime of a coupled reactive
transport simulation was improved between 14% and 42% using the lock-free DHT
as a surrogate model.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 18:25:20 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Lübke",
"Max",
""
],
[
"De Lucia",
"Marco",
""
],
[
"Petri",
"Stefan",
""
],
[
"Schnor",
"Bettina",
""
]
] |
2504.14375 | Kun Qian | Kun Qian, Maximillian Chen, Siyan Li, Arpit Sharma, Zhou Yu | Bottom-Up Synthesis of Knowledge-Grounded Task-Oriented Dialogues with
Iteratively Self-Refined Prompts | Accepted by NAACL 2025 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Training conversational question-answering (QA) systems requires a
substantial amount of in-domain data, which is often scarce in practice. A
common solution to this challenge is to generate synthetic data. Traditional
methods typically follow a top-down approach, where a large language model
(LLM) generates multi-turn dialogues from a broad prompt. Although this method
produces coherent conversations, it offers limited fine-grained control over
the content and is susceptible to hallucinations. We introduce a bottom-up
conversation synthesis approach, where QA pairs are generated first and then
combined into a coherent dialogue. This method offers greater control and
precision by dividing the process into two distinct steps, allowing refined
instructions and validations to be handled separately. Additionally, this
structure allows the use of non-local models in stages that do not involve
proprietary knowledge, enhancing the overall quality of the generated data.
Both human and automated evaluations demonstrate that our approach produces
more realistic and higher-quality dialogues compared to top-down methods.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 18:25:53 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Qian",
"Kun",
""
],
[
"Chen",
"Maximillian",
""
],
[
"Li",
"Siyan",
""
],
[
"Sharma",
"Arpit",
""
],
[
"Yu",
"Zhou",
""
]
] |
2504.14376 | Mohammed Shalaby | Mohammed Ayman Shalaby, Syed Shabbir Ahmed, Nicholas Dahdah, Charles
Champagne Cossette, Jerome Le Ny, and James Richard Forbes | MILUV: A Multi-UAV Indoor Localization dataset with UWB and Vision | 18 pages, 15 figures | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper introduces MILUV, a Multi-UAV Indoor Localization dataset with UWB
and Vision measurements. This dataset comprises 217 minutes of flight time over
36 experiments using three quadcopters, collecting ultra-wideband (UWB) ranging
data such as the raw timestamps and channel-impulse response data, vision data
from a stereo camera and a bottom-facing monocular camera, inertial measurement
unit data, height measurements from a laser rangefinder, magnetometer data, and
ground-truth poses from a motion-capture system. The UWB data is collected from
up to 12 transceivers affixed to mobile robots and static tripods in both
line-of-sight and non-line-of-sight conditions. The UAVs fly at a maximum speed
of 4.418 m/s in an indoor environment with visual fiducial markers as features.
MILUV is versatile and can be used for a wide range of applications beyond
localization, but the primary purpose of MILUV is for testing and validating
multi-robot UWB- and vision-based localization algorithms. The dataset can be
downloaded at https://doi.org/10.25452/figshare.plus.28386041.v1. A development
kit is presented alongside the MILUV dataset, which includes benchmarking
algorithms such as visual-inertial odometry, UWB-based localization using an
extended Kalman filter, and classification of CIR data using machine learning
approaches. The development kit can be found at
https://github.com/decargroup/miluv, and is supplemented with a website
available at https://decargroup.github.io/miluv/.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 18:29:59 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Shalaby",
"Mohammed Ayman",
""
],
[
"Ahmed",
"Syed Shabbir",
""
],
[
"Dahdah",
"Nicholas",
""
],
[
"Cossette",
"Charles Champagne",
""
],
[
"Ny",
"Jerome Le",
""
],
[
"Forbes",
"James Richard",
""
]
] |
2504.14377 | Anne Rubbens | Anne Rubbens, Julien M. Hendrickx, Adrien Taylor | A constructive approach to strengthen algebraic descriptions of function
and operator classes | null | null | null | null | math.OC | http://creativecommons.org/licenses/by/4.0/ | It is well known that functions (resp. operators) satisfying a property~$p$
on a subset $Q\subset \mathbb{R}^d$ cannot necessarily be extended to a
function (resp. operator) satisfying~$p$ on the whole of~$\mathbb{R}^d$. Given
$Q \subseteq \mathbb{R}^d$, this work considers the problem of obtaining
necessary and ideally sufficient conditions to be satisfied by a function
(resp. operator) on $Q$, ensuring the existence of an extension of this
function (resp. operator) satisfying $p$ on $\mathbb{R}^d$.
More precisely, given some property $p$, we present a refinement procedure to
obtain stronger necessary conditions to be imposed on $Q$. This procedure can
be applied iteratively until the stronger conditions are also sufficient. We
illustrate the procedure on a few examples, including the strengthening of
existing descriptions for the classes of smooth functions satisfying a
\L{}ojasiewicz condition, convex blockwise smooth functions, Lipschitz monotone
operators, strongly monotone cocoercive operators, and uniformly convex
functions.
