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": " We propose a new method for input variable selection in nonlinear regression.\nThe method is embedded into a kernel regression machine that can model general\nnonlinear functions, not being a priori limited to additive models. This is the\nfirst kernel-based variable selection method applicable to large datasets. It\nsidesteps the typical poor scaling properties of kernel methods by mapping the\ninputs into a relatively low-dimensional space of random features. The\nalgorithm discovers the variables relevant for the regression task together\nwith learning the prediction model through learning the appropriate nonlinear\nrandom feature maps. We demonstrate the outstanding performance of our method\non a set of large-scale synthetic and real datasets.\n",
"title": "Large-scale Nonlinear Variable Selection via Kernel Random Features"
}
| null | null | null | null | true | null |
10801
| null |
Default
| null | null |
null |
{
"abstract": " This article proposes a new way to construct computationally efficient\n`wrappers' around fine scale, microscopic, detailed descriptions of dynamical\nsystems, such as molecular dynamics, to make predictions at the macroscale\n`continuum' level. It is often significantly easier to code a microscale\nsimulator with periodicity: so the challenge addressed here is to develop a\nscheme that uses only a given periodic microscale simulator; specifically, one\nfor atomistic dynamics. Numerical simulations show that applying a suitable\nproportional controller within `action regions' of a patch of atomistic\nsimulation effectively predicts the macroscale transport of heat. Theoretical\nanalysis establishes that such an approach will generally be effective and\nefficient, and also determines good values for the strength of the proportional\ncontroller. This work has the potential to empower systematic analysis and\nunderstanding at a macroscopic system level when only a given microscale\nsimulator is available.\n",
"title": "Couple microscale periodic patches to simulate macroscale emergent dynamics"
}
| null | null |
[
"Mathematics"
] | null | true | null |
10802
| null |
Validated
| null | null |
null |
{
"abstract": " We study a mini-batch diversification scheme for stochastic gradient descent\n(SGD). While classical SGD relies on uniformly sampling data points to form a\nmini-batch, we propose a non-uniform sampling scheme based on the Determinantal\nPoint Process (DPP). The DPP relies on a similarity measure between data points\nand gives low probabilities to mini-batches which contain redundant data, and\nhigher probabilities to mini-batches with more diverse data. This\nsimultaneously balances the data and leads to stochastic gradients with lower\nvariance. We term this approach Diversified Mini-Batch SGD (DM-SGD). We show\nthat regular SGD and a biased version of stratified sampling emerge as special\ncases. Furthermore, DM-SGD generalizes stratified sampling to cases where no\ndiscrete features exist to bin the data into groups. We show experimentally\nthat our method results more interpretable and diverse features in unsupervised\nsetups, and in better classification accuracies in supervised setups.\n",
"title": "Determinantal Point Processes for Mini-Batch Diversification"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
10803
| null |
Validated
| null | null |
null |
{
"abstract": " Despite a very long history of meteor science, our understanding of meteor\nablation and its shocked plasma physics is still far from satisfactory as we\nare still missing the microphysics of meteor shock formation and its plasma\ndynamics. Here we argue that electrons and ions in the meteor plasma above\n$\\sim$100 km altitude undergo spatial separation due to electrons being trapped\nby gyration in the Earth's magnetic field, while the ions are carried by the\nmeteor as their dynamics is dictated by collisions. This separation process\ncharges the meteor and creates a strong local electric field. We show how\nacceleration of protons in this field leads to the collisional excitation of\nionospheric N$_2$ on the scale of many 100 m. This mechanism explains the\npuzzling large halo detected around Leonid meteors, while it also fits into the\ntheoretical expectations of several other unexplained meteor related phenomena.\nWe expect our work to lead to more advanced models of meteor-ionosphere\ninteraction, combined with the electrodynamics of meteor trail evolution.\n",
"title": "Proton-induced halo formation in charged meteors"
}
| null | null |
[
"Physics"
] | null | true | null |
10804
| null |
Validated
| null | null |
null |
{
"abstract": " In this article, we study a generalisation of the Seiberg-Witten equations,\nreplacing the spinor representation with a hyperKahler manifold equipped with\ncertain symmetries. Central to this is the construction of a (non-linear) Dirac\noperator acting on the sections of the non-linear fibre-bundle. For hyperKahler\nmanifolds admitting a hyperKahler potential, we derive a transformation formula\nfor the Dirac operator under the conformal change of metric on the base\nmanifold.\nAs an application, we show that when the hyperKahler manifold is of dimension\nfour, then away from a singular set, the equations can be expressed as a second\norder PDE in terms of almost-complex structure on the base manifold and a\nconformal factor. This extends a result of Donaldson to generalised\nSeiberg-Witten equations.\n",
"title": "Generalised Seiberg-Witten equations and almost-Hermitian geometry"
}
| null | null | null | null | true | null |
10805
| null |
Default
| null | null |
null |
{
"abstract": " Hyper-Kamiokande, the next generation large water Cherenkov detector in\nJapan, is planning to use approximately 80,000 20-inch photomultiplier tubes\n(PMTs). They are one of the major cost factors of the experiment. We propose a\nnovel enhanced photon trap design based on a smaller and more economical PMT in\ncombination with wavelength shifters, dichroic mirrors, and broadband mirrors.\nGEANT4 is utilized to obtain photon collection efficiencies and timing\nresolution of the photon traps. We compare the performance of different trap\nconfigurations and sizes. Our simulations indicate an enhanced photon trap with\na 12-inch PMT can match a 20-inch PMTs collection efficiency, however at a cost\nof reduced timing resolution. The photon trap might be suitable as detection\nmodule for the outer detector with large photo coverage area.\n",
"title": "Enhanced Photon Traps for Hyper-Kamiokande"
}
| null | null | null | null | true | null |
10806
| null |
Default
| null | null |
null |
{
"abstract": " We present MILABOT: a deep reinforcement learning chatbot developed by the\nMontreal Institute for Learning Algorithms (MILA) for the Amazon Alexa Prize\ncompetition. MILABOT is capable of conversing with humans on popular small talk\ntopics through both speech and text. The system consists of an ensemble of\nnatural language generation and retrieval models, including neural network and\ntemplate-based models. By applying reinforcement learning to crowdsourced data\nand real-world user interactions, the system has been trained to select an\nappropriate response from the models in its ensemble. The system has been\nevaluated through A/B testing with real-world users, where it performed\nsignificantly better than other systems. The results highlight the potential of\ncoupling ensemble systems with deep reinforcement learning as a fruitful path\nfor developing real-world, open-domain conversational agents.\n",
"title": "A Deep Reinforcement Learning Chatbot (Short Version)"
}
| null | null |
[
"Statistics"
] | null | true | null |
10807
| null |
Validated
| null | null |
null |
{
