text
null | inputs
dict | prediction
null | prediction_agent
null | annotation
list | annotation_agent
null | multi_label
bool 1
class | explanation
null | id
stringlengths 1
5
| metadata
null | status
stringclasses 2
values | event_timestamp
null | metrics
null |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
null |
{
"abstract": " Real-world machine learning applications often have complex test metrics, and\nmay have training and test data that follow different distributions. We propose\naddressing these issues by using a weighted loss function with a standard\nconvex loss, but with weights on the training examples that are learned to\noptimize the test metric of interest on the validation set. These\nmetric-optimized example weights can be learned for any test metric, including\nblack box losses and customized metrics for specific applications. We\nillustrate the performance of our proposal with public benchmark datasets and\nreal-world applications with domain shift and custom loss functions that\nbalance multiple objectives, impose fairness policies, and are non-convex and\nnon-decomposable.\n",
"title": "Metric-Optimized Example Weights"
}
| null | null | null | null | true | null |
4201
| null |
Default
| null | null |
null |
{
"abstract": " This paper proposes an efficient method for computing selected generalized\neigenpairs of a sparse Hermitian definite matrix pencil (A, B). Based on\nZolotarev's best rational function approximations of the signum function and\nconformal mapping techniques, we construct the best rational function\napproximation of a rectangular function supported on an arbitrary interval.\nThis new best rational function approximation is applied to construct spectrum\nfilters of (A, B). Combining fast direct solvers and the shift-invariant GMRES,\na hybrid fast algorithm is proposed to apply spectral filters efficiently.\nCompared to the state-of-the-art algorithm FEAST, the proposed rational\nfunction approximation is proved to be optimal among a larger function class,\nand the numerical implementation of the proposed method is also faster. The\nefficiency and stability of the proposed method are demonstrated by numerical\nexamples from computational chemistry.\n",
"title": "Interior Eigensolver for Sparse Hermitian Definite Matrices Based on Zolotarev's Functions"
}
| null | null |
[
"Computer Science",
"Mathematics"
] | null | true | null |
4202
| null |
Validated
| null | null |
null |
{
"abstract": " We establish a Polya-Vinogradov-type bound for finite periodic multipicative\ncharacters on the Gaussian integers.\n",
"title": "A Polya-Vinogradov-type inequality on $\\mathbb{Z}[i]$"
}
| null | null | null | null | true | null |
4203
| null |
Default
| null | null |
null |
{
"abstract": " Introductory and pedagogical treatmeant of the article : P. Broussous\n\"Distinction of the Steinberg representation\", with an appendix by François\nCourtès, IMRN 2014, no 11, 3140-3157. To appear in Proceedings of Chaire Jean\nMorlet, Dipendra Prasad, Volker Heiermann Ed. 2017. Contains modified and\nsimplified proofs of loc. cit. This article is written in memory of\nFrançois Courtès who passed away in september 2016.\n",
"title": "Distinction of representations via Bruhat-Tits buildings of p-adic groups"
}
| null | null | null | null | true | null |
4204
| null |
Default
| null | null |
null |
{
"abstract": " Generative adversarial networks (GANs) have received a tremendous amount of\nattention in the past few years, and have inspired applications addressing a\nwide range of problems. Despite its great potential, GANs are difficult to\ntrain. Recently, a series of papers (Arjovsky & Bottou, 2017a; Arjovsky et al.\n2017b; and Gulrajani et al. 2017) proposed using Wasserstein distance as the\ntraining objective and promised easy, stable GAN training across architectures\nwith minimal hyperparameter tuning. In this paper, we compare the performance\nof Wasserstein distance with other training objectives on a variety of GAN\narchitectures in the context of single image super-resolution. Our results\nagree that Wasserstein GAN with gradient penalty (WGAN-GP) provides stable and\nconverging GAN training and that Wasserstein distance is an effective metric to\ngauge training progress.\n",
"title": "Face Super-Resolution Through Wasserstein GANs"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
4205
| null |
Validated
| null | null |
null |
{
"abstract": " Experimental determination of protein function is resource-consuming. As an\nalternative, computational prediction of protein function has received\nattention. In this context, protein structural classification (PSC) can help,\nby allowing for determining structural classes of currently unclassified\nproteins based on their features, and then relying on the fact that proteins\nwith similar structures have similar functions. Existing PSC approaches rely on\nsequence-based or direct (\"raw\") 3-dimensional (3D) structure-based protein\nfeatures. In contrast, we first model 3D structures as protein structure\nnetworks (PSNs). Then, we use (\"processed\") network-based features for PSC. We\npropose the use of graphlets, state-of-the-art features in many domains of\nnetwork science, in the task of PSC. Moreover, because graphlets can deal only\nwith unweighted PSNs, and because accounting for edge weights when constructing\nPSNs could improve PSC accuracy, we also propose a deep learning framework that\nautomatically learns network features from the weighted PSNs. When evaluated on\na large set of ~9,400 CATH and ~12,800 SCOP protein domains (spanning 36 PSN\nsets), our proposed approaches are superior to existing PSC approaches in terms\nof accuracy, with comparable running time.\n",
"title": "Network-based protein structural classification"
}
| null | null | null | null | true | null |
4206
| null |
Default
| null | null |
null |
{
"abstract": " Eliminating the negative effect of non-stationary environmental noise is a\nlong-standing research topic for automatic speech recognition that stills\nremains an important challenge. Data-driven supervised approaches, including\nones based on deep neural networks, have recently emerged as potential\nalternatives to traditional unsupervised approaches and with sufficient\ntraining, can alleviate the shortcomings of the unsupervised methods in various\nreal-life acoustic environments. In this light, we review recently developed,\nrepresentative deep learning approaches for tackling non-stationary additive\nand convolutional degradation of speech with the aim of providing guidelines\nfor those involved in the development of environmentally robust speech\nrecognition systems. We separately discuss single- and multi-channel techniques\ndeveloped for the front-end and back-end of speech recognition systems, as well\nas joint front-end and back-end training frameworks.\n",
"title": "Deep Learning for Environmentally Robust Speech Recognition: An Overview of Recent Developments"
}
| null | null | null | null | true | null |
4207
| null |
Default
| null | null |
null |
{
"abstract": " We use a function field analogue of a method of Selberg to derive an\nasymptotic formula for the number of (square-free) monic polynomials in\n$\\mathbb{F}_q[X]$ of degree $n$ with precisely $k$ irreducible factors, in the\nlimit as $n$ tends to infinity. We then adapt this method to count such\npolynomials in arithmetic progressions and short intervals, and by making use\nof Weil's `Riemann hypothesis' for curves over $\\mathbb{F}_q$, obtain better\nranges for these formulae than are currently known for their analogues in the\nnumber field setting. Finally, we briefly discuss the regime in which $q$ tends\nto infinity.\n",
"title": "The function field Sathé-Selberg formula in arithmetic progressions and `short intervals'"
}
| null | null | null | null | true | null |
4208
| null |
Default
| null | null |
null |
{
"abstract": " We derive a correspondence between the eigenvalues of the adjacency matrix\n$A$ and the signless Laplacian matrix $Q$ of a graph $G$ when $G$ is\n$(d_1,d_2)$-biregular by using the relation $A^2=(Q-d_1I)(Q-d_2I)$. This\nmotivates asking when it is possible to have $X^r=f(Y)$ for $f$ a polynomial,\n$r>0$, and $X,\\ Y$ matrices associated to a graph $G$. It turns out that,\nessentially, this can only happen if $G$ is either regular or biregular.\n",
"title": "Polynomial Relations Between Matrices of Graphs"
}
| null | null | null | null | true | null |
4209
| null |
Default
| null | null |
null |
{
"abstract": " Deep learning has become a powerful and popular tool for a variety of machine\nlearning tasks. However, it is challenging to understand the mechanism of deep\nlearning from a theoretical perspective. In this work, we propose a random\nactive path model to study collective properties of deep neural networks with\nbinary synapses, under the removal perturbation of connections between layers.\nIn the model, the path from input to output is randomly activated, and the\ncorresponding input unit constrains the weights along the path into the form of\na $p$-weight interaction glass model. A critical value of the perturbation is\nobserved to separate a spin glass regime from a paramagnetic regime, with the\ntransition being of the first order. The paramagnetic phase is conjectured to\nhave a poor generalization performance.\n",
"title": "Random active path model of deep neural networks with diluted binary synapses"
}
| null | null | null | null | true | null |
4210
| null |
Default
| null | null |
null |
{
"abstract": " In nature or societies, the power-law is present ubiquitously, and then it is\nimportant to investigate the mathematical characteristics of power-laws in the\nrecent era of big data. In this paper we prove the superposition of\nnon-identical stochastic processes with power-laws converges in density to a\nunique stable distribution. This property can be used to explain the\nuniversality of stable laws such that the sums of the logarithmic return of\nnon-identical stock price fluctuations follow stable distributions.\n",
"title": "Super Generalized Central Limit Theorem: Limit distributions for sums of non-identical random variables with power-laws"
}
| null | null | null | null | true | null |
4211
| null |
Default
| null | null |
null |
{
"abstract": " We extend the results of Zhang et al. to show that $\\lambda$ is an eigenvalue\nof a $k$-uniform hypertree $(k \\geq 3)$ if and only if it is a root of a\nparticular matching polynomial for a connected induced subtree. We then use\nthis to provide a spectral characterization for power hypertrees. Notably, the\nsituation is quite different from that of ordinary trees, i.e., $2$-uniform\ntrees. We conclude by presenting an example (an $11$ vertex, $3$-uniform\nnon-power hypertree) illustrating these phenomena.\n",
"title": "On the Adjacency Spectra of Hypertrees"
}
| null | null | null | null | true | null |
4212
| null |
Default
| null | null |
null |
{
"abstract": " Plasma turbulence at scales of the order of the ion inertial length is\nmediated by several mechanisms, including linear wave damping, magnetic\nreconnection, formation and dissipation of thin current sheets, stochastic\nheating. It is now understood that the presence of localized coherent\nstructures enhances the dissipation channels and the kinetic features of the\nplasma. However, no formal way of quantifying the relationship between\nscale-to-scale energy transfer and the presence of spatial structures has so\nfar been presented. In this letter we quantify such relationship analyzing the\nresults of a two-dimensional high-resolution Hall-MHD simulation. In\nparticular, we employ the technique of space-filtering to derive a spectral\nenergy flux term which defines, in any point of the computational domain, the\nsigned flux of spectral energy across a given wavenumber. The characterization\nof coherent structures is performed by means of a traditional two-dimensional\nwavelet transformation. By studying the correlation between the spectral energy\nflux and the wavelet amplitude, we demonstrate the strong relationship between\nscale-to-scale transfer and coherent structures. Furthermore, by conditioning\none quantity with respect to the other, we are able for the first time to\nquantify the inhomogeneity of the turbulence cascade induced by topological\nstructures in the magnetic field. Taking into account the low filling-factor of\ncoherent structures (i.e. they cover a small portion of space), it emerges that\n80% of the spectral energy transfer (both in the direct and inverse cascade\ndirections) is localized in about 50% of space, and 50% of the energy transfer\nis localized in only 25% of space.\n",
"title": "Coherent structures and spectral energy transfer in turbulent plasma: a space-filter approach"
}
| null | null | null | null | true | null |
4213
| null |
Default
| null | null |
null |
{
"abstract": " Recurrent neural networks show state-of-the-art results in many text analysis\ntasks but often require a lot of memory to store their weights. Recently\nproposed Sparse Variational Dropout eliminates the majority of the weights in a\nfeed-forward neural network without significant loss of quality. We apply this\ntechnique to sparsify recurrent neural networks. To account for recurrent\nspecifics we also rely on Binary Variational Dropout for RNN. We report 99.5%\nsparsity level on sentiment analysis task without a quality drop and up to 87%\nsparsity level on language modeling task with slight loss of accuracy.\n",
"title": "Bayesian Sparsification of Recurrent Neural Networks"
}
| null | null | null | null | true | null |
4214
| null |
Default
| null | null |
null |
{
"abstract": " We build a deep reinforcement learning (RL) agent that can predict the\nlikelihood of an individual testing positive for malaria by asking questions\nabout their household. The RL agent learns to determine which survey question\nto ask next and when to stop to make a prediction about their likelihood of\nmalaria based on their responses hitherto. The agent incurs a small penalty for\neach question asked, and a large reward/penalty for making the correct/wrong\nprediction; it thus has to learn to balance the length of the survey with the\naccuracy of its final predictions. Our RL agent is a Deep Q-network that learns\na policy directly from the responses to the questions, with an action defined\nfor each possible survey question and for each possible prediction class. We\nfocus on Kenya, where malaria is a massive health burden, and train the RL\nagent on a dataset of 6481 households from the Kenya Malaria Indicator Survey\n2015. To investigate the importance of having survey questions be adaptive to\nresponses, we compare our RL agent to a supervised learning (SL) baseline that\nfixes its set of survey questions a priori. We evaluate on prediction accuracy\nand on the number of survey questions asked on a holdout set and find that the\nRL agent is able to predict with 80% accuracy, using only 2.5 questions on\naverage. In addition, the RL agent learns to survey adaptively to responses and\nis able to match the SL baseline in prediction accuracy while significantly\nreducing survey length.\n",