In most cases, these strengthened descriptions can be represented, or
relaxed, to semi-definite constraints, which can be used to formulate tractable
optimization problems on functions (resp. operators) within those classes.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 18:36:26 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Rubbens",
"Anne",
""
],
[
"Hendrickx",
"Julien M.",
""
],
[
"Taylor",
"Adrien",
""
]
] |
2504.14378 | Yue Li | Yue Li, Ye Wei, Alaukik Saxena, Markus K\"uhbach, Christoph Freysoldt,
Baptiste Gault | Machine learning enhanced atom probe tomography analysis: a snapshot
review | null | null | null | null | cond-mat.mtrl-sci cs.LG | http://creativecommons.org/licenses/by/4.0/ | Atom probe tomography (APT) is a burgeoning characterization technique that
provides compositional mapping of materials in three-dimensions at near-atomic
scale. Since its significant expansion in the past 30 years, we estimate that
one million APT datasets have been collected, each containing millions to
billions of individual ions. Their analysis and the extraction of
microstructural information has largely relied upon individual users whose
varied level of expertise causes clear and documented bias. Current practices
hinder efficient data processing, and make challenging standardization and the
deployment of data analysis workflows that would be compliant with FAIR data
principles. Over the past decade, building upon the long-standing expertise of
the APT community in the development of advanced data processing or data mining
techniques, there has been a surge of novel machine learning (ML) approaches
aiming for user-independence, and that are efficient, reproducible, and robust
from a statistics perspective. Here, we provide a snapshot review of this
rapidly evolving field. We begin with a brief introduction to APT and the
nature of the APT data. This is followed by an overview of relevant ML
algorithms and a comprehensive review of their applications to APT. We also
discuss how ML can enable discoveries beyond human capability, offering new
insights into the mechanisms within materials. Finally, we provide guidance for
future directions in this domain.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 18:37:26 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Li",
"Yue",
""
],
[
"Wei",
"Ye",
""
],
[
"Saxena",
"Alaukik",
""
],
[
"Kühbach",
"Markus",
""
],
[
"Freysoldt",
"Christoph",
""
],
[
"Gault",
"Baptiste",
""
]
] |
2504.14379 | Andrew Lee | Andrew Lee, Lihao Sun, Chris Wendler, Fernanda Vi\'egas, Martin
Wattenberg | The Geometry of Self-Verification in a Task-Specific Reasoning Model | null | null | null | null | cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | How do reasoning models verify their own answers? We study this question by
training a model using DeepSeek R1's recipe on the CountDown task. We leverage
the fact that preference tuning leads to mode collapse, resulting in a model
that always produces highly structured and easily parse-able chain-of-thought
sequences. With this setup, we do a top-down and bottom-up analysis to
reverse-engineer how the model verifies its outputs. Our top-down analysis
reveals Gated Linear Unit (GLU) weights encoding verification-related tokens,
such as ``success'' or ``incorrect'', which activate according to the
correctness of the model's reasoning steps. Our bottom-up analysis reveals that
``previous-token heads'' are mainly responsible for model verification. Our
analyses meet in the middle: drawing inspiration from inter-layer communication
channels, we use the identified GLU vectors to localize as few as three
attention heads that can disable model verification, pointing to a necessary
component of a potentially larger verification circuit.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 18:40:51 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Lee",
"Andrew",
""
],
[
"Sun",
"Lihao",
""
],
[
"Wendler",
"Chris",
""
],
[
"Viégas",
"Fernanda",
""
],
[
"Wattenberg",
"Martin",
""
]
] |
2504.14380 | David Blanik | David Blanik, Jos\'e Garre-Rubio, Norbert Schuch | Gauging Quantum Phases: A Matrix Product State Approach | 23 pages, 2 figures | null | null | null | quant-ph cond-mat.str-el | http://creativecommons.org/licenses/by/4.0/ | Utilizing the framework of matrix product states, we investigate gauging as a
method for exploring quantum phases of matter. Specifically, we describe how
symmetry-protected topological (SPT) phases and spontaneous symmetry breaking
(SSB) phases in one-dimensional spin systems behave under twisted gauging, a
generalization of the well-known gauging procedure for globally symmetric
states. Compared to previous, order parameter--based, approaches our analysis
is not limited to the case of maximally non-commutative (MNC) phases and we use
our findings to propose a generalization of the Kennedy-Tasaki transformation
to the non-MNC setting. A key result of our work is that gauging produces
configurations characterized by a combination of MNC order and symmetry
breaking, when applied to non-MNC SPT phases. More generally, we conjecture a
precise correspondence between SSB and non-MNC SPT phases, possibly enabling
the detection of such phases using local and string order parameters.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:04:26 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Blanik",
"David",
""
],
[
"Garre-Rubio",
"José",
""
],
[
"Schuch",
"Norbert",
""
]
] |
2504.14381 | Khuong Nguyen-An | Pham Nhat Minh, Khoa Nguyen, Willy Susilo, and Khuong Nguyen-An | Publicly Verifiable Secret Sharing: Generic Constructions and
Lattice-Based Instantiations in the Standard Model | 33 pages, 4 appendices | null | null | null | cs.CR cs.IT math.IT math.NT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Publicly verifiable secret sharing (PVSS) allows a dealer to share a secret
among a set of shareholders so that the secret can be reconstructed later from
any set of qualified participants. In addition, any public verifier should be
able to check the correctness of the sharing and reconstruction process. PVSS
has been demonstrated to yield various applications, such as e-voting,
distributed key generation, decentralized random number generation protocols,
and multi-party computation. Although many concrete PVSS protocols have been
proposed, their security is either proven in the random oracle model or relies
on quantum-vulnerable assumptions such as factoring or discrete logarithm. In
this work, we put forward a generic construction for PVSS that can be
instantiated in the standard model under the Learning With Errors (LWE)
assumption. Our instantiation provides the first post-quantum PVSS in the
standard model, with a reasonable level of asymptotic efficiency.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:05:41 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Minh",
"Pham Nhat",
""
],
[
"Nguyen",
"Khoa",
""
],
[
"Susilo",
"Willy",
""
],
[
"Nguyen-An",
"Khuong",
""
]
] |
2504.14382 | Madhuparna Pal | Sagnik Chakraborty, Madhuparna Pal | Monomial retracts of polynomial rings are polynomial rings | null | null | null | null | math.AC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Let $R$ be a ring and $B = R[X_1, \dots, X_n]$ the polynomial ring in $n$
variables over $R$. In this article, we consider retractions $\varphi : B
\longrightarrow B$ such that $\varphi(X_i)$ is either a monic monomial or $0$.