"abstract": " Quantification is a supervised learning task that consists in predicting,\ngiven a set of classes C and a set D of unlabelled items, the prevalence (or\nrelative frequency) p(c|D) of each class c in C. Quantification can in\nprinciple be solved by classifying all the unlabelled items and counting how\nmany of them have been attributed to each class. However, this \"classify and\ncount\" approach has been shown to yield suboptimal quantification accuracy;\nthis has established quantification as a task of its own, and given rise to a\nnumber of methods specifically devised for it. We propose a recurrent neural\nnetwork architecture for quantification (that we call QuaNet) that observes the\nclassification predictions to learn higher-order \"quantification embeddings\",\nwhich are then refined by incorporating quantification predictions of simple\nclassify-and-count-like methods. We test {QuaNet on sentiment quantification on\ntext, showing that it substantially outperforms several state-of-the-art\nbaselines.\n",
"title": "A Recurrent Neural Network for Sentiment Quantification"
}
| null | null | null | null | true | null |
10808
| null |
Default
| null | null |
null |
{
"abstract": " Data noising is an effective technique for regularizing neural network\nmodels. While noising is widely adopted in application domains such as vision\nand speech, commonly used noising primitives have not been developed for\ndiscrete sequence-level settings such as language modeling. In this paper, we\nderive a connection between input noising in neural network language models and\nsmoothing in $n$-gram models. Using this connection, we draw upon ideas from\nsmoothing to develop effective noising schemes. We demonstrate performance\ngains when applying the proposed schemes to language modeling and machine\ntranslation. Finally, we provide empirical analysis validating the relationship\nbetween noising and smoothing.\n",
"title": "Data Noising as Smoothing in Neural Network Language Models"
}
| null | null | null | null | true | null |
10809
| null |
Default
| null | null |
null |
{
"abstract": " We present a generative method to estimate 3D human motion and body shape\nfrom monocular video. Under the assumption that starting from an initial pose\noptical flow constrains subsequent human motion, we exploit flow to find\ntemporally coherent human poses of a motion sequence. We estimate human motion\nby minimizing the difference between computed flow fields and the output of an\nartificial flow renderer. A single initialization step is required to estimate\nmotion over multiple frames. Several regularization functions enhance\nrobustness over time. Our test scenarios demonstrate that optical flow\neffectively regularizes the under-constrained problem of human shape and motion\nestimation from monocular video.\n",
"title": "Optical Flow-based 3D Human Motion Estimation from Monocular Video"
}
| null | null | null | null | true | null |
10810
| null |
Default
| null | null |
null |
{
"abstract": " A quasi-order is a binary, reflexive and transitive relation. In the Journal\nof Pure and Applied Algebra 45 (1987), S.M. Fakhruddin introduced the notion of\n(totally) quasi-ordered fields and showed that each such field is either an\nordered field or else a valued field. Hence, quasi-ordered fields are very well\nsuited to treat ordered and valued fields simultaneously.\nIn this note, we will prove that the same dichotomy holds for commutative\nrings with 1 as well. For that purpose we first develop an appropriate notion\nof (totally) quasi-ordered rings. Our proof of the dichotomy then exploits\nFakhruddin's result that was mentioned above.\n",
"title": "Quasi-ordered Rings"
}
| null | null | null | null | true | null |
10811
| null |
Default
| null | null |
null |
{
"abstract": " In this paper, we study the controllability and stabilizability properties of\nthe Kolmogorov forward equation of a continuous time Markov chain (CTMC)\nevolving on a finite state space, using the transition rates as the control\nparameters. Firstly, we prove small-time local and global controllability from\nand to strictly positive equilibrium configurations when the underlying graph\nis strongly connected. Secondly, we show that there always exists a locally\nexponentially stabilizing decentralized linear (density-)feedback law that\ntakes zero valu at equilibrium and respects the graph structure, provided that\nthe transition rates are allowed to be negative and the desired target density\nlies in the interior of the set of probability densities. For bidirected\ngraphs, that is, graphs where a directed edge in one direction implies an edge\nin the opposite direction, we show that this linear control law can be realized\nusing a decentralized rational feedback law of the form k(x) = a(x) +\nb(x)f(x)/g(x) that also respects the graph structure and control constraints\n(positivity and zero at equilibrium). This enables the possibility of using\nLinear Matrix Inequality (LMI) based tools to algorithmically construct\ndecentralized density feedback controllers for stabilization of a robotic swarm\nto a target task distribution with no task-switching at equilibrium, as we\ndemonstrate with several numerical examples.\n",
"title": "Mean-Field Controllability and Decentralized Stabilization of Markov Chains, Part I: Global Controllability and Rational Feedbacks"
}
| null | null | null | null | true | null |
10812
| null |
Default
| null | null |
null |
{
"abstract": " In this article, we prove some total variation inequalities for maximal\nfunctions. Our results deal with two possible generalizations of the results\ncontained in Aldaz and Pérez Lázaro's work, one of whose considers a\nvariable truncation of the maximal function, and the other one interpolates the\ncentered and the uncentered maximal functions. In both contexts, we find sharp\nconstants for the desired inequalities, which can be viewed as progress towards\nthe conjecture that the best constant for the variation inequality in the\ncentered context is one. We also provide counterexamples showing that our\nmethods do not apply outside the stated parameter ranges.\n",
"title": "Sharp total variation results for maximal functions"
}
| null | null | null | null | true | null |
10813
| null |
Default
| null | null |
null |
{
"abstract": " We study bipartite community detection in networks, or more generally the\nnetwork biclustering problem. We present a fast two-stage procedure based on\nspectral initialization followed by the application of a pseudo-likelihood\nclassifier twice. Under mild regularity conditions, we establish the weak\nconsistency of the procedure (i.e., the convergence of the misclassification\nrate to zero) under a general bipartite stochastic block model. We show that\nthe procedure is optimal in the sense that it achieves the optimal convergence\nrate that is achievable by a biclustering oracle, adaptively over the whole\nclass, up to constants. This is further formalized by deriving a minimax lower\nbound over a class of biclustering problems. The optimal rate we obtain\nsharpens some of the existing results and generalizes others to a wide regime\nof average degree growth, from sparse networks with average degrees growing\narbitrarily slowly to fairly dense networks with average degrees of order\n$\\sqrt{n}$. As a special case, we recover the known exact recovery threshold in\nthe $\\log n$ regime of sparsity. To obtain the consistency result, as part of\nthe provable version of the algorithm, we introduce a sub-block partitioning\nscheme that is also computationally attractive, allowing for distributed\nimplementation of the algorithm without sacrificing optimality. The provable\nalgorithm is derived from a general class of pseudo-likelihood biclustering\nalgorithms that employ simple EM type updates. We show the effectiveness of\nthis general class by numerical simulations.\n",