"title": "Malaria Likelihood Prediction By Effectively Surveying Households Using Deep Reinforcement Learning"
}
| null | null | null | null | true | null |
4215
| null |
Default
| null | null |
null |
{
"abstract": " Curiosity is the strong desire to learn or know more about something or\nsomeone. Since learning is often a social endeavor, social dynamics in\ncollaborative learning may inevitably influence curiosity. There is a scarcity\nof research, however, focusing on how curiosity can be evoked in group learning\ncontexts. Inspired by a recently proposed theoretical framework that\narticulates an integrated socio-cognitive infrastructure of curiosity, in this\nwork, we use data-driven approaches to identify fine-grained social scaffolding\nof curiosity in child-child interaction, and propose how they can be used to\nelicit and maintain curiosity in technology-enhanced learning environments. For\nexample, we discovered sequential patterns of multimodal behaviors across group\nmembers and we describe those that maximize an individual's utility, or\nlikelihood, of demonstrating curiosity during open-ended problem-solving in\ngroup work. We also discovered, and describe here, behaviors that directly or\nin a mediated manner cause curiosity related conversational behaviors in the\ninteraction, with twice as many interpersonal causal influences compared to\nintrapersonal ones. We explain how these findings form a solid foundation for\ndeveloping curiosity-increasing learning technologies or even assisting a human\ncoach to induce curiosity among learners.\n",
"title": "Curious Minds Wonder Alike: Studying Multimodal Behavioral Dynamics to Design Social Scaffolding of Curiosity"
}
| null | null | null | null | true | null |
4216
| null |
Default
| null | null |
null |
{
"abstract": " We initiate the study of the communication complexity of fair division with\nindivisible goods. We focus on some of the most well-studied fairness notions\n(envy-freeness, proportionality, and approximations thereof) and valuation\nclasses (submodular, subadditive and unrestricted). Within these parameters,\nour results completely resolve whether the communication complexity of\ncomputing a fair allocation (or determining that none exist) is polynomial or\nexponential (in the number of goods), for every combination of fairness notion,\nvaluation class, and number of players, for both deterministic and randomized\nprotocols.\n",
"title": "Communication Complexity of Discrete Fair Division"
}
| null | null | null | null | true | null |
4217
| null |
Default
| null | null |
null |
{
"abstract": " We prove (and improve) the Muir-Suffridge conjecture for holomorphic convex\nmaps. Namely, let $F:\\mathbb B^n\\to \\mathbb C^n$ be a univalent map from the\nunit ball whose image $D$ is convex. Let $\\mathcal S\\subset \\partial \\mathbb\nB^n$ be the set of points $\\xi$ such that $\\lim_{z\\to \\xi}\\|F(z)\\|=\\infty$.\nThen we prove that $\\mathcal S$ is either empty, or contains one or two points\nand $F$ extends as a homeomorphism $\\tilde{F}:\\overline{\\mathbb B^n}\\setminus\n\\mathcal S\\to \\overline{D}$. Moreover, $\\mathcal S=\\emptyset$ if $D$ is\nbounded, $\\mathcal S$ has one point if $D$ has one connected component at\n$\\infty$ and $\\mathcal S$ has two points if $D$ has two connected components at\n$\\infty$ and, up to composition with an affine map, $F$ is an extension of the\nstrip map in the plane to higher dimension.\n",
"title": "A proof of the Muir-Suffridge conjecture for convex maps of the unit ball in $\\mathbb C^n$"
}
| null | null | null | null | true | null |
4218
| null |
Default
| null | null |
null |
{
"abstract": " In this article we give an approach to define continuous functional calculus\nfor bounded quaternionic normal operators defined on a right quaternionic\nHilbert space.\n",
"title": "Continuous Functional Calculus for Quaternionic Bounded Normal Operators"
}
| null | null |
[
"Mathematics"
] | null | true | null |
4219
| null |
Validated
| null | null |
null |
{
"abstract": " The interplay between spin-orbit coupling (SOC) and electron correlation\n($U$) is considered for many exotic phenomena in iridium oxides. We have\ninvestigated the evolution of structural, magnetic and electronic properties in\npyrochlore iridate Y$_2$Ir$_{2-x}$Ru$_{x}$O$_7$ where the substitution of Ru\nhas been aimed to tune this interplay. The Ru substitution does not introduce\nany structural phase transition, however, we do observe an evolution of lattice\nparameters with the doping level $x$. X-ray photoemission spectroscopy (XPS)\nstudy indicates Ru adopts charge state of Ru$^{4+}$ and replaces the Ir$^{4+}$\naccordingly. Magnetization data reveal both the onset of magnetic\nirreversibility and the magnetic moment decreases with progressive substitution\nof Ru. These materials show non-equilibrium low temperature magnetic state as\nrevealed by magnetic relaxation data. Interestingly, we find magnetic\nrelaxation rate increases with substitution of Ru. The electrical resistivity\nshows an insulating behavior in whole temperature range, however, resistivity\ndecreases with substitution of Ru. Nature of electronic conduction has been\nfound to follow power-law behavior for all the materials.\n",
"title": "Evolution of structure, magnetism and electronic transport in doped pyrochlore iridate Y$_2$Ir$_{2-x}$Ru$_{x}$O$_7$"
}
| null | null | null | null | true | null |
4220
| null |
Default
| null | null |
null |
{
"abstract": " We present a strong version of Abouzaid's No-Escape Lemma, which allows\nvarying contact forms on the boundary and which can be used instead of the\nMaximum Principle. Moreover, we give a clarified proof of Cieliebak's\nInvariance Theorem for Symplectic homology under subcritical handle attachment.\nFinally, we introduce the notion of asymptotically finitely generated contact\nstructures, which states essentially that the Symplectic homology in a certain\ndegree of any filling of such contact manifolds is uniformly generated by only\nfinitely many Reeb orbits. This property is then used to show that a large\nclass of manifolds carries infinitely many exactly fillable contact structures.\n",
"title": "Cieliebak's Invariance Theorem and contact structures via connected sums"
}
| null | null | null | null | true | null |
4221
| null |
Default
| null | null |
null |
{
"abstract": " Graph Weighted Models (GWMs) have recently been proposed as a natural\ngeneralization of weighted automata over strings and trees to arbitrary\nfamilies of labeled graphs (and hypergraphs). A GWM generically associates a\nlabeled graph with a tensor network and computes a value by successive\ncontractions directed by its edges. In this paper, we consider the problem of\nlearning GWMs defined over the graph family of pictures (or 2-dimensional\nwords). As a proof of concept, we consider regression and classification tasks\nover the simple Bars & Stripes and Shifting Bits picture languages and provide\nan experimental study investigating whether these languages can be learned in\nthe form of a GWM from positive and negative examples using gradient-based\nmethods. Our results suggest that this is indeed possible and that\ninvestigating the use of gradient-based methods to learn picture series and\nfunctions computed by GWMs over other families of graphs could be a fruitful\ndirection.\n",
"title": "Learning Graph Weighted Models on Pictures"
}
| null | null | null | null | true | null |
4222
| null |
Default
| null | null |
null |
{
"abstract": " In this paper, we are concerned with the existence of least energy solutions\nfor the following biharmonic equations: $$\\Delta^2 u+(\\lambda\nV(x)-\\delta)u=|u|^{p-2}u \\quad in\\quad \\mathbb{R}^N$$ where $N\\geq 5,\n2<p\\leq\\frac{2N}{N-4}, \\lambda>0$ is a parameter, $V(x)$ is a nonnegative\npotential function with nonempty zero sets $\\mbox{int} V^{-1}(0)$,\n$0<\\delta<\\mu_0$ and $\\mu_0$ is the principle eigenvalue of $\\Delta^2$ in the\nzero sets $\\mbox{int} V^{-1}(0)$ of $V(x)$. Here $\\mbox{int} V^{-1}(0)$ denotes\nthe interior part of the set $V^{-1}(0):=\\{x\\in \\mathbb{R}^N: V(x)=0\\}$. We\nprove that the above equation admits a least energy solution which is trapped\nnear the zero sets $\\mbox{int} V^{-1}(0)$ for $\\lambda>0$ large.\n",
"title": "Solutions for biharmonic equations with steep potential wells"
}
| null | null | null | null | true | null |
4223
| null |
Default
| null | null |
null |
{
"abstract": " N distinguishable players are randomly fitted with a white or black hat,\nwhere the probabilities of getting a white or black hat may be different for\neach player, but known to all the players. All players guess simultaneously the\ncolor of their own hat observing only the hat colors of the other N-1 players.\nIt is also allowed for each player to pass: no color is guessed. The team wins\nif at least one player guesses his hat color correctly and none of the players\nhas an incorrect guess. No communication of any sort is allowed, except for an\ninitial strategy session before the game begins. Our goal is to maximize the\nprobability of winning the game and to describe winning strategies, using the\nconcept of an adequate set. We find explicit solutions in case of N =3 and N\n=4.\n",
"title": "General three and four person two color Hat Game"
}
| null | null | null | null | true | null |
4224
| null |
Default
| null | null |
null |
{
"abstract": " Over any field $\\mathbb K$, there is a bijection between regular spreads of\nthe projective space ${\\rm PG}(3,{\\mathbb K})$ and $0$-secant lines of the\nKlein quadric in ${\\rm PG}(5,{\\mathbb K})$. Under this bijection, regular\nparallelisms of ${\\rm PG}(3,{\\mathbb K})$ correspond to hyperflock determining\nline sets (hfd line sets) with respect to the Klein quadric. An hfd line set is\ndefined to be \\emph{pencilled} if it is composed of pencils of lines. We\npresent a construction of pencilled hfd line sets, which is then shown to\ndetermine all such sets. Based on these results, we describe the corresponding\nregular parallelisms. These are also termed as being \\emph{pencilled}. Any\nClifford parallelism is regular and pencilled. From this, we derive necessary\nand sufficient algebraic conditions for the existence of pencilled hfd line\nsets.\n",
"title": "Pencilled regular parallelisms"
}
| null | null | null | null | true | null |
4225
| null |
Default
| null | null |
null |
{
"abstract": " While the optimization problem behind deep neural networks is highly\nnon-convex, it is frequently observed in practice that training deep networks\nseems possible without getting stuck in suboptimal points. It has been argued\nthat this is the case as all local minima are close to being globally optimal.\nWe show that this is (almost) true, in fact almost all local minima are\nglobally optimal, for a fully connected network with squared loss and analytic\nactivation function given that the number of hidden units of one layer of the\nnetwork is larger than the number of training points and the network structure\nfrom this layer on is pyramidal.\n",
"title": "The loss surface of deep and wide neural networks"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
4226
| null |
Validated
| null | null |
null |
{
"abstract": " This paper presents an analysis of rearward gap acceptance characteristics of\ndrivers of large trucks in highway lane change scenarios. The range between the\nvehicles was inferred from camera images using the estimated lane width\nobtained from the lane tracking camera as the reference. Six-hundred lane\nchange events were acquired from a large-scale naturalistic driving data set.\nThe kinematic variables from the image-based gap analysis were filtered by the\nweighted linear least squares in order to extrapolate them at the lane change\ntime. In addition, the time-to-collision and required deceleration were\ncomputed, and potential safety threshold values are provided. The resulting\nrange and range rate distributions showed directional discrepancies, i.e., in\nleft lane changes, large trucks are often slower than other vehicles in the\ntarget lane, whereas they are usually faster in right lane changes. Video\nobservations have confirmed that major motivations for changing lanes are\ndifferent depending on the direction of move, i.e., moving to the left (faster)\nlane occurs due to a slower vehicle ahead or a merging vehicle on the\nright-hand side, whereas right lane changes are frequently made to return to\nthe original lane after passing.\n",
"title": "Gap Acceptance During Lane Changes by Large-Truck Drivers-An Image-Based Analysis"
}
| null | null | null | null | true | null |
4227
| null |
Default
| null | null |
null |
{
"abstract": " We propose a simple yet highly effective method that addresses the\nmode-collapse problem in the Conditional Generative Adversarial Network (cGAN).\nAlthough conditional distributions are multi-modal (i.e., having many modes) in\npractice, most cGAN approaches tend to learn an overly simplified distribution\nwhere an input is always mapped to a single output regardless of variations in\nlatent code. To address such issue, we propose to explicitly regularize the\ngenerator to produce diverse outputs depending on latent codes. The proposed\nregularization is simple, general, and can be easily integrated into most\nconditional GAN objectives. Additionally, explicit regularization on generator\nallows our method to control a balance between visual quality and diversity. We\ndemonstrate the effectiveness of our method on three conditional generation\ntasks: image-to-image translation, image inpainting, and future video\nprediction. We show that simple addition of our regularization to existing\nmodels leads to surprisingly diverse generations, substantially outperforming\nthe previous approaches for multi-modal conditional generation specifically\ndesigned in each individual task.\n",
"title": "Diversity-Sensitive Conditional Generative Adversarial Networks"
}
| null | null | null | null | true | null |
4228
| null |
Default
| null | null |
null |
{