We prove that if $R$ is an integral domain, then any such retract is isomorphic
to $R^{[p]}$, the polynomial ring in $p$ variables over $R$, for some $0 \le p
\le n$. We also characterize different monomial retractions of $B$ which give
the same retract.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:07:02 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Chakraborty",
"Sagnik",
""
],
[
"Pal",
"Madhuparna",
""
]
] |
2504.14384 | Sam Nelson | Sam Nelson | A Quick Note On Homsets and Diagrams | Just a short note | null | null | null | math.GT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The homset invariant of a knot or link L with respect to an algebraic knot
coloring structure X can be identified with a set of colorings of a diagram of
L by elements of X via an identification of diagrammatic generators with
algebraic generators. In some cases, particularly when either L or X has a high
degree of symmetry, distinct homset elements can be represented by
superficially similar X-colored diagrams, potentially leading to confusion. In
this short note we examine and attempt to clarify this phenomenon.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:16:03 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Nelson",
"Sam",
""
]
] |
2504.14385 | Olga Maleva | Jan Kol\'a\v{r} and Olga Maleva | Chain rule for pointwise Lipschitz mappings | null | null | null | null | math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The classical Chain Rule formula $(f\circ g)'(x;u)=f'(g(x);g'(x;u))$ gives
the (partial, or directional) derivative of the composition of mappings $f$ and
$g$. We show how to get rid of the unnecessarily strong assumption of
differentiability at all of the relevant points: the mappings do not need to be
defined on the whole space and it is enough for them to be pointwise Lipschitz.
The price to pay is that the Chain Rule holds almost everywhere. We extend this
construction to infinite-dimensional spaces with good properties (Banach,
separable, Radon-Nikod\'ym) with an appropriate notion of almost everywhere.
Pointwise Lipschitzness is a local condition in contrast to the global
Lipschitz property: we do not need the mappings to be defined on the whole
space, or even locally in a neighbourhood, nor to know their behaviour far away
from the points we consider. This distinguishes our results from recent
research on the differentiation of the composition of Lipschitz mappings. The
methods we develop for the purpose of proving the Chain Rule also allow us to
strengthen the Rademacher-Stepanov type theorem on almost everywhere
differentiability of a mapping.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:16:54 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Kolář",
"Jan",
""
],
[
"Maleva",
"Olga",
""
]
] |
2504.14386 | Md Abtahi Majeed Chowdhury | Md Abtahi Majeed Chowdhury, Md Rifat Ur Rahman, Akil Ahmad Taki | LOOPE: Learnable Optimal Patch Order in Positional Embeddings for Vision
Transformers | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Positional embeddings (PE) play a crucial role in Vision Transformers (ViTs)
by providing spatial information otherwise lost due to the permutation
invariant nature of self attention. While absolute positional embeddings (APE)
have shown theoretical advantages over relative positional embeddings (RPE),
particularly due to the ability of sinusoidal functions to preserve spatial
inductive biases like monotonicity and shift invariance, a fundamental
challenge arises when mapping a 2D grid to a 1D sequence. Existing methods have
mostly overlooked or never explored the impact of patch ordering in positional
embeddings. To address this, we propose LOOPE, a learnable patch-ordering
method that optimizes spatial representation for a given set of frequencies,
providing a principled approach to patch order optimization. Empirical results
show that our PE significantly improves classification accuracy across various
ViT architectures. To rigorously evaluate the effectiveness of positional
embeddings, we introduce the "Three Cell Experiment", a novel benchmarking
framework that assesses the ability of PEs to retain relative and absolute
positional information across different ViT architectures. Unlike standard
evaluations, which typically report a performance gap of 4 to 6% between models
with and without PE, our method reveals a striking 30 to 35% difference,
offering a more sensitive diagnostic tool to measure the efficacy of PEs. Our
experimental analysis confirms that the proposed LOOPE demonstrates enhanced
effectiveness in retaining both relative and absolute positional information.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:20:47 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Chowdhury",
"Md Abtahi Majeed",
""
],
[
"Rahman",
"Md Rifat Ur",
""
],
[
"Taki",
"Akil Ahmad",
""
]
] |
2504.14387 | Sam Nelson | Anusha Kabra and Sam Nelson | New Polynomial Decategorifications of the Quandle Coloring Quiver | 9 pages | null | null | null | math.GT math.QA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce two new families of polynomial invariants of oriented classical
and virtual knots and links defined as decategorfications of the quandle
coloring quiver. We provide examples to illustrate the computation of the
invariants, show that they are not determined by the quandle counting
invariant, and compute tables of invariants values for some small quandles.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:23:58 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Kabra",
"Anusha",
""
],
[
"Nelson",
"Sam",
""
]
] |
2504.14388 | Xiaoyang Wang | Xiaoyang Wang, Christopher C. Yang | Balancing Fairness and Performance in Healthcare AI: A Gradient
Reconciliation Approach | Accepted to the 23rd International Conference on Artificial
Intelligence in Medicine (AIME 2025) | null | null | null | cs.LG cs.CY | http://creativecommons.org/licenses/by/4.0/ | The rapid growth of healthcare data and advances in computational power have
accelerated the adoption of artificial intelligence (AI) in medicine. However,
AI systems deployed without explicit fairness considerations risk exacerbating
existing healthcare disparities, potentially leading to inequitable resource
allocation and diagnostic disparities across demographic subgroups. To address
this challenge, we propose FairGrad, a novel gradient reconciliation framework
that automatically balances predictive performance and multi-attribute fairness
optimization in healthcare AI models. Our method resolves conflicting
optimization objectives by projecting each gradient vector onto the orthogonal
plane of the others, thereby regularizing the optimization trajectory to ensure
equitable consideration of all objectives. Evaluated on diverse real-world