"title": "Optimal Bipartite Network Clustering"
}
| null | null | null | null | true | null |
10814
| null |
Default
| null | null |
null |
{
"abstract": " This paper demonstrates end-to-end neural network architectures for\nVietnamese named entity recognition. Our best model is a combination of\nbidirectional Long Short-Term Memory (Bi-LSTM), Convolutional Neural Network\n(CNN), Conditional Random Field (CRF), using pre-trained word embeddings as\ninput, which achieves an F1 score of 88.59% on a standard test set. Our system\nis able to achieve a comparable performance to the first-rank system of the\nVLSP campaign without using any syntactic or hand-crafted features. We also\ngive an extensive empirical study on using common deep learning models for\nVietnamese NER, at both word and character level.\n",
"title": "End-to-end Recurrent Neural Network Models for Vietnamese Named Entity Recognition: Word-level vs. Character-level"
}
| null | null | null | null | true | null |
10815
| null |
Default
| null | null |
null |
{
"abstract": " We consider the task of estimating a high-dimensional directed acyclic graph,\ngiven observations from a linear structural equation model with arbitrary noise\ndistribution. By exploiting properties of common random graphs, we develop a\nnew algorithm that requires conditioning only on small sets of variables. The\nproposed algorithm, which is essentially a modified version of the\nPC-Algorithm, offers significant gains in both computational complexity and\nestimation accuracy. In particular, it results in more efficient and accurate\nestimation in large networks containing hub nodes, which are common in\nbiological systems. We prove the consistency of the proposed algorithm, and\nshow that it also requires a less stringent faithfulness assumption than the\nPC-Algorithm. Simulations in low and high-dimensional settings are used to\nillustrate these findings. An application to gene expression data suggests that\nthe proposed algorithm can identify a greater number of clinically relevant\ngenes than current methods.\n",
"title": "The Reduced PC-Algorithm: Improved Causal Structure Learning in Large Random Networks"
}
| null | null | null | null | true | null |
10816
| null |
Default
| null | null |
null |
{
"abstract": " We investigate the deexcitation of the $^{229}$Th nucleus via the excitation\nof an electron. Detailed calculations are performed for the enhancement of the\nnuclear decay width due to this so called electron bridge (EB) compared to the\ndirect photoemission from the nucleus. The results are obtianed for triply\nionized thorium by using a B-spline pseudo basis approach to solve the Dirac\nequation for a local $x_\\alpha$ potential. This approach allows for an\napproximation of the full electron propagator including the positive and\nnegative continuum. We show that the contribution of continua slightly\nincreases the enhancement compared to a propagator calculated by a direct\nsummation over bound states. Moreover we put special emphasis on the\ninterference between the direct and exchange Feynman diagrams that can have a\nstrong influence on the enhancement.\n",
"title": "Theoretical analysis of the electron bridge process in $^{229}$Th$^{3+}$"
}
| null | null | null | null | true | null |
10817
| null |
Default
| null | null |
null |
{
"abstract": " It is well known that the memory effect in flat spacetime is parametrized by\nthe BMS supertranslation. We investigate the relation between the memory effect\nand diffeomorphism in de Sitter spacetime. We find that gravitational memory is\nparametrized by a BMS-like supertranslation in the static patch of de Sitter\nspacetime. We also show a diffeomorphism that corresponds to gravitational\nmemory in the Poincare/cosmological patch. Our method does not need to assume\nthe separation between the source and the detector to be small compared with\nthe Hubble radius, and can potentially be applicable to other FLRW universes,\nas well as \"ordinary memory\" mediated by massive messenger particles.\n",
"title": "Memory in de Sitter space and BMS-like supertranslations"
}
| null | null |
[
"Physics"
] | null | true | null |
10818
| null |
Validated
| null | null |
null |
{
"abstract": " We consider the problem of estimating mutual information between dependent\ndata, an important problem in many science and engineering applications. We\npropose a data-driven, non-parametric estimator of mutual information in this\npaper. The main novelty of our solution lies in transforming the data to\nfrequency domain to make the problem tractable. We define a novel\nmetric--mutual information in frequency--to detect and quantify the dependence\nbetween two random processes across frequency using Cramér's spectral\nrepresentation. Our solution calculates mutual information as a function of\nfrequency to estimate the mutual information between the dependent data over\ntime. We validate its performance on linear and nonlinear models. In addition,\nmutual information in frequency estimated as a part of our solution can also be\nused to infer cross-frequency coupling in the data.\n",
"title": "Data-Driven Estimation Of Mutual Information Between Dependent Data"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
10819
| null |
Validated
| null | null |
null |
{
"abstract": " While Wigner functions forming phase space representation of quantum states\nis a well-known fact, their construction for noncommutative quantum mechanics\n(NCQM) remains relatively lesser known, in particular with respect to gauge\ndependencies. This paper deals with the construction of Wigner functions of\nNCQM for a system of 2-degrees of freedom using 2-parameter families of gauge\nequivalence classes of unitary irreducible representations (UIRs) of the Lie\ngroup $\\g$ which has been identified as the kinematical symmetry group of NCQM\nin an earlier paper. This general construction of Wigner functions for NCQM, in\nturn, yields the special cases of Landau and symmetric gauges of NCQM.\n",
"title": "Wigner functions for gauge equivalence classes of unitary irreducible representations of noncommutative quantum mechanics"
}
| null | null | null | null | true | null |
10820
| null |
Default
| null | null |
null |
{
"abstract": " Intracranial carotid artery calcification (ICAC) is a major risk factor for\nstroke, and might contribute to dementia and cognitive decline. Reliance on\ntime-consuming manual annotation of ICAC hampers much demanded further research\ninto the relationship between ICAC and neurological diseases. Automation of\nICAC segmentation is therefore highly desirable, but difficult due to the\nproximity of the lesions to bony structures with a similar attenuation\ncoefficient. In this paper, we propose a method for automatic segmentation of\nICAC; the first to our knowledge. Our method is based on a 3D fully\nconvolutional neural network that we extend with two regularization techniques.\nFirstly, we use deep supervision (hidden layers supervision) to encourage\ndiscriminative features in the hidden layers. Secondly, we augment the network\nwith skip connections, as in the recently developed ResNet, and dropout layers,\ninserted in a way that skip connections circumvent them. We investigate the\neffect of skip connections and dropout. In addition, we propose a simple\nproblem-specific modification of the network objective function that restricts\nthe focus to the most important image regions and simplifies the optimization.\nWe train and validate our model using 882 CT scans and test on 1,000. Our\nregularization techniques and objective improve the average Dice score by 7.1%,\nyielding an average Dice of 76.2% and 97.7% correlation between predicted ICAC\nvolumes and manual annotations.\n",