"abstract": " Global Style Tokens (GSTs) are a recently-proposed method to learn latent\ndisentangled representations of high-dimensional data. GSTs can be used within\nTacotron, a state-of-the-art end-to-end text-to-speech synthesis system, to\nuncover expressive factors of variation in speaking style. In this work, we\nintroduce the Text-Predicted Global Style Token (TP-GST) architecture, which\ntreats GST combination weights or style embeddings as \"virtual\" speaking style\nlabels within Tacotron. TP-GST learns to predict stylistic renderings from text\nalone, requiring neither explicit labels during training nor auxiliary inputs\nfor inference. We show that, when trained on a dataset of expressive speech,\nour system generates audio with more pitch and energy variation than two\nstate-of-the-art baseline models. We further demonstrate that TP-GSTs can\nsynthesize speech with background noise removed, and corroborate these analyses\nwith positive results on human-rated listener preference audiobook tasks.\nFinally, we demonstrate that multi-speaker TP-GST models successfully factorize\nspeaker identity and speaking style. We provide a website with audio samples\nfor each of our findings.\n",
"title": "Predicting Expressive Speaking Style From Text In End-To-End Speech Synthesis"
}
| null | null | null | null | true | null |
4229
| null |
Default
| null | null |
null |
{
"abstract": " Online social networks are more and more studied. The links between users of\na social network are important and have to be well qualified in order to detect\ncommunities and find influencers for example. In this paper, we present an\napproach based on the theory of belief functions to estimate the degrees of\ncognitive independence between users in a social network. We experiment the\nproposed method on a large amount of data gathered from the Twitter social\nnetwork.\n",
"title": "Independence of Sources in Social Networks"
}
| null | null |
[
"Computer Science"
] | null | true | null |
4230
| null |
Validated
| null | null |
null |
{
"abstract": " We review possible mechanisms for energy transfer based on 'rare' or\n'non-perturbative' effects, in physical systems that present a many-body\nlocalized phenomenology. The main focus is on classical systems, with or\nwithout quenched disorder. For non-quantum systems, the breakdown of\nlocalization is usually not regarded as an issue, and we thus aim at\nidentifying the fastest channels for transport. Next, we contemplate the\npossibility of applying the same mechanisms in quantum systems, including\ndisorder free systems (e.g. Bose-Hubbard chain), disordered many-body localized\nsystems with mobility edges at energies below the edge, and strongly disordered\nlattice systems in $d>1$. For quantum mechanical systems, the relevance of\nthese considerations for transport is currently a matter of debate.\n",
"title": "Classical and quantum systems: transport due to rare events"
}
| null | null | null | null | true | null |
4231
| null |
Default
| null | null |
null |
{
"abstract": " We consider a priori generalization bounds developed in terms of\ncross-validation estimates and the stability of learners. In particular, we\nfirst derive an exponential Efron-Stein type tail inequality for the\nconcentration of a general function of n independent random variables. Next,\nunder some reasonable notion of stability, we use this exponential tail bound\nto analyze the concentration of the k-fold cross-validation (KFCV) estimate\naround the true risk of a hypothesis generated by a general learning rule.\nWhile the accumulated literature has often attributed this concentration to the\nbias and variance of the estimator, our bound attributes this concentration to\nthe stability of the learning rule and the number of folds k. This insight\nraises valid concerns related to the practical use of KFCV and suggests\nresearch directions to obtain reliable empirical estimates of the actual risk.\n",
"title": "An a Priori Exponential Tail Bound for k-Folds Cross-Validation"
}
| null | null | null | null | true | null |
4232
| null |
Default
| null | null |
null |
{
"abstract": " We propose a novel class of statistical divergences called \\textit{Relaxed\nWasserstein} (RW) divergence. RW divergence generalizes Wasserstein divergence\nand is parametrized by a class of strictly convex and differentiable functions.\nWe establish for RW divergence several probabilistic properties, which are\ncritical for the success of Wasserstein divergence. In particular, we show that\nRW divergence is dominated by Total Variation (TV) and Wasserstein-$L^2$\ndivergence, and that RW divergence has continuity, differentiability and\nduality representation. Finally, we provide a nonasymptotic moment estimate and\na concentration inequality for RW divergence.\nOur experiments on the image generation task demonstrate that RW divergence\nis a suitable choice for GANs. Indeed, the performance of RWGANs with\nKullback-Leibler (KL) divergence is very competitive with other\nstate-of-the-art GANs approaches. Furthermore, RWGANs possess better\nconvergence properties than the existing WGANs with competitive inception\nscores. To the best of our knowledge, our new conceptual framework is the first\nto not only provide the flexibility in designing effective GANs scheme, but\nalso the possibility in studying different losses under a unified mathematical\nframework.\n",
"title": "Relaxed Wasserstein with Applications to GANs"
}
| null | null | null | null | true | null |
4233
| null |
Default
| null | null |
null |
{
"abstract": " We study the existence and nonexistence of maximizers for variational problem\nconcerning to the Moser--Trudinger inequality of Adimurthi--Druet type in\n$W^{1,N}(\\mathbb R^N)$ \\[ MT(N,\\beta, \\alpha) =\\sup_{u\\in W^{1,N}(\\mathbb R^N),\n\\|\\nabla u\\|_N^N + \\|u\\|_N^N\\leq 1} \\int_{\\mathbb R^N} \\Phi_N(\\beta(1+\\alpha\n\\|u\\|_N^N)^{\\frac1{N-1}} |u|^{\\frac N{N-1}}) dx, \\] where $\\Phi_N(t) =e^{t}\n-\\sum_{k=0}^{N-2} \\frac{t^k}{k!}$, $0\\leq \\alpha < 1$ both in the subcritical\ncase $\\beta < \\beta_N$ and critical case $\\beta =\\beta_N$ with $\\beta_N = N\n\\omega_{N-1}^{\\frac1{N-1}}$ and $\\omega_{N-1}$ denotes the surface area of the\nunit sphere in $\\mathbb R^N$. We will show that $MT(N,\\beta,\\alpha)$ is\nattained in the subcritical case if $N\\geq 3$ or $N=2$ and $\\beta \\in\n(\\frac{2(1+2\\alpha)}{(1+\\alpha)^2 B_2},\\beta_2)$ with $B_2$ is the best\nconstant in a Gagliardo--Nirenberg inequality in $W^{1,2}(\\mathbb R^2)$. We\nalso show that $MT(2,\\beta,\\alpha)$ is not attained for $\\beta$ small which is\ndifferent from the context of bounded domains. In the critical case, we prove\nthat $MT(N,\\beta_N,\\alpha)$ is attained for $\\alpha\\geq 0$ small enough. To\nprove our results, we first establish a lower bound for $MT(N,\\beta,\\alpha)$\nwhich excludes the concentrating or vanishing behaviors of their maximizer\nsequences. This implies the attainability of $MT(N,\\beta,\\alpha)$ in the\nsubcritical case. The proof in the critical case is based on the blow-up\nanalysis method. Finally, by using the Moser sequence together the scaling\nargument, we show that $MT(N,\\beta_N,1) =\\infty$. Our results settle the\nquestions left open in \\cite{doO2015,doO2016}.\n",
"title": "Extremal functions for the Moser--Trudinger inequality of Adimurthi--Druet type in $W^{1,N}(\\mathbb R^N)$"
}
| null | null | null | null | true | null |
4234
| null |
Default
| null | null |
null |
{
"abstract": " This paper introduces Dex, a reinforcement learning environment toolkit\nspecialized for training and evaluation of continual learning methods as well\nas general reinforcement learning problems. We also present the novel continual\nlearning method of incremental learning, where a challenging environment is\nsolved using optimal weight initialization learned from first solving a similar\neasier environment. We show that incremental learning can produce vastly\nsuperior results than standard methods by providing a strong baseline method\nacross ten Dex environments. We finally develop a saliency method for\nqualitative analysis of reinforcement learning, which shows the impact\nincremental learning has on network attention.\n",
"title": "Dex: Incremental Learning for Complex Environments in Deep Reinforcement Learning"
}
| null | null | null | null | true | null |
4235
| null |
Default
| null | null |
null |
{
"abstract": " In this paper we discuss the first order partial differential equations\nresolved with any derivatives. At first, we transform the first order partial\ndifferential equation resolved with respect to a time derivative into a system\nof linear equations. Secondly, we convert it into a system of the first order\nlinear partial differential equations with constant coefficients and nonlinear\nalgebraic equations. Thirdly, we solve them by the Fourier transform and\nconvert them into the equivalent integral equations. At last, we extend to\ndiscuss the first order partial differential equations resolved with respect to\ntime derivatives and the general scenario resolved with any derivatives.\n",
"title": "The first order partial differential equations resolved with any derivatives"
}
| null | null | null | null | true | null |
4236
| null |
Default
| null | null |
null |
{
"abstract": " We prove the genus-one restriction of the all-genus\nLandau-Ginzburg/Calabi-Yau conjecture of Chiodo and Ruan, stated in terms of\nthe geometric quantization of an explicit symplectomorphism determined by\ngenus-zero invariants. This provides the first evidence supporting the\nhigher-genus Landau-Ginzburg/Calabi-Yau correspondence for the quintic\nthreefold, and exhibits the first instance of the \"genus zero controls higher\ngenus\" principle, in the sense of Givental's quantization formalism, for\nnon-semisimple cohomological field theories.\n",
"title": "The Genus-One Global Mirror Theorem for the Quintic Threefold"
}
| null | null | null | null | true | null |
4237
| null |
Default
| null | null |
null |
{
"abstract": " We present a novel approach for mobile manipulator self-calibration using\ncontact information. Our method, based on point cloud registration, is applied\nto estimate the extrinsic transform between a fixed vision sensor mounted on a\nmobile base and an end effector. Beyond sensor calibration, we demonstrate that\nthe method can be extended to include manipulator kinematic model parameters,\nwhich involves a non-rigid registration process. Our procedure uses on-board\nsensing exclusively and does not rely on any external measurement devices,\nfiducial markers, or calibration rigs. Further, it is fully automatic in the\ngeneral case. We experimentally validate the proposed method on a custom mobile\nmanipulator platform, and demonstrate centimetre-level post-calibration\naccuracy in positioning of the end effector using visual guidance only. We also\ndiscuss the stability properties of the registration algorithm, in order to\ndetermine the conditions under which calibration is possible.\n",
"title": "Self-Calibration of Mobile Manipulator Kinematic and Sensor Extrinsic Parameters Through Contact-Based Interaction"
}
| null | null | null | null | true | null |
4238
| null |
Default
| null | null |
null |
{
"abstract": " The paper presents first results of the CitEcCyr project funded by RANEPA.\nThe project aims to create a source of open citation data for research papers\nwritten in Russian. Compared to existing sources of citation data, CitEcCyr is\nworking to provide the following added values: a) a transparent and distributed\narchitecture of a technology that generates the citation data; b) an openness\nof all built/used software and created citation data; c) an extended set of\ncitation data sufficient for the citation content analysis; d) services for\npublic control over a quality of the citation data and a citing activity of\nresearchers.\n",
"title": "Towards Open Data for the Citation Content Analysis"
}
| null | null | null | null | true | null |
4239
| null |
Default
| null | null |
null |
{
"abstract": " The majority of everyday tasks involve interacting with unstructured\nenvironments. This implies that, in order for robots to be truly useful they\nmust be able to handle contacts. This paper explores how a particle filter can\nbe used to localize a contact point and estimate the external force. We\ndemonstrate the capability of the particle filter on a simulated 4DoF planar\nrobotic arm, and compare it to a well-established analytical approach.\n",
"title": "Real Time Collision Detection and Identification for Robotic Manipulators"
}
| null | null | null | null | true | null |
4240
| null |
Default
| null | null |
null |
{
"abstract": " In this article, weak convergence of the general non-Markov state transition\nprobability estimator by Titman (2015) is established which, up to now, has not\nbeen verified yet for other general non-Markov estimators. A similar theorem is\nshown for the bootstrap, yielding resampling-based inference methods for\nstatistical functionals. Formulas of the involved covariance functions are\npresented in detail. Particular applications include the conditional expected\nlength of stay in a specific state, given occupation of another state in the\npast, as well as the construction of time-simultaneous confidence bands for the\ntransition probabilities. The expected lengths of stay in the two-sample liver\ncirrhosis data-set by Andersen et al. (1993) are compared and confidence\nintervals for their difference are constructed. With borderline significance\nand in comparison to the placebo group, the treatment group has an elevated\nexpected length of stay in the healthy state given an earlier disease state\noccupation. In contrast, the Aalen-Johansen estimator-based confidence\ninterval, which relies on a Markov assumption, leads to a drastically different\nconclusion. Also, graphical illustrations of confidence bands for the\ntransition probabilities demonstrate the biasedness of the Aalen-Johansen\nestimator in this data example. The reliability of these results is assessed in\na simulation study.\n",
"title": "Time-dynamic inference for non-Markov transition probabilities under independent right-censoring"
}
| null | null | null | null | true | null |
4241
| null |
Default
| null | null |
null |
{
"abstract": " Segmental conditional random fields (SCRFs) and connectionist temporal\nclassification (CTC) are two sequence labeling methods used for end-to-end\ntraining of speech recognition models. Both models define a transcription\nprobability by marginalizing decisions about latent segmentation alternatives\nto derive a sequence probability: the former uses a globally normalized joint\nmodel of segment labels and durations, and the latter classifies each frame as\neither an output symbol or a \"continuation\" of the previous label. In this\npaper, we train a recognition model by optimizing an interpolation between the\nSCRF and CTC losses, where the same recurrent neural network (RNN) encoder is\nused for feature extraction for both outputs. We find that this multitask\nobjective improves recognition accuracy when decoding with either the SCRF or\nCTC models. Additionally, we show that CTC can also be used to pretrain the RNN\nencoder, which improves the convergence rate when learning the joint model.\n",