healthcare datasets and predictive tasks - including Substance Use Disorder
(SUD) treatment and sepsis mortality - FairGrad achieved statistically
significant improvements in multi-attribute fairness metrics (e.g., equalized
odds) while maintaining competitive predictive accuracy. These results
demonstrate the viability of harmonizing fairness and utility in
mission-critical medical AI applications.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:24:34 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Wang",
"Xiaoyang",
""
],
[
"Yang",
"Christopher C.",
""
]
] |
2504.14389 | Kartal Nagy | Kartal Nagy | A new intersection condition in extremal set theory | 15 pages | null | null | null | math.CO | http://creativecommons.org/licenses/by/4.0/ | We call a family $\mathcal{F}$ $(3,2,\ell)$-intersecting if $|A \cap B|+|B
\cap C|+|C \cap A| \geq \ell$ for all $A$, $B$, $C \in \mathcal{F}$. We try to
look for the maximum size of such a family $\mathcal{F}$ in case when
$\mathcal{F} \subset {[n] \choose k}$ or $\mathcal{F} \subset 2^{[n]}$. In the
uniform case we show that if $\mathcal{F}$ is $(3,2,2)$-intersecting, then
$\vert \mathcal{F} \vert \leq {n+1 \choose k-1}+{n \choose k-2}$ and if
$\mathcal{F}$ is $(3,2,3)$-intersecting, then $|\mathcal{F}| \leq {n \choose
k-1} + 2 {n \choose k-3} + 3 {n-1 \choose k-3}$. For the lower bound we
construct a $(3,2,\ell)$-intersecting family and we show that this bound is
sharp when $\ell=2$ or $3$ and $n$ is sufficiently large compared to $k$. In
the non-uniform case we give an upper bound for a $(3,2,n-x)$-intersecting
family, when $n$ is sufficiently large compared to $x$.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:30:36 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Nagy",
"Kartal",
""
]
] |
2504.14390 | Grzegorz Gutowski | Steven Chaplick, Grzegorz Gutowski, Tomasz Krawczyk | A Note on the Complexity of Defensive Domination | null | null | null | null | cs.CC | http://creativecommons.org/licenses/by/4.0/ | In a graph G, a k-attack A is any set of at most k vertices and l-defense D
is a set of at most l vertices. We say that defense D counters attack A if each
a in A can be matched to a distinct defender d in D with a equal to d or a
adjacent to d in G. In the defensive domination problem, we are interested in
deciding, for a graph G and positive integers k and l given on input, if there
exists an l-defense that counters every possible k-attack on G. Defensive
domination is a natural resource allocation problem and can be used to model
network robustness and security, disaster response strategies, and redundancy
designs.
The defensive domination problem is naturally in the complexity class
$\Sigma^P_2$. The problem was known to be NP-hard in general, and
polynomial-time algorithms were found for some restricted graph classes. In
this note we prove that the defensive domination problem is
$\Sigma^P_2$-complete. We also introduce a natural variant of the defensive
domination problem in which the defense is allowed to be a multiset of
vertices. This variant is also $\Sigma^P_2$-complete, but we show that it
admits a polynomial-time algorithm in the class of interval graphs. A similar
result was known for the original setting in the class of proper interval
graphs.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:30:59 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Chaplick",
"Steven",
""
],
[
"Gutowski",
"Grzegorz",
""
],
[
"Krawczyk",
"Tomasz",
""
]
] |
2504.14391 | Rahul Thapa | Rahul Thapa, Andrew Li, Qingyang Wu, Bryan He, Yuki Sahashi, Christina
Binder, Angela Zhang, Ben Athiwaratkun, Shuaiwen Leon Song, David Ouyang,
James Zou | How Well Can General Vision-Language Models Learn Medicine By Watching
Public Educational Videos? | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Publicly available biomedical videos, such as those on YouTube, serve as
valuable educational resources for medical students. Unlike standard machine
learning datasets, these videos are designed for human learners, often mixing
medical imagery with narration, explanatory diagrams, and contextual framing.
In this work, we investigate whether such pedagogically rich, yet
non-standardized and heterogeneous videos can effectively teach general-domain
vision-language models biomedical knowledge. To this end, we introduce
OpenBiomedVi, a biomedical video instruction tuning dataset comprising 1031
hours of video-caption and Q/A pairs, curated through a multi-step
human-in-the-loop pipeline. Diverse biomedical video datasets are rare, and
OpenBiomedVid fills an important gap by providing instruction-style supervision
grounded in real-world educational content. Surprisingly, despite the informal
and heterogeneous nature of these videos, the fine-tuned Qwen-2-VL models
exhibit substantial performance improvements across most benchmarks. The 2B
model achieves gains of 98.7% on video tasks, 71.2% on image tasks, and 0.2% on
text tasks. The 7B model shows improvements of 37.09% on video and 11.2% on
image tasks, with a slight degradation of 2.7% on text tasks compared to their
respective base models. To address the lack of standardized biomedical video
evaluation datasets, we also introduce two new expert curated benchmarks,
MIMICEchoQA and SurgeryVideoQA. On these benchmarks, the 2B model achieves
gains of 99.1% and 98.1%, while the 7B model shows gains of 22.5% and 52.1%,
respectively, demonstrating the models' ability to generalize and perform
biomedical video understanding on cleaner and more standardized datasets than
those seen during training. These results suggest that educational videos
created for human learning offer a surprisingly effective training signal for
biomedical VLMs.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:32:15 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Thapa",
"Rahul",
""
],
[
"Li",
"Andrew",
""
],
[
"Wu",
"Qingyang",
""
],
[
"He",
"Bryan",
""
],
[
"Sahashi",
"Yuki",
""
],
[
"Binder",
"Christina",
""
],
[
"Zhang",
"Angela",
""
],
[
"Athiwaratkun",
"Ben",
""
],
[
"Song",
"Shuaiwen Leon",
""
],
[
"Ouyang",
"David",
""
],
[
"Zou",
"James",
""
]
] |
2504.14392 | Liangjun Weng | Xinqun Mei, Guofang Wang, Liangjun Weng | Convex capillary hypersurfaces of prescribed curvature problem | 23 pages | null | null | null | math.DG math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we study the prescribed $k$-th Weingarten curvature problem
for convex capillary hypersurfaces in $\overline{\mathbb{R}^{n+1}_+}$. This
problem naturally extends the prescribed $k$-th Weingarten curvature problem
for closed convex hypersurfaces, previously investigated by Guan-Guan in [19],
to the capillary setting. We reformulate the problem as the solvability of a
Hessian quotient equation with a Robin boundary condition on a spherical cap.