"title": "Segmentation of Intracranial Arterial Calcification with Deeply Supervised Residual Dropout Networks"
}
| null | null |
[
"Computer Science"
] | null | true | null |
10821
| null |
Validated
| null | null |
null |
{
"abstract": " The field of structural bioinformatics has seen significant advances with the\nuse of Molecular Dynamics (MD) simulations of biological systems. The MD\nmethodology has allowed to explain and discover molecular mechanisms in a wide\nrange of natural processes. There is an impending need to readily share the\never-increasing amount of MD data, which has been hindered by the lack of\nspecialized tools in the past. To solve this problem, we present HTMoL, a\nstate-of-the-art plug-in-free hardware-accelerated web application specially\ndesigned to efficiently transfer and visualize raw MD trajectory files on a web\nbrowser. Now, individual research labs can publish MD data on the Internet, or\nuse HTMoL to profoundly improve scientific reports by including supplemental MD\ndata in a journal publication. HTMoL can also be used as a visualization\ninterface to access MD trajectories generated on a high-performance computer\ncenter directly.\nAvailability: HTMoL is available free of charge for academic use. All major\nbrowsers are supported. A complete online documentation including instructions\nfor download, installation, configuration, and examples is available at the\nHTMoL website this http URL. Supplementary data are available\nonline. Corresponding author: [email protected]\n",
"title": "HTMoL: full-stack solution for remote access, visualization, and analysis of Molecular Dynamics trajectory data"
}
| null | null | null | null | true | null |
10822
| null |
Default
| null | null |
null |
{
"abstract": " The auction method developed by Bertsekas in the late 1970s is a relaxation\ntechnique for solving integer-valued assignment problems. It resembles a\ncompetitive bidding process, where unsatisfied persons (bidders) attempt to\nclaim the objects (lots) offering the best value. By transforming\ninteger-valued transport problems into assignment problems, the auction method\ncan be extended to compute optimal transport solutions. We propose a more\ngeneral auction method that can be applied directly to real-valued transport\nproblems. We prove termination and provide a priori error bounds for the\ngeneral auction method. Our numerical results indicate that the complexity of\nthe general auction is roughly comparable to that of the original auction\nmethod, when the latter is applicable.\n",
"title": "General auction method for real-valued optimal transport"
}
| null | null | null | null | true | null |
10823
| null |
Default
| null | null |
null |
{
"abstract": " Levitated optomechanics is showing potential for precise force measurements.\nHere, we report a case study, to show experimentally the capacity of such a\nforce sensor. Using an electric field as a tool to detect a Coulomb force\napplied onto a levitated nanosphere. We experimentally observe the spatial\ndisplacement of up to 6.6 nm of the levitated nanosphere by imposing a DC\nfield. We further apply an AC field and demonstrate resonant enhancement of\nforce sensing when a driving frequency, $\\omega_{AC}$, and the frequency of the\nlevitated mechanical oscillator, $\\omega_0$, converge. We directly measure a\nforce of $3.0 \\pm 1.5 \\times 10^{-20}$ N with 10 second integration time, at a\ncentre of mass temperature of 3 K and at a pressure of $1.6 \\times 10^{-5}$\nmbar.\n",
"title": "Force sensing with an optically levitated charged nanoparticle"
}
| null | null | null | null | true | null |
10824
| null |
Default
| null | null |
null |
{
"abstract": " The purpose of the present paper is to show that: Eilenberg-type\ncorrespondences = Birkhoff's theorem for (finite) algebras + duality. We\nconsider algebras for a monad T on a category D and we study (pseudo)varieties\nof T-algebras. Pseudovarieties of algebras are also known in the literature as\nvarieties of finite algebras. Two well-known theorems that characterize\nvarieties and pseudovarieties of algebras play an important role here:\nBirkhoff's theorem and Birkhoff's theorem for finite algebras, the latter also\nknown as Reiterman's theorem. We prove, under mild assumptions, a categorical\nversion of Birkhoff's theorem for (finite) algebras to establish a one-to-one\ncorrespondence between (pseudo)varieties of T-algebras and (pseudo)equational\nT-theories. Now, if C is a category that is dual to D and B is the comonad on C\nthat is the dual of T, we get a one-to-one correspondence between\n(pseudo)equational T-theories and their dual, (pseudo)coequational B-theories.\nParticular instances of (pseudo)coequational B-theories have been already\nstudied in language theory under the name of \"varieties of languages\" to\nestablish Eilenberg-type correspondences. All in all, we get a one-to-one\ncorrespondence between (pseudo)varieties of T-algebras and (pseudo)coequational\nB-theories, which will be shown to be exactly the nature of Eilenberg-type\ncorrespondences.\n",
"title": "Unveiling Eilenberg-type Correspondences: Birkhoff's Theorem for (finite) Algebras + Duality"
}
| null | null | null | null | true | null |
10825
| null |
Default
| null | null |
null |
{
"abstract": " CONTEXT. It is theoretically possible for rings to have formed around\nextrasolar planets in a similar way to that in which they formed around the\ngiant planets in our solar system. However, no such rings have been detected to\ndate.\nAIMS: We aim to test the possibility of detecting rings around exoplanets by\ninvestigating the photometric and spectroscopic ring signatures in\nhigh-precision transit signals.\nMETHODS: The photometric and spectroscopic transit signals of a ringed planet\nis expected to show deviations from that of a spherical planet. We used these\ndeviations to quantify the detectability of rings. We present SOAP3.0 which is\na numerical tool to simulate ringed planet transits and measure ring\ndetectability based on amplitudes of the residuals between the ringed planet\nsignal and best fit ringless model.\nRESULTS: We find that it is possible to detect the photometric and\nspectroscopic signature of near edge-on rings especially around planets with\nhigh impact parameter. Time resolution $\\leq$ 7 mins is required for the\nphotometric detection, while 15 mins is sufficient for the spectroscopic\ndetection. We also show that future instruments like CHEOPS and ESPRESSO, with\nprecisions that allow ring signatures to be well above their noise-level,\npresent good prospects for detecting rings.\n",
"title": "Detecting transit signatures of exoplanetary rings using SOAP3.0"
}
| null | null |
[
"Physics"
] | null | true | null |
10826
| null |
Validated
| null | null |
null |
{
"abstract": " Scientific legacy code in MATLAB/Octave not compatible with modernization of\nresearch workflows is vastly abundant throughout academic community.\nPerformance of non-vectorized code written in MATLAB/Octave represents a major\nburden. A new programming language for technical computing Julia, promises to\naddress these issues. Although Julia syntax is similar to MATLAB/Octave,\nporting code to Julia may be cumbersome for researchers. Here we present\nMatlabCompat.jl - a library aimed at simplifying the conversion of your\nMATLAB/Octave code to Julia. We show using a simplistic image analysis use case\nthat MATLAB/Octave code can be easily ported to high performant Julia using\nMatlabCompat.jl.\n",
"title": "MatlabCompat.jl: helping Julia understand Your Matlab/Octave Code"
}
| null | null | null | null | true | null |
10827
| null |