"title": "Multitask Learning with CTC and Segmental CRF for Speech Recognition"
}
| null | null | null | null | true | null |
4242
| null |
Default
| null | null |
null |
{
"abstract": " The use of eco-friendly materials for the environment has been addressed as a\ncritical issue in the development of systems for renewable energy applications.\nIn this regard, the investigation of organic photovoltaic (OPV) molecules for\nthe implementation in solar cells, has become a subject of intense research in\nthe last years. The present work is a systematic study at the B3LYP level of\ntheory performed for a series of 50 OPV materials. Full geometry optimizations\nrevealed that those systems with a twisted geometry are the most energetically\nstable. Nuclear independent Chemical shifts (NICS) values show a strong\naromatic character along the series, indicating a possible polymerization in\nsolid-state, via a {\\pi}-{\\pi} stacking, which may be relevant in the design of\na solar cell device. The absorption spectra in the series was also computed\nusing Time Dependent DFT at the same level of theory, indicating that all\nspectra are red-shifted along the series. This is a promissory property that\nmay be directly implemented in a photovoltaic material, since it is possible to\nabsorb a larger range of visible light. The computed HOMO-LUMO gaps as a\nmeasurement of the band gap in semiconductors, show a reasonable agreement with\nthose found in experiment, predicting candidate materials that may be directly\nused in photovoltaic applications. Non-linear optical (NLO) properties were\nalso estimated with the aid of a PCBM molecule as a model of an acceptor, and a\nfinal set of optimal systems was identified as potential candidates to be\nimplemented as photovoltaic materials. The methodological approach presented in\nthis work may aid in the in silico assisted-design of OPV materials.\n",
"title": "Electronic structure and non-linear optical properties of organic photovoltaic systems with potential applications on solar cell devices: A DFT approach"
}
| null | null | null | null | true | null |
4243
| null |
Default
| null | null |
null |
{
"abstract": " While much of the work in the design of convolutional networks over the last\nfive years has revolved around the empirical investigation of the importance of\ndepth, filter sizes, and number of feature channels, recent studies have shown\nthat branching, i.e., splitting the computation along parallel but distinct\nthreads and then aggregating their outputs, represents a new promising\ndimension for significant improvements in performance. To combat the complexity\nof design choices in multi-branch architectures, prior work has adopted simple\nstrategies, such as a fixed branching factor, the same input being fed to all\nparallel branches, and an additive combination of the outputs produced by all\nbranches at aggregation points.\nIn this work we remove these predefined choices and propose an algorithm to\nlearn the connections between branches in the network. Instead of being chosen\na priori by the human designer, the multi-branch connectivity is learned\nsimultaneously with the weights of the network by optimizing a single loss\nfunction defined with respect to the end task. We demonstrate our approach on\nthe problem of multi-class image classification using three different datasets\nwhere it yields consistently higher accuracy compared to the state-of-the-art\n\"ResNeXt\" multi-branch network given the same learning capacity.\n",
"title": "Connectivity Learning in Multi-Branch Networks"
}
| null | null | null | null | true | null |
4244
| null |
Default
| null | null |
null |
{
"abstract": " A rigorous bridge between spiking-level and macroscopic quantities is an\non-going and well-developed story for asynchronously firing neurons, but focus\nhas shifted to include neural populations exhibiting varying synchronous\ndynamics. Recent literature has used the Ott--Antonsen ansatz (2008) to great\neffect, allowing a rigorous derivation of an order parameter for large\noscillator populations. The ansatz has been successfully applied using several\nmodels including networks of Kuramoto oscillators, theta models, and\nintegrate-and-fire neurons, along with many types of network topologies. In the\npresent study, we take a converse approach: given the mean field dynamics of\nslow synapses, predict the synchronization properties of finite neural\npopulations. The slow synapse assumption is amenable to averaging theory and\nthe method of multiple timescales. Our proposed theory applies to two\nheterogeneous populations of N excitatory n-dimensional and N inhibitory\nm-dimensional oscillators with homogeneous synaptic weights. We then\ndemonstrate our theory using two examples. In the first example we take a\nnetwork of excitatory and inhibitory theta neurons and consider the case with\nand without heterogeneous inputs. In the second example we use Traub models\nwith calcium for the excitatory neurons and Wang-Buzs{á}ki models for the\ninhibitory neurons. We accurately predict phase drift and phase locking in each\nexample even when the slow synapses exhibit non-trivial mean-field dynamics.\n",
"title": "A multiple timescales approach to bridging spiking- and population-level dynamics"
}
| null | null | null | null | true | null |
4245
| null |
Default
| null | null |
null |
{
"abstract": " The recognition of actions from video sequences has many applications in\nhealth monitoring, assisted living, surveillance, and smart homes. Despite\nadvances in sensing, in particular related to 3D video, the methodologies to\nprocess the data are still subject to research. We demonstrate superior results\nby a system which combines recurrent neural networks with convolutional neural\nnetworks in a voting approach. The gated-recurrent-unit-based neural networks\nare particularly well-suited to distinguish actions based on long-term\ninformation from optical tracking data; the 3D-CNNs focus more on detailed,\nrecent information from video data. The resulting features are merged in an SVM\nwhich then classifies the movement. In this architecture, our method improves\nrecognition rates of state-of-the-art methods by 14% on standard data sets.\n",
"title": "Two-Stream RNN/CNN for Action Recognition in 3D Videos"
}
| null | null | null | null | true | null |
4246
| null |
Default
| null | null |
null |
{
"abstract": " Digital information can be encoded in the building-block sequence of\nmacromolecules, such as RNA and single-stranded DNA. Methods of \"writing\" and\n\"reading\" macromolecular strands are currently available, but they are slow and\nexpensive. In an ideal molecular data storage system, routine operations such\nas write, read, erase, store, and transfer must be done reliably and at high\nspeed within an integrated chip. As a first step toward demonstrating the\nfeasibility of this concept, we report preliminary results of DNA readout\nexperiments conducted in miniaturized chambers that are scalable to even\nsmaller dimensions. We show that translocation of a single-stranded DNA\nmolecule (consisting of 50 adenosine bases followed by 100 cytosine bases)\nthrough an ion-channel yields a characteristic signal that is attributable to\nthe 2-segment structure of the molecule. We also examine the dependence of the\ntranslocation rate and speed on the adjustable parameters of the experiment.\n",
"title": "DNA translocation through alpha-haemolysin nano-pores with potential application to macromolecular data storage"
}
| null | null | null | null | true | null |
4247
| null |
Default
| null | null |
null |
{
"abstract": " Traditionally categorical data analysis (e.g. generalized linear models)\nworks with simple, flat datasets akin to a single table in a database with no\nnotion of missing data or conflicting versions. In contrast, modern data\nanalysis must deal with distributed databases with many partial local tables\nthat need not always agree. The computational agents tabulating these tables\nare spatially separated, with binding speed-of-light constraints and data\narriving too rapidly for these distributed views ever to be fully informed and\nglobally consistent. Contextuality is a mathematical property which describes a\nkind of inconsistency arising in quantum mechanics (e.g. in Bell's theorem). In\nthis paper we show how contextuality can arise in common data collection\nscenarios, including missing data and versioning (as in low-latency distributed\ndatabases employing snapshot isolation). In the companion paper, we develop\nstatistical models adapted to this regime.\n",
"title": "Contextuality from missing and versioned data"
}
| null | null | null | null | true | null |
4248
| null |
Default
| null | null |
null |
{
"abstract": " In this article we present an idea of using liquid scintillator Cherenkov\nneutrino detectors to detect the mantle and K-40 components of geoneutrinos.\nLiquid scintillator Cherenkov detectors feature both energy and direction\nmeasurement for charge particles. Geoneutrinos can be detected with the elastic\nscattering process of neutrino and electron. With the directionality, the\ndominant intrinsic background originated from solar neutrinos in common liquid\nscintillator detectors can be suppressed. The mantle geoneutrinos can be\ndistinguished because they come mainly underneath. The K-40 geoneutrinos can\nalso be identified, if the detection threshold for direction measurement can be\nlower than, for example, 0.8 MeV. According to our calculation, a moderate,\nkilo-ton scale, detector can observe tens of candidates, and is a practical\nstart for an experiment.\n",
"title": "Reveal the Mantle and K-40 Components of Geoneutrinos with Liquid Scintillator Cherenkov Neutrino Detectors"
}
| null | null | null | null | true | null |
4249
| null |
Default
| null | null |
null |
{
"abstract": " This paper contributes a new machine learning solution for stock movement\nprediction, which aims to predict whether the price of a stock will be up or\ndown in the near future. The key novelty is that we propose to employ\nadversarial training to improve the generalization of a recurrent neural\nnetwork model. The rationality of adversarial training here is that the input\nfeatures to stock prediction are typically based on stock price, which is\nessentially a stochastic variable and continuously changed with time by nature.\nAs such, normal training with stationary price-based features (e.g. the closing\nprice) can easily overfit the data, being insufficient to obtain reliable\nmodels. To address this problem, we propose to add perturbations to simulate\nthe stochasticity of continuous price variable, and train the model to work\nwell under small yet intentional perturbations. Extensive experiments on two\nreal-world stock data show that our method outperforms the state-of-the-art\nsolution with 3.11% relative improvements on average w.r.t. accuracy, verifying\nthe usefulness of adversarial training for stock prediction task. Codes will be\nmade available upon acceptance.\n",
"title": "Improving Stock Movement Prediction with Adversarial Training"
}
| null | null | null | null | true | null |
4250
| null |
Default
| null | null |
null |
{
"abstract": " We characterize strong type and weak type inequalities with two weights for\npositive operators on filtered measure spaces. These estimates are\nprobabilistic analogues of two-weight inequalities for positive operators\nassociated to the dyadic cubes in $\\mathbb R^n$ due to Lacey, Sawyer and\nUriarte-Tuero \\cite{LaSaUr}. Several mixed bounds for the Doob maximal operator\non filtered measure spaces are also obtained. In fact, Hytönen-Pérez\ntype and Lerner-Moen type norm estimates for Doob maximal operator are\nestablished. Our approaches are mainly based on the construction of principal\nsets.\n",
"title": "Weighted estimates for positive operators and Doob maximal operators on filtered measure spaces"
}
| null | null | null | null | true | null |
4251
| null |
Default
| null | null |
null |
{
"abstract": " Sentiment analysis is the Natural Language Processing (NLP) task dealing with\nthe detection and classification of sentiments in texts. While some tasks deal\nwith identifying the presence of sentiment in the text (Subjectivity analysis),\nother tasks aim at determining the polarity of the text categorizing them as\npositive, negative and neutral. Whenever there is a presence of sentiment in\nthe text, it has a source (people, group of people or any entity) and the\nsentiment is directed towards some entity, object, event or person. Sentiment\nanalysis tasks aim to determine the subject, the target and the polarity or\nvalence of the sentiment. In our work, we try to automatically extract\nsentiment (positive or negative) from Facebook posts using a machine learning\napproach.While some works have been done in code-mixed social media data and in\nsentiment analysis separately, our work is the first attempt (as of now) which\naims at performing sentiment analysis of code-mixed social media text. We have\nused extensive pre-processing to remove noise from raw text. Multilayer\nPerceptron model has been used to determine the polarity of the sentiment. We\nhave also developed the corpus for this task by manually labeling Facebook\nposts with their associated sentiments.\n",
"title": "Sentiment Identification in Code-Mixed Social Media Text"
}
| null | null |
[
"Computer Science"
] | null | true | null |
4252
| null |
Validated
| null | null |
null |
{
"abstract": " Separating two sources from an audio mixture is an important task with many\napplications. It is a challenging problem since only one signal channel is\navailable for analysis. In this paper, we propose a novel framework for singing\nvoice separation using the generative adversarial network (GAN) with a\ntime-frequency masking function. The mixture spectra is considered to be a\ndistribution and is mapped to the clean spectra which is also considered a\ndistribtution. The approximation of distributions between mixture spectra and\nclean spectra is performed during the adversarial training process. In contrast\nwith current deep learning approaches for source separation, the parameters of\nthe proposed framework are first initialized in a supervised setting and then\noptimized by the training procedure of GAN in an unsupervised setting.\nExperimental results on three datasets (MIR-1K, iKala and DSD100) show that\nperformance can be improved by the proposed framework consisting of\nconventional networks.\n",