Under a natural sufficient condition, we establish the existence of a strictly
convex capillary hypersurface with the prescribed $k$-th Weingarten curvature.
This also extends our recent work on the capillary Minkowski problem in [40].
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:33:34 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Mei",
"Xinqun",
""
],
[
"Wang",
"Guofang",
""
],
[
"Weng",
"Liangjun",
""
]
] |
2504.14393 | Ashley Tharp | Emily Barnard, Nathan Reading, Ashley M. Tharp | Noncrossing arc diagrams of type B | 41 pages, 17 figures | null | null | null | math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Noncrossing arc diagrams are combinatorial models for permutations that
encode information about lattice congruences of the weak order and about the
associated discrete geometry. In this paper, we consider two related, analogous
models for signed permutations. One model features centrally symmetric
noncrossing arc diagrams, while the other features their quotients modulo the
central symmetry. We demonstrate the utility of the models by applying them to
various questions about lattice quotients of the weak order.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:39:35 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Barnard",
"Emily",
""
],
[
"Reading",
"Nathan",
""
],
[
"Tharp",
"Ashley M.",
""
]
] |
2504.14394 | Chao Chen | Chao Chen and Thomas Chaffey and Rodolphe Sepulchre | Graphical Dominance Analysis for Linear Systems: A Frequency-Domain
Approach | null | null | null | null | eess.SY cs.SY math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose a frequency-domain approach to dominance analysis for multi-input
multi-output (MIMO) linear time-invariant systems. The dominance of a MIMO
system is defined to be the number of its poles in the open right half-plane.
Our approach is graphical: we define a frequency-wise notion of the
recently-introduced scaled graph of a MIMO system plotted in a complex plane.
The scaled graph provides a bound of the eigenloci of the system, which can be
viewed as a robust MIMO extension of the classical Nyquist plot. Our main
results characterize sufficient conditions for quantifying the dominance of a
closed-loop system based upon separation of scaled graphs of two open-loop
systems in a frequency-wise manner. The results reconcile existing small gain,
small phase and passivity theorems for feedback dominance analysis.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:40:37 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Chen",
"Chao",
""
],
[
"Chaffey",
"Thomas",
""
],
[
"Sepulchre",
"Rodolphe",
""
]
] |
2504.14395 | Chung-En Yu | Chung-En (Johnny) Yu, Hsuan-Chih (Neil) Chen, Brian Jalaian, Nathaniel
D. Bastian | Hydra: An Agentic Reasoning Approach for Enhancing Adversarial
Robustness and Mitigating Hallucinations in Vision-Language Models | null | null | null | null | cs.CV cs.AI cs.MA | http://creativecommons.org/licenses/by/4.0/ | To develop trustworthy Vision-Language Models (VLMs), it is essential to
address adversarial robustness and hallucination mitigation, both of which
impact factual accuracy in high-stakes applications such as defense and
healthcare. Existing methods primarily focus on either adversarial defense or
hallucination post-hoc correction, leaving a gap in unified robustness
strategies. We introduce \textbf{Hydra}, an adaptive agentic framework that
enhances plug-in VLMs through iterative reasoning, structured critiques, and
cross-model verification, improving both resilience to adversarial
perturbations and intrinsic model errors. Hydra employs an Action-Critique
Loop, where it retrieves and critiques visual information, leveraging
Chain-of-Thought (CoT) and In-Context Learning (ICL) techniques to refine
outputs dynamically. Unlike static post-hoc correction methods, Hydra adapts to
both adversarial manipulations and intrinsic model errors, making it robust to
malicious perturbations and hallucination-related inaccuracies. We evaluate
Hydra on four VLMs, three hallucination benchmarks, two adversarial attack
strategies, and two adversarial defense methods, assessing performance on both
clean and adversarial inputs. Results show that Hydra surpasses plug-in VLMs
and state-of-the-art (SOTA) dehallucination methods, even without explicit
adversarial defenses, demonstrating enhanced robustness and factual
consistency. By bridging adversarial resistance and hallucination mitigation,
Hydra provides a scalable, training-free solution for improving the reliability
of VLMs in real-world applications.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:51:20 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Chung-En",
"",
"",
"Johnny"
],
[
"Yu",
"",
"",
"Neil"
],
[
"Hsuan-Chih",
"",
"",
"Neil"
],
[
"Chen",
"",
""
],
[
"Jalaian",
"Brian",
""
],
[
"Bastian",
"Nathaniel D.",
""
]
] |
2504.14396 | Minho Park | Minho Park, Taewoong Kang, Jooyeol Yun, Sungwon Hwang, Jaegul Choo | SphereDiff: Tuning-free Omnidirectional Panoramic Image and Video
Generation via Spherical Latent Representation | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | The increasing demand for AR/VR applications has highlighted the need for
high-quality 360-degree panoramic content. However, generating high-quality
360-degree panoramic images and videos remains a challenging task due to the
severe distortions introduced by equirectangular projection (ERP). Existing
approaches either fine-tune pretrained diffusion models on limited ERP datasets
or attempt tuning-free methods that still rely on ERP latent representations,
leading to discontinuities near the poles. In this paper, we introduce
SphereDiff, a novel approach for seamless 360-degree panoramic image and video
generation using state-of-the-art diffusion models without additional tuning.