Default
| null | null |
null |
{
"abstract": " It is customary to conceive the interactions of all the constituents of a\nmolecular system, i.e. electrons and nuclei, as Coulombic. However, in a more\ndetailed analysis one may always find small but non-negligible non-Coulombic\ninteractions in molecular systems originating from the finite size of nuclei,\nmagnetic interactions, etc. While such small modifications of the Coulombic\ninteractions do not seem to alter the nature of a molecular system in real\nworld seriously, they are a serious obstacle for quantum chemical theories and\nmethodologies which their formalism is strictly confined to the Coulombic\ninteractions. Although the quantum theory of atoms in molecules (QTAIM) has\nbeen formulated originally for the Coulombic systems, some recent studies have\ndemonstrated that apart from basin energy of an atom in a molecule, its\ntheoretical ingredients are not sensitive to the explicit form of the potential\nenergy operator. In this study, it is demonstrated that the basin energy may be\ndefined not only for coulombic systems but for all real-space subsystems of\nthose systems that are described by any member of the set of the homogeneous\npotential energy functions. On the other hand, this extension opens the door\nfor seeking novel real-space subsystems, apart from atoms in molecules, in\nnon-Coulombic systems. These novel real-space subsystems call for an extended\nformalism that goes beyond the orthodox QTAIM, which is not confined to the\nCoulombic systems nor to the atoms in molecules as the sole real-space\nsubsystems. It is termed the quantum theory of real-space open subsystems\n(QTROS) and its potential applications are detailed. The harmonic trap model,\ncontaining non-interacting fermions or bosons, is considered as an example for\nthe QTROS analysis. The QTROS analysis of bosonic systems is particularly quite\nunprecedented, not attempted before.\n",
"title": "Extending the topological analysis and seeking the real-space subsystems in non-Coulombic systems with homogeneous potential energy functions"
}
| null | null | null | null | true | null |
10828
| null |
Default
| null | null |
null |
{
"abstract": " Future sea-level rise drives severe risks for many coastal communities.\nStrategies to manage these risks hinge on a sound characterization of the\nuncertainties. For example, recent studies suggest that large fractions of the\nAntarctic ice sheet (AIS) may rapidly disintegrate in response to rising global\ntemperatures, leading to potentially several meters of sea-level rise during\nthe next few centuries. It is deeply uncertain, for example, whether such an\nAIS disintegration will be triggered, how much this would increase sea-level\nrise, whether extreme storm surges intensify in a warming climate, or which\nemissions pathway future societies will choose. Here, we assess the impacts of\nthese deep uncertainties on projected flooding probabilities for a levee ring\nin New Orleans, Louisiana. We use 18 scenarios, presenting probabilistic\nprojections within each one, to sample key deeply uncertain future projections\nof sea-level rise, radiative forcing pathways, storm surge characterization,\nand contributions from rapid AIS mass loss. The implications of these deep\nuncertainties for projected flood risk are thus characterized by a set of 18\nprobability distribution functions. We use a global sensitivity analysis to\nassess which mechanisms contribute to uncertainty in projected flood risk over\nthe course of a 50-year design life. In line with previous work, we find that\nthe uncertain storm surge drives the most substantial risk, followed by general\nAIS dynamics, in our simple model for future flood risk for New Orleans.\n",
"title": "Deep Uncertainty Surrounding Coastal Flood Risk Projections: A Case Study for New Orleans"
}
| null | null | null | null | true | null |
10829
| null |
Default
| null | null |
null |
{
"abstract": " The approximation power of general feedforward neural networks with piecewise\nlinear activation functions is investigated. First, lower bounds on the size of\na network are established in terms of the approximation error and network depth\nand width. These bounds improve upon state-of-the-art bounds for certain\nclasses of functions, such as strongly convex functions. Second, an upper bound\nis established on the difference of two neural networks with identical weights\nbut different activation functions.\n",
"title": "Bounds on the Approximation Power of Feedforward Neural Networks"
}
| null | null |
[
"Statistics"
] | null | true | null |
10830
| null |
Validated
| null | null |
null |
{
"abstract": " We define the standard Borel space of free Araki-Woods factors and prove that\ntheir isomorphism relation is not classifiable by countable structures. We also\nprove that equality of $\\tau$-topologies, arising as invariants of type III\nfactors, as well as coycle and outer conjugacy of actions of abelian groups on\nfree product factors are not classifiable by countable structures.\n",
"title": "Non-classification of free Araki-Woods factors and $τ$-invariants"
}
| null | null | null | null | true | null |
10831
| null |
Default
| null | null |
null |
{
"abstract": " Robots such as autonomous underwater vehicles (AUVs) and autonomous surface\nvehicles (ASVs) have been used for sensing and monitoring aquatic environments\nsuch as oceans and lakes. Environmental sampling is a challenging task because\nthe environmental attributes to be observed can vary both spatially and\ntemporally, and the target environment is usually a large and continuous domain\nwhereas the sampling data is typically sparse and limited. The challenges\nrequire that the sampling method must be informative and efficient enough to\ncatch up with the environmental dynamics. In this paper we present a planning\nand learning method that enables a sampling robot to perform persistent\nmonitoring tasks by learning and refining a dynamic \"data map\" that models a\nspatiotemporal environment attribute such as ocean salinity content. Our\nenvironmental sampling framework consists of two components: to maximize the\ninformation collected, we propose an informative planning component that\nefficiently generates sampling waypoints that contain the maximal information;\nTo alleviate the computational bottleneck caused by large-scale data\naccumulated, we develop a component based on a sparse Gaussian Process whose\nhyperparameters are learned online by taking advantage of only a subset of data\nthat provides the greatest contribution. We validate our method with both\nsimulations running on real ocean data and field trials with an ASV in a lake\nenvironment. Our experiments show that the proposed framework is both accurate\nin learning the environmental data map and efficient in catching up with the\ndynamic environmental changes.\n",
"title": "Data-Driven Learning and Planning for Environmental Sampling"
}
| null | null | null | null | true | null |
10832
| null |
Default
| null | null |
null |
{
"abstract": " In this paper, we present a spectral graph wavelet approach for shape\nanalysis of carpal bones of human wrist. We apply a metric called global\nspectral graph wavelet signature for representation of cortical surface of the\ncarpal bone based on eigensystem of Laplace-Beltrami operator. Furthermore, we\npropose a heuristic and efficient way of aggregating local descriptors of a\ncarpal bone surface to global descriptor. The resultant global descriptor is\nnot only isometric invariant, but also much more efficient and requires less\nmemory storage. We perform experiments on shape of the carpal bones of ten\nwomen and ten men from a publicly-available database. Experimental results show\nthe excellency of the proposed GSGW compared to recent proposed GPS embedding\napproach for comparing shapes of the carpal bones across populations.\n",