"title": "SVSGAN: Singing Voice Separation via Generative Adversarial Network"
}
| null | null | null | null | true | null |
4253
| null |
Default
| null | null |
null |
{
"abstract": " In this article, we first derive the wavevector- and frequency-dependent,\nmicroscopic current response tensor which corresponds to the \"macroscopic\"\nansatz $\\vec D = \\varepsilon_0 \\varepsilon_{\\mathrm{eff}} \\vec E$ and $\\vec B =\n\\mu_0 \\mu_{\\mathrm{eff}} \\vec H$ with wavevector- and frequency-independent,\n\"effective\" material constants $\\varepsilon_{\\mathrm{eff}}$ and\n$\\mu_{\\mathrm{eff}}$. We then deduce the electromagnetic and optical properties\nof this effective material model by employing exact, microscopic response\nrelations. In particular, we argue that for recovering the standard relation\n$n^2 = \\varepsilon_{\\mathrm{eff}} \\mu_{\\mathrm{eff}}$ between the refractive\nindex and the effective material constants, it is imperative to start from the\nmicroscopic wave equation in terms of the transverse dielectric function,\n$\\varepsilon_{\\mathrm{T}}(\\vec k, \\omega) = 0$. On the phenomenological side,\nour result is especially relevant for metamaterials research, which draws\ndirectly on the standard relation for the refractive index in terms of\neffective material constants. Since for a wide class of materials the current\nresponse tensor can be calculated from first principles and compared to the\nmodel expression derived here, this work also paves the way for a systematic\nsearch for new metamaterials.\n",
"title": "Microscopic theory of refractive index applied to metamaterials: Effective current response tensor corresponding to standard relation $n^2 = \\varepsilon_{\\mathrm{eff}} μ_{\\mathrm{eff}}$"
}
| null | null | null | null | true | null |
4254
| null |
Default
| null | null |
null |
{
"abstract": " An optimization procedure for multi-transmitter (MISO) wireless power\ntransfer (WPT) systems based on tight semidefinite relaxation (SDR) is\npresented. This method ensures physical realizability of MISO WPT systems\ndesigned via convex optimization -- a robust, semi-analytical and intuitive\nroute to optimizing such systems. To that end, the nonconvex constraints\nrequiring that power is fed into rather than drawn from the system via all\ntransmitter ports are incorporated in a convex semidefinite relaxation, which\nis efficiently and reliably solvable by dedicated algorithms. A test of the\nsolution then confirms that this modified problem is equivalent (tight\nrelaxation) to the original (nonconvex) one and that the true global optimum\nhas been found. This is a clear advantage over global optimization methods\n(e.g. genetic algorithms), where convergence to the true global optimum cannot\nbe ensured or tested. Discussions of numerical results yielded by both the\nclosed-form expressions and the refined technique illustrate the importance and\npracticability of the new method. It, is shown that this technique offers a\nrigorous optimization framework for a broad range of current and emerging WPT\napplications.\n",
"title": "Semidefinite Relaxation-Based Optimization of Multiple-Input Wireless Power Transfer Systems"
}
| null | null | null | null | true | null |
4255
| null |
Default
| null | null |
null |
{
"abstract": " We consider the problem of estimating counterfactual quantities when prior\nknowledge is available in the form of disjunctive statements. These include\ndisjunction of conditions (e.g., \"the patient is more than 60 years of age\") as\nwell as disjuction of antecedants (e.g., \"had the patient taken either drug A\nor drug B\"). Focusing on linear structural equation models (SEM) and imperfect\ncontrol plans, we extend the counterfactual framework of Balke and Pearl (1995)\n, Chen and Pearl (2015), and Pearl (2009, pp. 389-391) from unconditional to\nconditional plans, from a univariate treatment to a set of treatments, and from\npoint type knowledge to disjunctive knowledge. Finally, we provide improved\nmatrix representations of the resulting counterfactual parameters, and improved\ncomputational methods of their evaluation.\n",
"title": "Counterfactual Reasoning with Disjunctive Knowledge in a Linear Structural Equation Model"
}
| null | null | null | null | true | null |
4256
| null |
Default
| null | null |
null |
{
"abstract": " Text generation is increasingly common but often requires manual post-editing\nwhere high precision is critical to end users. However, manual editing is\nexpensive so we want to ensure this effort is focused on high-value tasks. And\nwe want to maintain stylistic consistency, a particular challenge in crowd\nsettings. We present a case study, analysing human post-editing in the context\nof a template-based biography generation system. An edit flow visualisation\ncombined with manual characterisation of edits helps identify and prioritise\nwork for improving end-to-end efficiency and accuracy.\n",
"title": "Post-edit Analysis of Collective Biography Generation"
}
| null | null | null | null | true | null |
4257
| null |
Default
| null | null |
null |
{
"abstract": " A means of building safe critical systems consists of formally modeling the\nrequirements formulated by stakeholders and ensuring their consistency with\nrespect to application domain properties. This paper proposes a metamodel for\nan ontology modeling formalism based on OWL and PLIB. This modeling formalism\nis part of a method for modeling the domain of systems whose requirements are\ncaptured through SysML/KAOS. The formal semantics of SysML/KAOS goals are\nrepresented using Event-B specifications. Goals provide the set of events,\nwhile domain models will provide the structure of the system state of the\nEvent-B specification. Our proposal is illustrated through a case study dealing\nwith a Cycab localization component specification. The case study deals with\nthe specification of a localization software component that uses GPS,Wi-Fi and\nsensor technologies for the realtime localization of the Cycab vehicle, an\nautonomous ground transportation system designed to be robust and completely\nindependent.\n",
"title": "The SysML/KAOS Domain Modeling Approach"
}
| null | null |
[
"Computer Science"
] | null | true | null |
4258
| null |
Validated
| null | null |
null |
{
"abstract": " The topic of this paper is modeling and analyzing dependence in stochastic\nsocial networks. Using a latent variable block model allows the analysis of\ndependence between blocks via the analysis of a latent graphical model. Our\napproach to the analysis of the graphical model then is based on the idea\nunderlying the neighborhood selection scheme put forward by Meinshausen and\nBühlmann (2006). However, because of the latent nature of our model,\nestimates have to be used in lieu of the unobserved variables. This leads to a\nnovel analysis of graphical models under uncertainty, in the spirit of\nRosenbaum et al. (2010), or Belloni et al. (2017). Lasso-based selectors, and a\nclass of Dantzig-type selectors are studied.\n",
"title": "Neighborhood selection with application to social networks"
}
| null | null | null | null | true | null |
4259
| null |
Default
| null | null |
null |
{
"abstract": " Nodal-line semimetals, one of the topological semimetals, have degeneracy\nalong nodal lines where the band gap is closed. In many cases, the nodal lines\nappear accidentally, and in such cases it is impossible to determine whether\nthe nodal lines appear or not, only from the crystal symmetry and the electron\nfilling. In this paper, for spinless systems, we show that in specific space\ngroups at $4N+2$ fillings ($8N+4$ fillings including the spin degree of\nfreedom), presence of the nodal lines is required regardless of the details of\nthe systems. Here, the spinless systems refer to crystals where the spin-orbit\ncoupling is negligible and the spin degree of freedom can be omitted because of\nthe SU(2) spin degeneracy. In this case the shape of the band structure around\nthese nodal lines is like an hourglass, and we call this a spinless hourglass\nnodal-line semimetal. We construct a model Hamiltonian as an example and we\nshow that it is always in the spinless hourglass nodal-line semimetal phase\neven when the model parameters are changed without changing the symmetries of\nthe system. We also establish a list of all the centrosymmetric space groups,\nunder which spinless systems always have hourglass nodal lines, and illustrate\nwhere the nodal lines are located. We propose that Al$_3$FeSi$_2$, whose\nspace-group symmetry is Pbcn (No. 60), is one of the nodal-line semimetals\narising from this mechanism.\n",
"title": "Spinless hourglass nodal-line semimetals"
}
| null | null | null | null | true | null |
4260
| null |
Default
| null | null |
null |
{
"abstract": " We analyze the list-decodability, and related notions, of random linear\ncodes. This has been studied extensively before: there are many different\nparameter regimes and many different variants. Previous works have used\ncomplementary styles of arguments---which each work in their own parameter\nregimes but not in others---and moreover have left some gaps in our\nunderstanding of the list-decodability of random linear codes. In particular,\nnone of these arguments work well for list-recovery, a generalization of\nlist-decoding that has been useful in a variety of settings.\nIn this work, we present a new approach, which works across parameter regimes\nand further generalizes to list-recovery. Our main theorem can establish better\nlist-decoding and list-recovery results for low-rate random linear codes over\nlarge fields; list-recovery of high-rate random linear codes; and it can\nrecover the rate bounds of Guruswami, Hastad, and Kopparty for constant-rate\nrandom linear codes (although with large list sizes).\n",
"title": "Average-radius list-recovery of random linear codes: it really ties the room together"
}
| null | null | null | null | true | null |
4261
| null |
Default
| null | null |
null |
{
"abstract": " Comparing different neural network representations and determining how\nrepresentations evolve over time remain challenging open questions in our\nunderstanding of the function of neural networks. Comparing representations in\nneural networks is fundamentally difficult as the structure of representations\nvaries greatly, even across groups of networks trained on identical tasks, and\nover the course of training. Here, we develop projection weighted CCA\n(Canonical Correlation Analysis) as a tool for understanding neural networks,\nbuilding off of SVCCA, a recently proposed method (Raghu et al., 2017). We\nfirst improve the core method, showing how to differentiate between signal and\nnoise, and then apply this technique to compare across a group of CNNs,\ndemonstrating that networks which generalize converge to more similar\nrepresentations than networks which memorize, that wider networks converge to\nmore similar solutions than narrow networks, and that trained networks with\nidentical topology but different learning rates converge to distinct clusters\nwith diverse representations. We also investigate the representational dynamics\nof RNNs, across both training and sequential timesteps, finding that RNNs\nconverge in a bottom-up pattern over the course of training and that the hidden\nstate is highly variable over the course of a sequence, even when accounting\nfor linear transforms. Together, these results provide new insights into the\nfunction of CNNs and RNNs, and demonstrate the utility of using CCA to\nunderstand representations.\n",
"title": "Insights on representational similarity in neural networks with canonical correlation"
}
| null | null | null | null | true | null |
4262
| null |
Default
| null | null |
null |
{
"abstract": " We axiomatize and study the matrices of type $H\\in M_N(A)$, having unitary\nentries, $H_{ij}\\in U(A)$, and whose rows and columns are subject to\northogonality type conditions. Here $A$ can be any $C^*$-algebra, for instance\n$A=\\mathbb C$, where we obtain the usual complex Hadamard matrices, or\n$A=C(X)$, where we obtain the continuous families of complex Hadamard matrices.\nOur formalism allows the construction of a quantum permutation group $G\\subset\nS_N^+$, whose structure and computation is discussed here.\n",
"title": "Complex Hadamard matrices with noncommutative entries"
}
| null | null | null | null | true | null |
4263
| null |
Default
| null | null |
null |
{
"abstract": " This paper argues that the judicial use of formal language theory and\ngrammatical inference are invaluable tools in understanding how deep neural\nnetworks can and cannot represent and learn long-term dependencies in temporal\nsequences. Learning experiments were conducted with two types of Recurrent\nNeural Networks (RNNs) on six formal languages drawn from the Strictly Local\n(SL) and Strictly Piecewise (SP) classes. The networks were Simple RNNs\n(s-RNNs) and Long Short-Term Memory RNNs (LSTMs) of varying sizes. The SL and\nSP classes are among the simplest in a mathematically well-understood hierarchy\nof subregular classes. They encode local and long-term dependencies,\nrespectively. The grammatical inference algorithm Regular Positive and Negative\nInference (RPNI) provided a baseline. According to earlier research, the LSTM\narchitecture should be capable of learning long-term dependencies and should\noutperform s-RNNs. The results of these experiments challenge this narrative.\nFirst, the LSTMs' performance was generally worse in the SP experiments than in\nthe SL ones. Second, the s-RNNs out-performed the LSTMs on the most complex SP\nexperiment and performed comparably to them on the others.\n",
"title": "Subregular Complexity and Deep Learning"
}
| null | null | null | null | true | null |
4264
| null |
Default
| null | null |
null |
{
"abstract": " Topological crystalline insulators have been recently predicted and observed\nin rock-salt structure SnSe $\\{111\\}$ thin films. Previous studies have\nsuggested that the Se-terminated surface of this thin film with hydrogen\npassivation, has a reduced surface energy and is thus a preferred\nconfiguration. In this paper, synchrotron-based angle-resolved photoemission\nspectroscopy, along with density functional theory calculations, are used to\ndemonstrate conclusively that a rock-salt SnSe $\\{111\\}$ thin film\nepitaxially-grown on \\ce{Bi2Se3} has a stable Sn-terminated surface. These\nobservations are supported by low energy electron diffraction (LEED)\nintensity-voltage measurements and dynamical LEED calculations, which further\nshow that the Sn-terminated SnSe $\\{111\\}$ thin film has undergone a surface\nstructural relaxation of the interlayer spacing between the Sn and Se atomic\nplanes. In sharp contrast to the Se-terminated counterpart, the observed Dirac\nsurface state in the Sn-terminated SnSe $\\{111\\}$ thin film is shown to yield a\nhigh Fermi velocity, $0.50\\times10^6$m/s, which suggests a potential mechanism\nof engineering the Dirac surface state of topological materials by tuning the\nsurface configuration.\n",