We define a spherical latent representation that ensures uniform distribution
across all perspectives, mitigating the distortions inherent in ERP. We extend
MultiDiffusion to spherical latent space and propose a spherical latent
sampling method to enable direct use of pretrained diffusion models. Moreover,
we introduce distortion-aware weighted averaging to further improve the
generation quality in the projection process. Our method outperforms existing
approaches in generating 360-degree panoramic content while maintaining high
fidelity, making it a robust solution for immersive AR/VR applications. The
code is available here. https://github.com/pmh9960/SphereDiff
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 19:59:11 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Park",
"Minho",
""
],
[
"Kang",
"Taewoong",
""
],
[
"Yun",
"Jooyeol",
""
],
[
"Hwang",
"Sungwon",
""
],
[
"Choo",
"Jaegul",
""
]
] |
2504.14397 | Sarah J. Witherspoon | A. V. Shepler and S. Witherspoon | Deformation Theory and Hopf Actions on Koszul Algebras | 47 pages | null | null | null | math.RA math.RT | http://creativecommons.org/licenses/by/4.0/ | We consider graded deformations and PBW deformations of algebras defined over
noncommutative algebras. We explain how fibers of graded deformations
correspond to filtered algebras admitting a PBW property, with focus on smash
product algebras for Hopf algebras acting on quadratic algebras. In particular,
we describe all PBW deformations arising from Hopf actions on Koszul algebras,
giving Hopf-Koszul Hecke algebras. Alexander-Whitney and Eilenberg-Zilber maps
for twisted tensor products transfer homological information between
resolutions and convert conditions on Hochschild cocycles and Gerstenhaber
brackets into explicit PBW conditions.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 20:13:59 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Shepler",
"A. V.",
""
],
[
"Witherspoon",
"S.",
""
]
] |
2504.14398 | Teresa Head-Gordon | Selim Sami, Maximilian F.S.J. Menger, Yingze Wang, Teresa Head-Gordon | Improved Treatment of 1-4 interactions in Force Fields for Molecular
Dynamics Simulations | null | null | null | null | physics.chem-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Traditional force fields commonly use a combination of bonded torsional terms
and empirically scaled non-bonded interactions to capture 1-4 energies and
forces of atoms separated by three bonds in a molecule. While this approach can
yield accurate torsional energy barriers, it often leads to inaccurate forces
and erroneous geometries, and creates an interdependence between dihedral terms
and non-bonded interactions, complicating parameterization and reducing
transferability. In this paper, we demonstrate that 1-4 interactions can be
accurately modeled using only bonded coupling terms, eliminating the need for
arbitrarily scaled non-bonded interactions altogether. Furthermore by
leveraging the automated parameterization capabilities of the Q-Force toolkit,
we efficiently determine the necessary coupling terms without the need for
manual adjustment. Our approach is validated on a range of representative
molecular systems, encompassing both flexible and rigid structures, and shows a
significant improvement in force field accuracy, obtaining sub-kcal/mol mean
absolute error for every molecule tested. These results suggest that a
bonded-only model for 1-4 interactions can provide a more accurate alternative
to that used in traditional force field models for chemical, biomolecular, and
materials systems.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 20:24:36 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Sami",
"Selim",
""
],
[
"Menger",
"Maximilian F. S. J.",
""
],
[
"Wang",
"Yingze",
""
],
[
"Head-Gordon",
"Teresa",
""
]
] |
2504.14399 | Philip Wiese | Philip Wiese, Maurus Item, Luca Bertaccini, Yvan Tortorella, Angelo
Garofalo, Luca Benini | RedMulE-FT: A Reconfigurable Fault-Tolerant Matrix Multiplication Engine | Accepted to CF25-OSHW: Workshop on Open-Source Hardware (3rd
Edition), co-located with Computing Frontiers 2025 | null | 10.1145/3706594.3726981 | null | cs.AR | http://creativecommons.org/licenses/by/4.0/ | As safety-critical applications increasingly rely on data-parallel
floating-point computations, there is an increasing need for flexible and
configurable fault tolerance in parallel floating-point accelerators such as
tensor engines. While replication-based methods ensure reliability but incur
high area and power costs, error correction codes lack the flexibility to trade
off robustness against performance. This work presents RedMulE-FT, a
runtime-configurable fault-tolerant extension of the RedMulE matrix
multiplication accelerator, balancing fault tolerance, area overhead, and
performance impacts. The fault tolerance mode is configured in a shadowed
context register file before task execution. By combining replication with
error-detecting codes to protect the data path, RedMulE-FT achieves an 11x
uncorrected fault reduction with only 2.3% area overhead. Full protection
extends to control signals, resulting in no functional errors after 1M
injections during our extensive fault injection simulation campaign, with a
total area overhead of 25.2% while maintaining a 500 MHz frequency in a 12 nm
technology.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 20:25:33 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Wiese",
"Philip",
""
],
[
"Item",
"Maurus",
""
],
[
"Bertaccini",
"Luca",
""
],
[
"Tortorella",
"Yvan",
""
],
[
"Garofalo",
"Angelo",
""
],
[
"Benini",
"Luca",
""
]
] |
2504.14401 | Mouly Dewan | Mouly Dewan, Jiqun Liu, Aditya Gautam, Chirag Shah | LLM-Driven Usefulness Judgment for Web Search Evaluation | null | null | null | null | cs.IR | http://creativecommons.org/licenses/by/4.0/ | Evaluation is fundamental in optimizing search experiences and supporting
diverse user intents in Information Retrieval (IR). Traditional search
evaluation methods primarily rely on relevance labels, which assess how well
retrieved documents match a user's query. However, relevance alone fails to
capture a search system's effectiveness in helping users achieve their search
goals, making usefulness a critical evaluation criterion. In this paper, we
explore an alternative approach: LLM-generated usefulness labels, which
incorporate both implicit and explicit user behavior signals to evaluate
document usefulness. We propose Task-aware Rubric-based Usefulness Evaluation