"title": "Global spectral graph wavelet signature for surface analysis of carpal bones"
}
| null | null | null | null | true | null |
10833
| null |
Default
| null | null |
null |
{
"abstract": " This paper studies some robust regression problems associated with the\n$q$-norm loss ($q\\ge1$) and the $\\epsilon$-insensitive $q$-norm loss in the\nreproducing kernel Hilbert space. We establish a variance-expectation bound\nunder a priori noise condition on the conditional distribution, which is the\nkey technique to measure the error bound. Explicit learning rates will be given\nunder the approximation ability assumptions on the reproducing kernel Hilbert\nspace.\n",
"title": "Learning Rates of Regression with q-norm Loss and Threshold"
}
| null | null | null | null | true | null |
10834
| null |
Default
| null | null |
null |
{
"abstract": " Thin liquid films are ubiquitous in natural phenomena and technological\napplications. They have been extensively studied via deterministic hydrodynamic\nequations, but thermal fluctuations often play a crucial role that needs to be\nunderstood. An example of this is dewetting, which involves the rupture of a\nthin liquid film and the formation of droplets. Such a process is thermally\nactivated and requires fluctuations to be taken into account self-consistently.\nIn this work we present an analytical and numerical study of a stochastic\nthin-film equation derived from first principles. Following a brief review of\nthe derivation, we scrutinise the behaviour of the equation in the limit of\nperfectly correlated noise along the wall-normal direction. The stochastic\nthin-film equation is also simulated by adopting a numerical scheme based on a\nspectral collocation method. The scheme allows us to explore the fluctuating\ndynamics of the thin film and the behaviour of its free energy in the vicinity\nof rupture. Finally, we also study the effect of the noise intensity on the\nrupture time, which is in agreement with previous works.\n",
"title": "Instability, rupture and fluctuations in thin liquid films: Theory and computations"
}
| null | null |
[
"Physics"
] | null | true | null |
10835
| null |
Validated
| null | null |
null |
{
"abstract": " The long range movement of certain organisms in the presence of a\nchemoattractant can be governed by long distance runs, according to an\napproximate Levy distribution. This article clarifies the form of biologically\nrelevant model equations: We derive Patlak-Keller-Segel-like equations\ninvolving nonlocal, fractional Laplacians from a microscopic model for cell\nmovement. Starting from a power-law distribution of run times, we derive a\nkinetic equation in which the collision term takes into account the long range\nbehaviour of the individuals. A fractional chemotactic equation is obtained in\na biologically relevant regime. Apart from chemotaxis, our work has\nimplications for biological diffusion in numerous processes.\n",
"title": "Fractional Patlak-Keller-Segel equations for chemotactic superdiffusion"
}
| null | null | null | null | true | null |
10836
| null |
Default
| null | null |
null |
{
"abstract": " We study trend filtering, a relatively recent method for univariate\nnonparametric regression. For a given positive integer $r$, the $r$-th order\ntrend filtering estimator is defined as the minimizer of the sum of squared\nerrors when we constrain (or penalize) the sum of the absolute $r$-th order\ndiscrete derivatives of the fitted function at the design points. For $r=1$,\nthe estimator reduces to total variation regularization which has received much\nattention in the statistics and image processing literature. In this paper, we\nstudy the performance of the trend filtering estimator for every positive\ninteger $r$, both in the constrained and penalized forms. Our main results show\nthat in the strong sparsity setting when the underlying function is a\n(discrete) spline with few \"knots\", the risk (under the global squared error\nloss) of the trend filtering estimator (with an appropriate choice of the\ntuning parameter) achieves the parametric $n^{-1}$ rate, up to a logarithmic\n(multiplicative) factor. Our results therefore provide support for the use of\ntrend filtering, for every $r$, in the strong sparsity setting.\n",
"title": "Adaptive Risk Bounds in Univariate Total Variation Denoising and Trend Filtering"
}
| null | null | null | null | true | null |
10837
| null |
Default
| null | null |
null |
{
"abstract": " In search of a reliable methodology for the prediction of light absorption\nand emission of Ce$^{3+}$-doped luminescent materials, 13 representative\nmaterials are studied with first-principles and semiempirical approaches. In\nthe first-principles approach, that combines constrained density-functional\ntheory and $\\Delta$SCF, the atomic positions are obtained for both ground and\nexcited states of the Ce$^{3+}$ ion. The structural information is fed into\nDorenbos' semiempirical model. Absorption and emission energies are calculated\nwith both methods and compared with experiment. The first-principles approach\nmatches experiment within 0.3 eV, with two exceptions at 0.5 eV. In contrast,\nthe semiempirical approach does not perform as well (usually more than 0.5 eV\nerror). The general applicability of the present first-principles scheme, with\nan encouraging predictive power, opens a novel avenue for crystal site\nengineering and high-throughput search for new phosphors and scintillators.\n",
"title": "Assessment of First-Principles and Semiempirical Methodologies for Absorption and Emission Energies of Ce$^{3+}$-Doped Luminescent Materials"
}
| null | null |
[
"Physics"
] | null | true | null |
10838
| null |
Validated
| null | null |
null |
{
"abstract": " Adversarial training has been shown to regularize deep neural networks in\naddition to increasing their robustness to adversarial examples. However, its\nimpact on very deep state of the art networks has not been fully investigated.\nIn this paper, we present an efficient approach to perform adversarial training\nby perturbing intermediate layer activations and study the use of such\nperturbations as a regularizer during training. We use these perturbations to\ntrain very deep models such as ResNets and show improvement in performance both\non adversarial and original test data. Our experiments highlight the benefits\nof perturbing intermediate layer activations compared to perturbing only the\ninputs. The results on CIFAR-10 and CIFAR-100 datasets show the merits of the\nproposed adversarial training approach. Additional results on WideResNets show\nthat our approach provides significant improvement in classification accuracy\nfor a given base model, outperforming dropout and other base models of larger\nsize.\n",
"title": "Regularizing deep networks using efficient layerwise adversarial training"
}
| null | null | null | null | true | null |
10839
| null |
Default
| null | null |
null |
{