"title": "Observation of oscillatory relaxation in the Sn-terminated surface of epitaxial rock-salt SnSe $\\{111\\}$ topological crystalline insulator"
}
| null | null |
[
"Physics"
] | null | true | null |
4265
| null |
Validated
| null | null |
null |
{
"abstract": " Experimental records of active bundle motility are used to demonstrate the\npresence of a low-dimensional chaotic attractor in hair cell dynamics.\nDimensionality tests from dynamic systems theory are applied to estimate the\nnumber of independent variables sufficient for modeling the hair cell response.\nPoincare maps are constructed to observe a quasiperiodic transition from chaos\nto order with increasing amplitudes of mechanical forcing. The onset of this\ntransition is accompanied by a reduction of Kolmogorov entropy in the system\nand an increase in mutual information between the stimulus and the hair bundle,\nindicative of signal detection. A simple theoretical model is used to describe\nthe observed chaotic dynamics. The model exhibits an enhancement of sensitivity\nto weak stimuli when the system is poised in the chaotic regime. We propose\nthat chaos may play a role in the hair cell's ability to detect low-amplitude\nsounds.\n",
"title": "Chaotic Dynamics of Inner Ear Hair Cells"
}
| null | null | null | null | true | null |
4266
| null |
Default
| null | null |
null |
{
"abstract": " Weyl and Dirac (semi)metals in three dimensions have robust gapless\nelectronic band structures. Their massless single-body energy spectra are\nprotected by symmetries such as lattice translation, (screw) rotation and time\nreversal. In this manuscript, we discuss many-body interactions in these\nsystems. We focus on strong interactions that preserve symmetries and are\noutside the single-body mean-field regime. By mapping a Dirac (semi)metal to a\nmodel based on a three dimensional array of coupled Dirac wires, we show (1)\nthe Dirac (semi)metal can acquire a many-body excitation energy gap without\nbreaking the relevant symmetries, and (2) interaction can enable an anomalous\nWeyl (semi)metallic phase that is otherwise forbidden by symmetries in the\nsingle-body setting and can only be present holographically on the boundary of\na four dimensional weak topological insulator. Both of these topological states\nsupport fractional gapped (gapless) bulk (resp. boundary) quasiparticle\nexcitations.\n",
"title": "From Dirac semimetals to topological phases in three dimensions: a coupled wire construction"
}
| null | null | null | null | true | null |
4267
| null |
Default
| null | null |
null |
{
"abstract": " We have developed FFT beamforming techniques for the CHIME radio telescope,\nto search for and localize the astrophysical signals from Fast Radio Bursts\n(FRBs) over a large instantaneous field-of-view (FOV) while maintaining the\nfull angular resolution of CHIME. We implement a hybrid beamforming pipeline in\na GPU correlator, synthesizing 256 FFT-formed beams in the North-South\ndirection by four formed beams along East-West via exact phasing, tiling a sky\narea of ~250 square degrees. A zero-padding approximation is employed to\nimprove chromatic beam alignment across the wide bandwidth of 400 to 800 MHz.\nWe up-channelize the data in order to achieve fine spectral resolution of\n$\\Delta\\nu$=24 kHz and time cadence of 0.983 ms, desirable for detecting\ntransient and dispersed signals such as those from FRBs.\n",
"title": "CHIME FRB: An application of FFT beamforming for a radio telescope"
}
| null | null | null | null | true | null |
4268
| null |
Default
| null | null |
null |
{
"abstract": " Magnetic Resonance Imaging (MRI) and Positron Emission Tomography (PET)\nautomatic 3-D registration is implemented and validated for small animal image\nvolumes so that the high-resolution anatomical MRI information can be fused\nwith the low spatial resolution of functional PET information for the\nlocalization of lesion that is currently in high demand in the study of tumor\nof cancer (oncology) and its corresponding preparation of pharmaceutical drugs.\nThough many registration algorithms are developed and applied on human brain\nvolumes, these methods may not be as efficient on small animal datasets due to\nlack of intensity information and often the high anisotropy in voxel\ndimensions. Therefore, a fully automatic registration algorithm which can\nregister not only assumably rigid small animal volumes such as brain but also\ndeformable organs such as kidney, cardiac and chest is developed using a\ncombination of global affine and local B-spline transformation models in which\nmutual information is used as a similarity criterion. The global affine\nregistration uses a multi-resolution pyramid on image volumes of 3 levels\nwhereas in local B-spline registration, a multi-resolution scheme is applied on\nthe B-spline grid of 2 levels on the finest resolution of the image volumes in\nwhich only the transform itself is affected rather than the image volumes.\nSince mutual information lacks sufficient spatial information, PCA is used to\ninject it by estimating initial translation and rotation parameters. It is\ncomputationally efficient since it is implemented using C++ and ITK library,\nand is qualitatively and quantitatively shown that this PCA-initialized global\nregistration followed by local registration is in close agreement with expert\nmanual registration and outperforms the one without PCA initialization tested\non small animal brain and kidney.\n",
"title": "MRI-PET Registration with Automated Algorithm in Pre-clinical Studies"
}
| null | null |
[
"Computer Science"
] | null | true | null |
4269
| null |
Validated
| null | null |
null |
{
"abstract": " The goal of this paper is to extend the classical and multiplicative\nfractional derivatives. For this purpose, it is introduced the new extended\nmodified Bessel function and also given an important relation between this new\nfunction I(v,q;x) and the confluent hypergeometric function. Besides, it is\nused to generalize the hypergeometric, the confluent hypergeometric and the\nextended beta functions by using the new extended modified Bessel function.\nAlso, the asymptotic formulae and the generating function of the extended\nmodified Bessel function are obtained. The extensions of classical and\nmultiplicative fractional derivatives are defined via extended modified Bessel\nfunction and, first time the fractional derivative of rational functions is\nexplicitly given via complex partial fraction decomposition.\n",
"title": "Generalization of Special Functions and its Applications to Multiplicative and Ordinary Fractional Derivatives"
}
| null | null | null | null | true | null |
4270
| null |
Default
| null | null |
null |
{
"abstract": " Considering the problem of color distortion caused by the defogging algorithm\nbased on dark channel prior, an improved algorithm was proposed to calculate\nthe transmittance of all channels respectively. First, incident light\nfrequency's effect on the transmittance of various color channels was analyzed\naccording to the Beer-Lambert's Law, from which a proportion among various\nchannel transmittances was derived; afterwards, images were preprocessed by\ndown-sampling to refine transmittance, and then the original size was restored\nto enhance the operational efficiency of the algorithm; finally, the\ntransmittance of all color channels was acquired in accordance with the\nproportion, and then the corresponding transmittance was used for image\nrestoration in each channel. The experimental results show that compared with\nthe existing algorithm, this improved image defogging algorithm could make\nimage colors more natural, solve the problem of slightly higher color\nsaturation caused by the existing algorithm, and shorten the operation time by\nfour to nine times.\n",
"title": "Incident Light Frequency-based Image Defogging Algorithm"
}
| null | null | null | null | true | null |
4271
| null |
Default
| null | null |
null |
{
"abstract": " We present a generalisation of C. Bishop and P. Jones' result in [BJ1], where\nthey give a characterisation of the tangent points of a Jordan curve in terms\nof $\\beta$ numbers. Instead of the $L^\\infty$ Jones' $\\beta$ numbers, we use an\naveraged version of them, firstly introduced by J. Azzam and R. Schul in [AS1].\nA fundamental tool in the proof will be the Reifenberg parameterisation Theorem\nof G. David and T. Toro (see [DT1]).\n",
"title": "Tangent points of d-lower content regular sets and $β$ numbers"
}
| null | null | null | null | true | null |
4272
| null |
Default
| null | null |
null |
{
"abstract": " Flexible estimation of heterogeneous treatment effects lies at the heart of\nmany statistical challenges, such as personalized medicine and optimal resource\nallocation. In this paper, we develop a general class of two-step algorithms\nfor heterogeneous treatment effect estimation in observational studies. We\nfirst estimate marginal effects and treatment propensities in order to form an\nobjective function that isolates the causal component of the signal. Then, we\noptimize this data-adaptive objective function. Our approach has several\nadvantages over existing methods. From a practical perspective, our method is\nflexible and easy to use: In both steps, we can use any loss-minimization\nmethod, e.g., penalized regression, deep neutral networks, or boosting;\nmoreover, these methods can be fine-tuned by cross validation. Meanwhile, in\nthe case of penalized kernel regression, we show that our method has a\nquasi-oracle property: Even if the pilot estimates for marginal effects and\ntreatment propensities are not particularly accurate, we achieve the same error\nbounds as an oracle who has a priori knowledge of these two nuisance\ncomponents. We implement variants of our approach based on both penalized\nregression and boosting in a variety of simulation setups, and find promising\nperformance relative to existing baselines.\n",
"title": "Quasi-Oracle Estimation of Heterogeneous Treatment Effects"
}
| null | null | null | null | true | null |
4273
| null |
Default
| null | null |
null |
{
"abstract": " In hierarchical searches for continuous gravitational waves, clustering of\ncandidates is an important postprocessing step because it reduces the number of\nnoise candidates that are followed-up at successive stages [1][7][12]. Previous\nclustering procedures bundled together nearby candidates ascribing them to the\nsame root cause (be it a signal or a disturbance), based on a predefined\ncluster volume. In this paper, we present a procedure that adapts the cluster\nvolume to the data itself and checks for consistency of such volume with what\nis expected from a signal. This significantly improves the noise rejection\ncapabilities at fixed detection threshold, and at fixed computing resources for\nthe follow-up stages, this results in an overall more sensitive search. This\nnew procedure was employed in the first Einstein@Home search on data from the\nfirst science run of the advanced LIGO detectors (O1) [11].\n",
"title": "Adaptive clustering procedure for continuous gravitational wave searches"
}
| null | null | null | null | true | null |
4274
| null |
Default
| null | null |
null |
{
"abstract": " We study the heat trace for both the drifting Laplacian as well as\nSchrödinger operators on compact Riemannian manifolds. In the case of a\nfinite regularity potential or weight function, we prove the existence of a\npartial (six term) asymptotic expansion of the heat trace for small times as\nwell as a suitable remainder estimate. We also demonstrate that the more\nprecise asymptotic behavior of the remainder is determined by and conversely\ndistinguishes higher (Sobolev) regularity on the potential or weight function.\nIn the case of a smooth weight function, we determine the full asymptotic\nexpansion of the heat trace for the drifting Laplacian for small times. We then\nuse the heat trace to study the asymptotics of the eigenvalue counting\nfunction. In both cases the Weyl law coincides with the Weyl law for the\nRiemannian manifold with the standard Laplace-Beltrami operator. We conclude by\ndemonstrating isospectrality results for the drifting Laplacian on compact\nmanifolds.\n",
"title": "The heat trace for the drifting Laplacian and Schrödinger operators on manifolds"
}
| null | null | null | null | true | null |
4275
| null |
Default
| null | null |
null |
{
"abstract": " As online systems based on machine learning are offered to public or paid\nsubscribers via application programming interfaces (APIs), they become\nvulnerable to frequent exploits and attacks. This paper studies adversarial\nmachine learning in the practical case when there are rate limitations on API\ncalls. The adversary launches an exploratory (inference) attack by querying the\nAPI of an online machine learning system (in particular, a classifier) with\ninput data samples, collecting returned labels to build up the training data,\nand training an adversarial classifier that is functionally equivalent and\nstatistically close to the target classifier. The exploratory attack with\nlimited training data is shown to fail to reliably infer the target classifier\nof a real text classifier API that is available online to the public. In\nreturn, a generative adversarial network (GAN) based on deep learning is built\nto generate synthetic training data from a limited number of real training data\nsamples, thereby extending the training data and improving the performance of\nthe inferred classifier. The exploratory attack provides the basis to launch\nthe causative attack (that aims to poison the training process) and evasion\nattack (that aims to fool the classifier into making wrong decisions) by\nselecting training and test data samples, respectively, based on the confidence\nscores obtained from the inferred classifier. These stealth attacks with small\nfootprint (using a small number of API calls) make adversarial machine learning\npractical under the realistic case with limited training data available to the\nadversary.\n",
"title": "Generative Adversarial Networks for Black-Box API Attacks with Limited Training Data"
}