(TRUE), a rubric-driven evaluation method that employs iterative sampling and
reasoning to model complex search behavior patterns. Our findings show that (i)
LLMs can generate moderate usefulness labels by leveraging comprehensive search
session history incorporating personalization and contextual understanding, and
(ii) fine-tuned LLMs improve usefulness judgments when provided with structured
search session contexts. Additionally, we examine whether LLMs can distinguish
between relevance and usefulness, particularly in cases where this divergence
impacts search success. We also conduct an ablation study to identify key
metrics for accurate usefulness label generation, optimizing for token
efficiency and cost-effectiveness in real-world applications. This study
advances LLM-based usefulness evaluation by refining key user metrics,
exploring LLM-generated label reliability, and ensuring feasibility for
large-scale search systems.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 20:38:09 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Dewan",
"Mouly",
""
],
[
"Liu",
"Jiqun",
""
],
[
"Gautam",
"Aditya",
""
],
[
"Shah",
"Chirag",
""
]
] |
2504.14402 | Albert Huber | Albert Huber | General Shells and Generalized Functions | 25 pages, no figures, matches published version, to be published in
CQG | null | null | null | gr-qc | http://creativecommons.org/licenses/by/4.0/ | In this work, standard methods of the mixed thin-shell foramlism are refined
using the framework of Colombeau's theory of generalized functions. To this
end, systematic use is made of smooth generalized functions, in particular
regularizations of the Heaviside step function and the delta distribution,
instead of working directly with the corresponding Schwartz distributions.
Based on this change of method, the resulting extended thin shell formalism is
shown to offer a decisive advantage over traditional approaches to the subject:
it avoids dealing with ill-defined products of distributions in the calculation
of nonlinear curvature expressions, thereby allowing for the treatment of
problems that prove intractable with the 'conventional' thin-shell formalism.
This includes, in particular, the problem of matching singular spacetimes with
distributional metrics (containing a delta distribution term) across a joint
boundary hypersurface in spacetime, the problem of setting up the dominant
energy condition for thin shells, and the problem of defining reasonably
rigorously nonlinear distribution-valued curvature invariants needed in
higher-derivative theories of gravity. Eventually, as a further application,
close links to Penrose's cut-and-paste method are established by proving that
results of said method can be re-derived using the generalized formalism
presented.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 20:42:11 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Huber",
"Albert",
""
]
] |
2504.14403 | Moritz Jirak | Moritz Jirak | Weak dependence and optimal quantitative self-normalized central limit
theorems | Preprint of accepted version. DOI 10.4171/JEMS/1573 | null | null | null | math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Consider a stationary, weakly dependent sequence of random variables. Given
only mild conditions, allowing for polynomial decay of the autocovariance
function, we show a Berry-Esseen bound of optimal order $n^{-1/2}$ for
studentized (self-normalized) partial sums, both for the Kolmogorov and
Wasserstein (and $L^p$) distance. The results show that, in general, (minimax)
optimal estimators of the long-run variance lead to suboptimal bounds in the
central limit theorem, that is, the rate $n^{-1/2}$ cannot be reached. This can
be salvaged by simple methods: In order to maintain the optimal speed of
convergence $n^{-1/2}$, simple over-smoothing within a certain range is
necessary and sufficient. The setup contains many prominent dynamical systems
and time series models, including random walks on the general linear group,
products of positive random matrices, functionals of Garch models of any order,
functionals of dynamical systems arising from SDEs, iterated random functions
and many more.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 20:48:29 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Jirak",
"Moritz",
""
]
] |
2504.14404 | Pavan Dharanipragada | Pavan Dharanipragada, B. Sathiapalan | Yang-Mills interaction from boundary vector model | 10 pages + appendices | null | null | null | hep-th | http://creativecommons.org/licenses/by/4.0/ | We construct a non-abelian spin 1 gauge theory with a cubic interaction in
AdS$_4$ from the Exact Renormalisation Group (ERG) flow of a CFT in 3
dimensions. The latter is the ${USp}(2N)$ singlet sector of the free field
theory of $2N$ massless complex scalars. The quadratic and cubic terms in the
bulk action are those of a gauge fixed version of a (local) gauge invariant
action. By construction this bulk action is the evolution operator for the ERG
equation of the boundary theory and thus is guaranteed to reproduce the correct
boundary correlators using the usual AdS/CFT prescription. This work expands on
the programme first set out in arXiv:1706.03371.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 20:51:29 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Dharanipragada",
"Pavan",
""
],
[
"Sathiapalan",
"B.",
""
]
] |
2504.14405 | Abhinav Prakash Gahlot | Abhinav Prakash Gahlot and Huseyin Tuna Erdinc and Felix J. Herrmann | Sensitivity-aware rock physics enhanced digital shadow for
underground-energy storage monitoring | null | null | null | null | physics.geo-ph physics.comp-ph | http://creativecommons.org/licenses/by/4.0/ | Underground energy storage, which includes storage of hydrogen, compressed
air, and CO2, requires careful monitoring to track potential leakage pathways,
a situation where time-lapse seismic imaging alone may be inadequate. A
recently developed Digital Shadow (DS) enhances forecasting using machine
learning and Bayesian inference, yet their accuracy depends on assumed rock
physics models, the mismatch of which can lead to unreliable predictions for
the reservoir's state (saturation/pressure). Augmenting DS training with
multiple rock physics models mitigates errors but averages over uncertainties,
obscuring their sources. To address this challenge, we introduce context-aware
sensitivity analysis inspired by amortized Bayesian inference, allowing the DS