"abstract": " The Android OS has become the most popular mobile operating system leading to\na significant increase in the spread of Android malware. Consequently, several\nstatic and dynamic analysis systems have been developed to detect Android\nmalware. With dynamic analysis, efficient test input generation is needed in\norder to trigger the potential run-time malicious behaviours. Most existing\ndynamic analysis systems employ random-based input generation methods usually\nbuilt using the Android Monkey tool. Random-based input generation has several\nshortcomings including limited code coverage, which motivates us to explore\ncombining it with a state-based method in order to improve efficiency. Hence,\nin this paper, we present a novel hybrid test input generation approach\ndesigned to improve dynamic analysis on real devices. We implemented the hybrid\nsystem by integrating a random based tool (Monkey) with a state based tool\n(DroidBot) in order to improve code coverage and potentially uncover more\nmalicious behaviours. The system is evaluated using 2,444 Android apps\ncontaining 1222 benign and 1222 malware samples from the Android malware genome\nproject. Three scenarios, random only, state-based only, and our proposed\nhybrid approach were investigated to comparatively evaluate their performances.\nOur study shows that the hybrid approach significantly improved the amount of\ndynamic features extracted from both benign and malware samples over the\nstate-based and commonly used random test input generation method.\n",
"title": "Improving Dynamic Analysis of Android Apps Using Hybrid Test Input Generation"
}
| null | null | null | null | true | null |
10840
| null |
Default
| null | null |
null |
{
"abstract": " Process Monitoring involves tracking a system's behaviors, evaluating the\ncurrent state of the system, and discovering interesting events that require\nimmediate actions. In this paper, we consider monitoring temporal system state\nsequences to help detect the changes of dynamic systems, check the divergence\nof the system development, and evaluate the significance of the deviation. We\nbegin with discussions of data reduction, symbolic data representation, and the\nanomaly detection in temporal discrete sequences. Time-series representation\nmethods are also discussed and used in this paper to discretize raw data into\nsequences of system states. Markov Chains and stationary state distributions\nare continuously generated from temporal sequences to represent snapshots of\nthe system dynamics in different time frames. We use generalized Jensen-Shannon\nDivergence as the measure to monitor changes of the stationary symbol\nprobability distributions and evaluate the significance of system deviations.\nWe prove that the proposed approach is able to detect deviations of the systems\nwe monitor and assess the deviation significance in probabilistic manner.\n",
"title": "Process Monitoring Using Maximum Sequence Divergence"
}
| null | null | null | null | true | null |
10841
| null |
Default
| null | null |
null |
{
"abstract": " In this article we go on to discuss about various proper extensions of\nKannan's two different fixed point theorems, introducing the new concept of\n$\\sigma_c$-function; which is independent of the three notions of simulation\nfunction, manageable functions and R-functions. These results are the analogous\nto some well known theorems, and extends several known results in this\nliterature.\n",
"title": "Some generalizations of Kannan's theorems via $σ_c$-function"
}
| null | null | null | null | true | null |
10842
| null |
Default
| null | null |
null |
{
"abstract": " In 2002, Biss investigated on a kind of fibration which is called rigid\ncovering fibration (we rename it by rigid fibration) with properties similar to\ncovering spaces. In this paper, we obtain a relation between arbitrary\ntopological spaces and its rigid fibrations. Using this relation we obtain a\ncommutative diagram of homotopy groups and quasitopological homotopy groups and\ndeduce some results in this field.\n",
"title": "On Exact Sequences of the Rigid Fibrations"
}
| null | null | null | null | true | null |
10843
| null |
Default
| null | null |
null |
{
"abstract": " Analytic gradient routines are a desirable feature for quantum mechanical\nmethods, allowing for efficient determination of equilibrium and transition\nstate structures and several other molecular properties. In this work, we\npresent analytical gradients for multiconfiguration pair-density functional\ntheory (MC-PDFT) when used with a state-specific complete active space\nself-consistent field reference wave function. Our approach constructs a\nLagrangian that is variational in all wave function parameters. We find that\nMC-PDFT locates equilibrium geometries for several small- to medium-sized\norganic molecules that are similar to those located by complete active space\nsecond-order perturbation theory but that are obtained with decreased\ncomputational cost.\n",
"title": "Analytic Gradients for Complete Active Space Pair-Density Functional Theory"
}
| null | null | null | null | true | null |
10844
| null |
Default
| null | null |
null |
{
"abstract": " We report the analysis of the $10-1000$ TeV large-scale sidereal anisotropy\nof Galactic cosmic rays (GCRs) with the data collected by the Tibet Air Shower\nArray from October, 1995 to February, 2010. In this analysis, we improve the\nenergy estimate and extend the declination range down to $-30^{\\circ}$. We find\nthat the anisotropy maps above 100 TeV are distinct from that at multi-TeV\nband. The so-called \"tail-in\" and \"loss-cone\" features identified at low\nenergies get less significant and a new component appears at $\\sim100$ TeV. The\nspatial distribution of the GCR intensity with an excess (7.2$\\sigma$\npre-trial, 5.2$\\sigma$ post-trial) and a deficit ($-5.8\\sigma$ pre-trial) are\nobserved in the 300 TeV anisotropy map, in a good agreement with IceCube's\nresults at 400 TeV. Combining the Tibet results in the northern sky with\nIceCube's results in the southern sky, we establish a full-sky picture of the\nanisotropy in hundreds of TeV band. We further find that the amplitude of the\nfirst order anisotropy increases sharply above $\\sim100$ TeV, indicating a new\ncomponent of the anisotropy. All these results may shed new light on\nunderstanding the origin and propagation of GCRs.\n",
"title": "Northern sky Galactic Cosmic Ray anisotropy between 10-1000 TeV with the Tibet Air Shower Array"
}
| null | null | null | null | true | null |
10845
| null |
Default
| null | null |
null |
{
"abstract": " Counting objects in digital images is a process that should be replaced by\nmachines. This tedious task is time consuming and prone to errors due to\nfatigue of human annotators. The goal is to have a system that takes as input\nan image and returns a count of the objects inside and justification for the\nprediction in the form of object localization. We repose a problem, originally\nposed by Lempitsky and Zisserman, to instead predict a count map which contains\nredundant counts based on the receptive field of a smaller regression network.\nThe regression network predicts a count of the objects that exist inside this\nframe. By processing the image in a fully convolutional way each pixel is going\nto be accounted for some number of times, the number of windows which include\nit, which is the size of each window, (i.e., 32x32 = 1024). To recover the true\ncount we take the average over the redundant predictions. Our contribution is\nredundant counting instead of predicting a density map in order to average over\nerrors. We also propose a novel deep neural network architecture adapted from\nthe Inception family of networks called the Count-ception network. Together our\napproach results in a 20% relative improvement (2.9 to 2.3 MAE) over the state\nof the art method by Xie, Noble, and Zisserman in 2016.\n",
"title": "Count-ception: Counting by Fully Convolutional Redundant Counting"
}
| null | null | null | null | true | null |
10846
| null |
Default
| null | null |
null |
{