| null | null | null | null | true | null |
4276
| null |
Default
| null | null |
null |
{
"abstract": " We show that every ($P_6$, diamond, $K_4$)-free graph is $6$-colorable.\nMoreover, we give an example of a ($P_6$, diamond, $K_4$)-free graph $G$ with\n$\\chi(G) = 6$. This generalizes some known results in the literature.\n",
"title": "Coloring ($P_6$, diamond, $K_4$)-free graphs"
}
| null | null | null | null | true | null |
4277
| null |
Default
| null | null |
null |
{
"abstract": " The nearby exoplanet Proxima Centauri b will be a prime future target for\ncharacterization, despite questions about its retention of water. Climate\nmodels with static oceans suggest that an Earth-like Proxima b could harbor a\nsmall dayside region of surface liquid water at fairly warm temperatures\ndespite its weak instellation. We present the first 3-dimensional climate\nsimulations of Proxima b with a dynamic ocean. We find that an ocean-covered\nProxima b could have a much broader area of surface liquid water but at much\ncolder temperatures than previously suggested, due to ocean heat transport and\ndepression of the freezing point by salinity. Elevated greenhouse gas\nconcentrations do not necessarily produce more open ocean area because of\npossible dynamic regime transitions. For an evolutionary path leading to a\nhighly saline present ocean, Proxima b could conceivably be an inhabited,\nmostly open ocean planet dominated by halophilic life. For an ocean planet in\n3:2 spin-orbit resonance, a permanent tropical waterbelt exists for moderate\neccentricity. Simulations of Proxima Centauri b may also be a model for the\nhabitability of planets receiving similar instellation from slightly cooler or\nwarmer stars, e.g., in the TRAPPIST-1, LHS 1140, GJ 273, and GJ 3293 systems.\n",
"title": "Habitable Climate Scenarios for Proxima Centauri b With a Dynamic Ocean"
}
| null | null | null | null | true | null |
4278
| null |
Default
| null | null |
null |
{
"abstract": " There is a long-standing belief that the modular tensor categories\n$\\mathcal{C}(\\mathfrak{g},k)$, for $k\\in\\mathbb{Z}_{\\geq1}$ and\nfinite-dimensional simple complex Lie algebras $\\mathfrak{g}$, contain\nexceptional connected étale algebras at only finitely many levels $k$. This\npremise has known implications for the study of relations in the Witt group of\nnondegenerate braided fusion categories, modular invariants of conformal field\ntheories, and the classification of subfactors in the theory of von Neumann\nalgebras. Here we confirm this conjecture when $\\mathfrak{g}$ has rank 2,\ncontributing proofs and explicit bounds when $\\mathfrak{g}$ is of type $B_2$ or\n$G_2$, adding to the previously known positive results for types $A_1$ and\n$A_2$.\n",
"title": "Level bounds for exceptional quantum subgroups in rank two"
}
| null | null |
[
"Mathematics"
] | null | true | null |
4279
| null |
Validated
| null | null |
null |
{
"abstract": " Most of mathematic forgetting curve models fit well with the forgetting data\nunder the learning condition of one time rather than repeated. In the paper, a\nconvolution model of forgetting curve is proposed to simulate the memory\nprocess during learning. In this model, the memory ability (i.e. the central\nprocedure in the working memory model) and learning material (i.e. the input in\nthe working memory model) is regarded as the system function and the input\nfunction, respectively. The status of forgetting (i.e. the output in the\nworking memory model) is regarded as output function or the convolution result\nof the memory ability and learning material. The model is applied to simulate\nthe forgetting curves in different situations. The results show that the model\nis able to simulate the forgetting curves not only in one time learning\ncondition but also in multi-times condition. The model is further verified in\nthe experiments of Mandarin tone learning for Japanese learners. And the\npredicted curve fits well on the test points.\n",
"title": "Convolution Forgetting Curve Model for Repeated Learning"
}
| null | null | null | null | true | null |
4280
| null |
Default
| null | null |
null |
{
"abstract": " An important step in the efficient computation of multi-dimensional theta\nfunctions is the construction of appropriate symplectic transformations for a\ngiven Riemann matrix assuring a rapid convergence of the theta series. An\nalgorithm is presented to approximately map the Riemann matrix to the Siegel\nfundamental domain. The shortest vector of the lattice generated by the Riemann\nmatrix is identified exactly, and the algorithm ensures that its length is\nlarger than $\\sqrt{3}/2$. The approach is based on a previous algorithm by\nDeconinck et al. using the LLL algorithm for lattice reductions. Here, the LLL\nalgorithm is replaced by exact Minkowski reductions for small genus and an\nexact identification of the shortest lattice vector for larger values of the\ngenus.\n",
"title": "Efficient computation of multidimensional theta functions"
}
| null | null | null | null | true | null |
4281
| null |
Default
| null | null |
null |
{
"abstract": " An imperative aspect of modern science is that scientific institutions act\nfor the benefit of a common scientific enterprise, rather than for the personal\ngain of individuals within them. This implies that science should not\nperpetuate existing or historical unequal social orders. Some scientific\nterminology, though, gives a very different impression. I will give two\nexamples of terminology invented recently for the field of quantum information\nwhich use language associated with subordination, slavery, and racial\nsegregation: 'ancilla qubit' and 'quantum supremacy'.\n",
"title": "The careless use of language in quantum information"
}
| null | null |
[
"Physics"
] | null | true | null |
4282
| null |
Validated
| null | null |
null |
{
"abstract": " Stardust grains recovered from meteorites provide high-precision snapshots of\nthe isotopic composition of the stellar environment in which they formed.\nAttributing their origin to specific types of stars, however, often proves\ndifficult. Intermediate-mass stars of 4-8 solar masses are expected to\ncontribute a large fraction of meteoritic stardust. However, no grains have\nbeen found with characteristic isotopic compositions expected from such stars.\nThis is a long-standing puzzle, which points to serious gaps in our\nunderstanding of the lifecycle of stars and dust in our Galaxy. Here we show\nthat the increased proton-capture rate of $^{17}$O reported by a recent\nunderground experiment leads to $^{17}$O/$^{16}$O isotopic ratios that match\nthose observed in a population of stardust grains, for proton-burning\ntemperatures of 60-80 million K. These temperatures are indeed achieved at the\nbase of the convective envelope during the late evolution of intermediate-mass\nstars of 4-8 solar masses, which reveals them as the most likely site of origin\nof the grains. This result provides the first direct evidence that these stars\ncontributed to the dust inventory from which the Solar System formed.\n",
"title": "Origin of meteoritic stardust unveiled by a revised proton-capture rate of $^{17}$O"
}
| null | null | null | null | true | null |
4283
| null |
Default
| null | null |
null |
{
"abstract": " We present a new proof of a fundamental result concerning cycles of random\npermutations which gives some intuition for the connection between Touchard\npolynomials and the Poisson distribution. We also introduce a rather novel\npermutation statistic and study its distribution. This quantity, indexed by\n$m$, is the number of sets of size $m$ fixed by the permutation. This leads to\na new and simpler derivation of the exponential generating function for the\nnumber of covers of certain multisets.\n",
"title": "Some Connections Between Cycles and Permutations that Fix a Set and Touchard Polynomials and Covers of Multisets"
}
| null | null | null | null | true | null |
4284
| null |
Default
| null | null |
null |
{
"abstract": " Animal telemetry data are often analysed with discrete time movement models\nassuming rotation in the movement. These models are defined with equidistant\ndistant time steps. However, telemetry data from marine animals are observed\nirregularly. To account for irregular data, a time-irregularised\nfirst-difference correlated random walk model with drift is introduced. The\nmodel generalizes the commonly used first-difference correlated random walk\nwith regular time steps by allowing irregular time steps, including a drift\nterm, and by allowing different autocorrelation in the two coordinates. The\nmodel is applied to data from a ringed seal collected through the Argos\nsatellite system, and is compared to related movement models through\nsimulations. Accounting for irregular data in the movement model results in\naccurate parameter estimates and reconstruction of movement paths. Measured by\ndistance, the introduced model can provide more accurate movement paths than\nthe regular time counterpart. Extracting accurate movement paths from uncertain\ntelemetry data is important for evaluating space use patterns for marine\nanimals, which in turn is crucial for management. Further, handling irregular\ndata directly in the movement model allows efficient simultaneous analysis of\nseveral animals.\n",
"title": "Generalizing the first-difference correlated random walk for marine animal movement data"
}
| null | null |
[
"Quantitative Biology"
] | null | true | null |
4285
| null |
Validated
| null | null |
null |
{
"abstract": " Chiral magnets with topologically nontrivial spin order such as Skyrmions\nhave generated enormous interest in both fundamental and applied sciences. We\nreport broadband microwave spectroscopy performed on the insulating chiral\nferrimagnet Cu$_{2}$OSeO$_{3}$. For the damping of magnetization dynamics we\nfind a remarkably small Gilbert damping parameter of about $1\\times10^{-4}$ at\n5 K. This value is only a factor of 4 larger than the one reported for the best\ninsulating ferrimagnet yttrium iron garnet. We detect a series of sharp\nresonances and attribute them to confined spin waves in the mm-sized samples.\nConsidering the small damping, insulating chiral magnets turn out to be\npromising candidates when exploring non-collinear spin structures for high\nfrequency applications.\n",
"title": "Low spin wave damping in the insulating chiral magnet Cu$_{2}$OSeO$_{3}$"
}
| null | null | null | null | true | null |
4286
| null |
Default
| null | null |
null |
{
"abstract": " For human pose estimation in monocular images, joint occlusions and\noverlapping upon human bodies often result in deviated pose predictions. Under\nthese circumstances, biologically implausible pose predictions may be produced.\nIn contrast, human vision is able to predict poses by exploiting geometric\nconstraints of joint inter-connectivity. To address the problem by\nincorporating priors about the structure of human bodies, we propose a novel\nstructure-aware convolutional network to implicitly take such priors into\naccount during training of the deep network. Explicit learning of such\nconstraints is typically challenging. Instead, we design discriminators to\ndistinguish the real poses from the fake ones (such as biologically implausible\nones). If the pose generator (G) generates results that the discriminator fails\nto distinguish from real ones, the network successfully learns the priors.\n",
"title": "Adversarial PoseNet: A Structure-aware Convolutional Network for Human Pose Estimation"
}
| null | null | null | null | true | null |
4287
| null |
Default
| null | null |
null |
{
"abstract": " Among underwater perceptual sensors, imaging sonar has been highlighted for\nits perceptual robustness underwater. The major challenge of imaging sonar,\nhowever, arises from the difficulty in defining visual features despite limited\nresolution and high noise levels. Recent developments in deep learning provide\na powerful solution for computer-vision researches using optical images.\nUnfortunately, deep learning-based approaches are not well established for\nimaging sonars, mainly due to the scant data in the training phase. Unlike the\nabundant publically available terrestrial images, obtaining underwater images\nis often costly, and securing enough underwater images for training is not\nstraightforward. To tackle this issue, this paper presents a solution to this\nfield's lack of data by introducing a novel end-to-end image-synthesizing\nmethod in the training image preparation phase. The proposed method present\nimage synthesizing scheme to the images captured by an underwater simulator.\nOur synthetic images are based on the sonar imaging models and noisy\ncharacteristics to represent the real data obtained from the sea. We validate\nthe proposed scheme by training using a simulator and by testing the simulated\nimages with real underwater sonar images obtained from a water tank and the\nsea.\n",
"title": "Deep Learning from Shallow Dives: Sonar Image Generation and Training for Underwater Object Detection"
}
| null | null | null | null | true | null |
4288
| null |
Default
| null | null |
null |
{
"abstract": " Genome-wide association studies (GWAS) have achieved great success in the\ngenetic study of Alzheimer's disease (AD). Collaborative imaging genetics\nstudies across different research institutions show the effectiveness of\ndetecting genetic risk factors. However, the high dimensionality of GWAS data\nposes significant challenges in detecting risk SNPs for AD. Selecting relevant\nfeatures is crucial in predicting the response variable. In this study, we\npropose a novel Distributed Feature Selection Framework (DFSF) to conduct the\nlarge-scale imaging genetics studies across multiple institutions. To speed up\nthe learning process, we propose a family of distributed group Lasso screening\nrules to identify irrelevant features and remove them from the optimization.\nThen we select the relevant group features by performing the group Lasso\nfeature selection process in a sequence of parameters. Finally, we employ the\nstability selection to rank the top risk SNPs that might help detect the early\nstage of AD. To the best of our knowledge, this is the first distributed\nfeature selection model integrated with group Lasso feature selection as well\nas detecting the risk genetic factors across multiple research institutions\nsystem. Empirical studies are conducted on 809 subjects with 5.9 million SNPs\nwhich are distributed across several individual institutions, demonstrating the\nefficiency and effectiveness of the proposed method.\n",
"title": "Large-scale Feature Selection of Risk Genetic Factors for Alzheimer's Disease via Distributed Group Lasso Regression"