to learn explicit dependencies between seismic data, the reservoir state, e.g.,
CO2 saturation, and rock physics models. At inference time, this approach
allows for real-time ''what if'' scenario testing rather than relying on costly
retraining, thereby enhancing interpretability and decision-making for safer,
more reliable underground storage.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 21:10:00 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Gahlot",
"Abhinav Prakash",
""
],
[
"Erdinc",
"Huseyin Tuna",
""
],
[
"Herrmann",
"Felix J.",
""
]
] |
2504.14406 | Harry Ye | Runlong Ye, Patrick Yung Kang Lee, Matthew Varona, Oliver Huang, and
Carolina Nobre | ScholarMate: A Mixed-Initiative Tool for Qualitative Knowledge Work and
Information Sensemaking | accepted at CHIWORK 2025 | null | null | null | cs.HC cs.AI | http://creativecommons.org/licenses/by/4.0/ | Synthesizing knowledge from large document collections is a critical yet
increasingly complex aspect of qualitative research and knowledge work. While
AI offers automation potential, effectively integrating it into human-centric
sensemaking workflows remains challenging. We present ScholarMate, an
interactive system designed to augment qualitative analysis by unifying AI
assistance with human oversight. ScholarMate enables researchers to dynamically
arrange and interact with text snippets on a non-linear canvas, leveraging AI
for theme suggestions, multi-level summarization, and contextual naming, while
ensuring transparency through traceability to source documents. Initial pilot
studies indicated that users value this mixed-initiative approach, finding the
balance between AI suggestions and direct manipulation crucial for maintaining
interpretability and trust. We further demonstrate the system's capability
through a case study analyzing 24 papers. By balancing automation with human
control, ScholarMate enhances efficiency and supports interpretability,
offering a valuable approach for productive human-AI collaboration in demanding
sensemaking tasks common in knowledge work.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 21:11:40 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Ye",
"Runlong",
""
],
[
"Lee",
"Patrick Yung Kang",
""
],
[
"Varona",
"Matthew",
""
],
[
"Huang",
"Oliver",
""
],
[
"Nobre",
"Carolina",
""
]
] |
2504.14407 | Chao Chen | Chao Chen and Sei Zhen Khong and Rodolphe Sepulchre | Soft and Hard Scaled Relative Graphs for Nonlinear Feedback Stability | null | null | null | null | eess.SY cs.SY math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents input-output stability analysis of nonlinear feedback
systems based on the notion of soft and hard scaled relative graphs (SRGs). The
soft and hard SRGs acknowledge the distinction between incremental positivity
and incremental passivity and reconcile them from a graphical perspective. The
essence of our proposed analysis is that the separation of soft/hard SRGs of
two open-loop systems on the complex plane guarantees closed-loop stability.
The main results generalize an existing soft SRG separation theorem for bounded
open-loop systems which was proved based on interconnection properties of soft
SRGs under a chordal assumption. By comparison, our analysis does not require
this chordal assumption and applies to possibly unbounded open-loop systems.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 21:28:40 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Chen",
"Chao",
""
],
[
"Khong",
"Sei Zhen",
""
],
[
"Sepulchre",
"Rodolphe",
""
]
] |
2504.14408 | Andrei Romashchenko | Andrei Romashchenko | Algebraic Barriers to Halving Algorithmic Information Quantities in
Correlated Strings | 20 pages, 3 figures | null | null | null | cs.IT cs.DM math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the possibility of scaling down algorithmic information quantities
in tuples of correlated strings. In particular, we address a question raised by
Alexander Shen: whether, for any triple of strings \((a, b, c)\), there exists
a string \(z\) such that each of the values of conditional Kolmogorov
complexity \(C(a|z), C(b|z), C(c|z)\) is approximately half of the
corresponding unconditional Kolmogorov complexity. We provide a negative answer
to this question by constructing a triple \((a, b, c)\) for which no such
string \(z\) exists. Our construction is based on combinatorial properties of
incidences in finite projective planes and relies on recent bounds on
point-line incidences over prime fields. As an application, we show that this
impossibility implies lower bounds on the communication complexity of secret
key agreement protocols in certain settings. These results reveal algebraic
obstructions to efficient information exchange and highlight a separation in
the information-theoretic behavior of projective planes over fields with and
without proper subfields.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 21:34:16 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Romashchenko",
"Andrei",
""
]
] |
2504.14409 | Christopher Ick | Christopher Ick, Gordon Wichern, Yoshiki Masuyama, Fran\c{c}ois G.
Germain, Jonathan Le Roux | Data Augmentation Using Neural Acoustic Fields With Retrieval-Augmented
Pre-training | Presented at ICASSP 2025 GenDA Workshop | null | null | null | eess.AS cs.AI cs.CV cs.LG cs.SD | http://creativecommons.org/licenses/by-nc-sa/4.0/ | This report details MERL's system for room impulse response (RIR) estimation
submitted to the Generative Data Augmentation Workshop at ICASSP 2025 for
Augmenting RIR Data (Task 1) and Improving Speaker Distance Estimation (Task
2). We first pre-train a neural acoustic field conditioned by room geometry on
an external large-scale dataset in which pairs of RIRs and the geometries are
provided. The neural acoustic field is then adapted to each target room by
using the enrollment data, where we leverage either the provided room
geometries or geometries retrieved from the external dataset, depending on
availability. Lastly, we predict the RIRs for each pair of source and receiver
locations specified by Task 1, and use these RIRs to train the speaker distance
estimation model in Task 2.
| [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 21:43:56 GMT"
}
] | 2025-04-22T00:00:00 | [
[
"Ick",
"Christopher",
""
],
[
"Wichern",
"Gordon",
""
],
[
"Masuyama",
"Yoshiki",
""
],
[
"Germain",
"François G.",
""
],
[
"Roux",
"Jonathan Le",
""
]
] |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.