}
| null | null | null | null | true | null |
4289
| null |
Default
| null | null |
null |
{
"abstract": " This paper shows that for any random variables $X$ and $Y$, it is possible to\nrepresent $Y$ as a function of $(X,Z)$ such that $Z$ is independent of $X$ and\n$I(X;Z|Y)\\le\\log(I(X;Y)+1)+4$ bits. We use this strong functional\nrepresentation lemma (SFRL) to establish a bound on the rate needed for\none-shot exact channel simulation for general (discrete or continuous) random\nvariables, strengthening the results by Harsha et al. and Braverman and Garg,\nand to establish new and simple achievability results for one-shot\nvariable-length lossy source coding, multiple description coding and Gray-Wyner\nsystem. We also show that the SFRL can be used to reduce the channel with state\nnoncausally known at the encoder to a point-to-point channel, which provides a\nsimple achievability proof of the Gelfand-Pinsker theorem.\n",
"title": "Strong Functional Representation Lemma and Applications to Coding Theorems"
}
| null | null |
[
"Computer Science",
"Mathematics"
] | null | true | null |
4290
| null |
Validated
| null | null |
null |
{
"abstract": " Infants are experts at playing, with an amazing ability to generate novel\nstructured behaviors in unstructured environments that lack clear extrinsic\nreward signals. We seek to replicate some of these abilities with a neural\nnetwork that implements curiosity-driven intrinsic motivation. Using a simple\nbut ecologically naturalistic simulated environment in which the agent can move\nand interact with objects it sees, the agent learns a world model predicting\nthe dynamic consequences of its actions. Simultaneously, the agent learns to\ntake actions that adversarially challenge the developing world model, pushing\nthe agent to explore novel and informative interactions with its environment.\nWe demonstrate that this policy leads to the self-supervised emergence of a\nspectrum of complex behaviors, including ego motion prediction, object\nattention, and object gathering. Moreover, the world model that the agent\nlearns supports improved performance on object dynamics prediction and\nlocalization tasks. Our results are a proof-of-principle that computational\nmodels of intrinsic motivation might account for key features of developmental\nvisuomotor learning in infants.\n",
"title": "Emergence of Structured Behaviors from Curiosity-Based Intrinsic Motivation"
}
| null | null |
[
"Statistics"
] | null | true | null |
4291
| null |
Validated
| null | null |
null |
{
"abstract": " A modern aircraft may require on the order of thousands of custom shims to\nfill gaps between structural components in the airframe that arise due to\nmanufacturing tolerances adding up across large structures. These shims are\nnecessary to eliminate gaps, maintain structural performance, and minimize\npull-down forces required to bring the aircraft into engineering nominal\nconfiguration for peak aerodynamic efficiency. Gap filling is a time-consuming\nprocess, involving either expensive by-hand inspection or computations on vast\nquantities of measurement data from increasingly sophisticated metrology\nequipment. Either case amounts to significant delays in production, with much\nof the time spent in the critical path of aircraft assembly. This work presents\nan alternative strategy for predictive shimming, based on machine learning and\nsparse sensing to first learn gap distributions from historical data, and then\ndesign optimized sparse sensing strategies to streamline data collection and\nprocessing. This new approach is based on the assumption that patterns exist in\nshim distributions across aircraft, which may be mined and used to reduce the\nburden of data collection and processing in future aircraft. Specifically,\nrobust principal component analysis is used to extract low-dimensional patterns\nin the gap measurements while rejecting outliers. Next, optimized sparse\nsensors are obtained that are most informative about the dimensions of a new\naircraft in these low-dimensional principal components. We demonstrate the\nsuccess of the proposed approach, called PIXel Identification Despite\nUncertainty in Sensor Technology (PIXI-DUST), on historical production data\nfrom 54 representative Boeing commercial aircraft. Our algorithm successfully\npredicts $99\\%$ of shim gaps within the desired measurement tolerance using\n$3\\%$ of the laser scan points typically required; all results are\ncross-validated.\n",
"title": "Predicting shim gaps in aircraft assembly with machine learning and sparse sensing"
}
| null | null | null | null | true | null |
4292
| null |
Default
| null | null |
null |
{
"abstract": " Quantitative multivariate central limit theorems for general functionals of\npossibly non-symmetric and non-homogeneous infinite Rademacher sequences are\nproved by combining discrete Malliavin calculus with the smart path method for\nnormal approximation. In particular, a discrete multivariate second-order\nPoincaré inequality is developed. As a first application, the normal\napproximation of vectors of subgraph counting statistics in the\nErdős-Rényi random graph is considered. In this context, we further\nspecialize to the normal approximation of vectors of vertex degrees. In a\nsecond application we prove a quantitative multivariate central limit theorem\nfor vectors of intrinsic volumes induced by random cubical complexes.\n",
"title": "Multivariate central limit theorems for Rademacher functionals with applications"
}
| null | null |
[
"Mathematics"
] | null | true | null |
4293
| null |
Validated
| null | null |
null |
{
"abstract": " According to the DeGroot-Friedkin model of a social network, an individual's\nsocial power evolves as the network discusses individual opinions over a\nsequence of issues. Under mild assumptions on the connectivity of the network,\nthe social power of every individual converges to a constant strictly positive\nvalue as the number of issues discussed increases. If the network has a special\ntopology, termed \"star topology\", then all social power accumulates with the\nindividual at the centre of the star. This paper studies the strategic\nintroduction of new individuals and/or interpersonal relationships into a\nsocial network with star topology to reduce the social power of the centre\nindividual. In fact, several strategies are proposed. For each strategy, we\nderive necessary and sufficient conditions on the strength of the new\ninterpersonal relationships, based on local information, which ensures that the\ncentre individual no longer has the greatest social power within the social\nnetwork. Interpretations of these conditions show that the strategies are\nremarkably intuitive and that certain strategies are favourable compared to\nothers, all of which is sociologically expected.\n",
"title": "Modification of Social Dominance in Social Networks by Selective Adjustment of Interpersonal Weights"
}
| null | null | null | null | true | null |
4294
| null |
Default
| null | null |
null |
{
"abstract": " We prove existence of Abrikosov vortex lattice solutions of the\nGinzburg-Landau equations of superconductivity, with multiple magnetic flux\nquanta per a fundamental cell. We also revisit the existence proof for the\nAbrikosov vortex lattices, streamlining some arguments and providing some\nessential details missing in earlier proofs for a single magnetic flux quantum\nper a fundamental cell.\n",
"title": "On Abrikosov Lattice Solutions of the Ginzburg-Landau Equations"
}
| null | null | null | null | true | null |
4295
| null |
Default
| null | null |
null |
{
"abstract": " The Hilda asteroids are primitive bodies in resonance with Jupiter whose\norigin and physical properties are not well understood. Current models posit\nthat these asteroids formed in the outer Solar System and were scattered along\nwith the Jupiter Trojans into their present-day positions during a chaotic\nepisode of dynamical restructuring. In order to explore the surface composition\nof these enigmatic objects in comparison with an analogous study of Trojans\n(Emery et al. 2011), we present new near-infrared spectra (0.7-2.5 $\\mu$m) of\n25 Hilda asteroids. No discernible absorption features are apparent in the\ndata. Synthesizing the bimodalities in optical color and infrared reflectivity\nreported in previous studies, we classify 26 of the 28 Hildas in our spectral\nsample into the so-called less-red and red sub-populations and find that the\ntwo sub-populations have distinct average spectral shapes. Combining our\nresults with visible spectra, we find that Trojans and Hildas possess similar\noverall spectral shapes, suggesting that the two minor body populations share a\ncommon progenitor population. A more detailed examination reveals that while\nthe red Trojans and Hildas have nearly identical spectra, less-red Hildas are\nsystematically bluer in the visible and redder in the near-infrared than\nless-red Trojans, indicating a putative broad, shallow absorption feature\nbetween 0.5 and 1.0 $\\mu$m. We argue that the less-red and red objects found in\nboth Hildas and Trojans represent two distinct surface chemistries and\nattribute the small discrepancy between less-red Hildas and Trojans to the\ndifference in surface temperatures between the two regions.\n",
"title": "$0.7-2.5~μ$m spectra of Hilda asteroids"
}
| null | null |
[
"Physics"
] | null | true | null |
4296
| null |
Validated
| null | null |
null |
{
"abstract": " The properties of cold Bose gases at unitarity have been extensively\ninvestigated in the last few years both theoretically and experimentally. In\nthis paper we use a family of interactions tuned to two-body unitarity and very\nweak three-body binding to demonstrate the universal properties of both\nclusters and matter. We determine the universal properties of finite clusters\nup to 60 particles and, for the first time, explicitly demonstrate the\nsaturation of energy and density with particle number and compare with bulk\nproperties. At saturation in the bulk we determine the energy, density, two-\nand three-body contacts and the condensate fraction. We find that uniform\nmatter is more bound than three-body clusters by nearly two orders of\nmagnitude, the two-body contact is very large in absolute terms, and yet the\ncondensate fraction is also very large, greater than 90%. Equilibrium\nproperties of these systems may be experimentally accessible through rapid\nquenching of weakly-interacting boson superfluids.\n",
"title": "Ground-state properties of unitary bosons: from clusters to matter"
}
| null | null | null | null | true | null |
4297
| null |
Default
| null | null |
null |
{
"abstract": " The best known method to give a lower bound for the Noether number of a given\nfinite group is to use the fact that it is greater than or equal to the Noether\nnumber of any of the subgroups or factor groups. The results of the present\npaper show in particular that these inequalities are strict for proper\nsubgroups or factor groups. This is established by studying the algebra of\ncoinvariants of a representation induced from a representation of a subgroup.\n",
"title": "Lower bounds on the Noether number"
}
| null | null | null | null | true | null |
4298
| null |
Default
| null | null |
null |
{
"abstract": " Education is a key factor in ensuring economic growth, especially for\ncountries with growing economies. Today, students have become more\ntechnologically savvy as teaching and learning uses more advance technology day\nin, day out. Due to virtualize resources through the Internet, as well as\ndynamic scalability, cloud computing has continued to be adopted by more\norganizations. Despite the looming financial crisis, there has been increasing\npressure for educational institutions to deliver better services using minimal\nresources. Leaning institutions, both public and private can utilize the\npotential advantage of cloud computing to ensure high quality service\nregardless of the minimal resources available. Cloud computing is taking a\ncenter stage in academia because of its various benefits. Various learning\ninstitutions use different cloud-based applications provided by the service\nproviders to ensure that their students and other users can perform both\nacademic as well as business-related tasks. Thus, this research will seek to\nestablish the benefits associated with the use of cloud computing in learning\ninstitutions. The solutions provided by the cloud technology ensure that the\nresearch and development, as well as the teaching is more sustainable and\nefficient, thus positively influencing the quality of learning and teaching\nwithin educational institutions. This has led to various learning institutions\nadopting cloud technology as a solution to various technological challenges\nthey face on a daily routine.\n",
"title": "A Survey on the Adoption of Cloud Computing in Education Sector"
}
| null | null | null | null | true | null |
4299
| null |
Default
| null | null |
null |
{
"abstract": " It has long been known that a single-layer fully-connected neural network\nwith an i.i.d. prior over its parameters is equivalent to a Gaussian process\n(GP), in the limit of infinite network width. This correspondence enables exact\nBayesian inference for infinite width neural networks on regression tasks by\nmeans of evaluating the corresponding GP. Recently, kernel functions which\nmimic multi-layer random neural networks have been developed, but only outside\nof a Bayesian framework. As such, previous work has not identified that these\nkernels can be used as covariance functions for GPs and allow fully Bayesian\nprediction with a deep neural network.\nIn this work, we derive the exact equivalence between infinitely wide deep\nnetworks and GPs. We further develop a computationally efficient pipeline to\ncompute the covariance function for these GPs. We then use the resulting GPs to\nperform Bayesian inference for wide deep neural networks on MNIST and CIFAR-10.\nWe observe that trained neural network accuracy approaches that of the\ncorresponding GP with increasing layer width, and that the GP uncertainty is\nstrongly correlated with trained network prediction error. We further find that\ntest performance increases as finite-width trained networks are made wider and\nmore similar to a GP, and thus that GP predictions typically outperform those\nof finite-width networks. Finally we connect the performance of these GPs to\nthe recent theory of signal propagation in random neural networks.\n",
"title": "Deep Neural Networks as Gaussian Processes"
}
| null | null | null | null | true | null |
4300
| null |
Default
| null | null |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.