"abstract": " We propose a modification of the standard inverse scattering transform for\nthe focusing nonlinear Schrödinger equation (also other equations by natural\ngeneralization) formulated with nonzero boundary conditions at infinity. The\npurpose is to deal with arbitrary-order poles and potentially severe spectral\nsingularities in a simple and unified way. As an application, we use the\nmodified transform to place the Peregrine solution and related higher-order\n\"rogue wave\" solutions in an inverse-scattering context for the first time.\nThis allows one to directly study properties of these solutions such as their\ndynamical or structural stability, or their asymptotic behavior in the limit of\nhigh order. The modified transform method also allows rogue waves to be\ngenerated on top of other structures by elementary Darboux transformations,\nrather than the generalized Darboux transformations in the literature or other\nrelated limit processes.\n",
"title": "A robust inverse scattering transform for the focusing nonlinear Schrödinger equation"
}
| null | null | null | null | true | null |
10847
| null |
Default
| null | null |
null |
{
"abstract": " This paper investigates estimation of the mean vector under invariant\nquadratic loss for a spherically symmetric location family with a residual\nvector with density of the form $\nf(x,u)=\\eta^{(p+n)/2}f(\\eta\\{\\|x-\\theta\\|^2+\\|u\\|^2\\}) $, where $\\eta$ is\nunknown. We show that the natural estimator $x$ is admissible for $p=1,2$.\nAlso, for $p\\geq 3$, we find classes of generalized Bayes estimators that are\nadmissible within the class of equivariant estimators of the form\n$\\{1-\\xi(x/\\|u\\|)\\}x$. In the Gaussian case, a variant of the James--Stein\nestimator, $[1-\\{(p-2)/(n+2)\\}/\\{\\|x\\|^2/\\|u\\|^2+(p-2)/(n+2)+1\\}]x$, which\ndominates the natural estimator $x$, is also admissible within this class. We\nalso study the related regression model.\n",
"title": "Admissible Bayes equivariant estimation of location vectors for spherically symmetric distributions with unknown scale"
}
| null | null | null | null | true | null |
10848
| null |
Default
| null | null |
null |
{
"abstract": " We define a variety of abstract termination principles which form\ngeneralisations of simplification orders, and investigate their computational\ncontent. Simplification orders, which include the well-known multiset and\nlexicographic path orderings, are important techniques for proving that\ncomputer programs terminate. Moreover, an analysis of the proofs that these\norders are wellfounded can yield additional quantitative information: namely an\nupper bound on the complexity of programs reducing under these orders. In this\npaper we focus on extracting computational content from the typically\nnon-constructive wellfoundedness proofs of termination orders, with an eye\ntowards the establishment of general metatheorems which characterise bounds on\nthe derivational complexity induced by these orders. However, ultimately we\nhave a much broader goal, which is to explore a number of deep mathematical\nconcepts which underlie termination orders, including minimal-bad-sequence\nconstructions, modified realizability and bar recursion. We aim to describe how\nthese concepts all come together to form a particularly elegant illustration of\nthe bridge between proofs and programs.\n",
"title": "A proof theoretic study of abstract termination principles"
}
| null | null | null | null | true | null |
10849
| null |
Default
| null | null |
null |
{
"abstract": " The problem of the search for the satellites of the exoplanets (exomoons) is\ndiscussed recently. There are very many satellites in our Solar System. But in\ncontrary of our Solar system, exoplanets have significant eccentricity. In\nprocess of planetary migration, exoplanets can cross some resonances with\nfollowing growth of their orbital eccentricity. The stability of exomoons\ndecreases, and many of satellites were lost. Here we give a simple example of\nloss satellite when eccentricity increased. Finally, we can conclude that\nexomoons must be rare due to observed large eccentricities of exoplanets.\n",
"title": "Why exomoons must be rare?"
}
| null | null | null | null | true | null |
10850
| null |
Default
| null | null |
null |
{
"abstract": " Consider the following Kolmogorov type hypoelliptic operator $$ \\mathscr\nL_t:=\\mbox{$\\sum_{j=2}^n$}x_j\\cdot\\nabla_{x_{j-1}}+{\\rm Tr} (a_t\n\\cdot\\nabla^2_{x_n}), $$ where $n\\geq 2$, $x=(x_1,\\cdots,x_n)\\in(\\mathbb R^d)^n\n=\\mathbb R^{nd}$ and $a_t$ is a time-dependent constant symmetric $d\\times\nd$-matrix that is uniformly elliptic and bounded.. Let $\\{\\mathcal T_{s,t};\nt\\geq s\\}$ be the time-dependent semigroup associated with $\\mathscr L_t$; that\nis, $\\partial_s {\\mathcal T}_{s, t} f = - {\\mathscr L}_s {\\mathcal T}_{s, t}f$.\nFor any $p\\in(1,\\infty)$, we show that there is a constant $C=C(p,n,d)>0$ such\nthat for any $f(t, x)\\in L^p(\\mathbb R \\times \\mathbb R^{nd})=L^p(\\mathbb\nR^{1+nd})$ and every $\\lambda \\geq 0$, $$\n\\left\\|\\Delta_{x_j}^{{1}/{(1+2(n-j)})}\\int^{\\infty}_0 e^{-\\lambda t} {\\mathcal\nT}_{s, s+t }f(t+s, x)dt\\right\\|_p\\leq C\\|f\\|_p,\\quad j=1,\\cdots, n, $$ where\n$\\|\\cdot\\|_p$ is the usual $L^p$-norm in $L^p(\\mathbb R^{1+nd}; d s\\times d\nx)$. To show this type of estimates, we first study the propagation of\nregularity in $L^2$-space from variable $x_n$ to $x_1$ for the solution of the\ntransport equation $\\partial_t u+\\sum_{j=2}^nx_j\\cdot\\nabla_{x_{j-1}} u=f$.\n",
"title": "Propagation of regularity in $L^p$-spaces for Kolmogorov type hypoelliptic operators"
}
| null | null | null | null | true | null |
10851
| null |
Default
| null | null |
null |
{
"abstract": " Constraint answer set programming is a promising research direction that\nintegrates answer set programming with constraint processing. It is often\ninformally related to the field of satisfiability modulo theories. Yet, the\nexact formal link is obscured as the terminology and concepts used in these two\nresearch areas differ. In this paper, we connect these two research areas by\nuncovering the precise formal relation between them. We believe that this work\nwill booster the cross-fertilization of the theoretical foundations and the\nexisting solving methods in both areas. As a step in this direction we provide\na translation from constraint answer set programs with integer linear\nconstraints to satisfiability modulo linear integer arithmetic that paves the\nway to utilizing modern satisfiability modulo theories solvers for computing\nanswer sets of constraint answer set programs.\n",
"title": "On Relation between Constraint Answer Set Programming and Satisfiability Modulo Theories"
}
| null | null | null | null | true | null |
10852
| null |
Default
| null | null |
null |
{
"abstract": " Modern radio telescopes, such as the Square Kilometre Array (SKA), will probe\nthe radio sky over large fields-of-view, which results in large w-modulations\nof the sky image. This effect complicates the relationship between the measured\nvisibilities and the image under scrutiny. In algorithmic terms, it gives rise\nto massive memory and computational time requirements. Yet, it can be a\nblessing in terms of reconstruction quality of the sky image. In recent years,\nseveral works have shown that large w-modulations promote the spread spectrum\neffect. Within the compressive sensing framework, this effect increases the\nincoherence between the sensing basis and the sparsity basis of the signal to\nbe recovered, leading to better estimation of the sky image. In this article,\nwe revisit the w-projection approach using convex optimisation in realistic\nsettings, where the measurement operator couples the w-terms in Fourier and the\nde-gridding kernels. We provide sparse, thus fast, models of the Fourier part\nof the measurement operator through adaptive sparsification procedures.\nConsequently, memory requirements and computational cost are significantly\nalleviated, at the expense of introducing errors on the radio-interferometric\ndata model. We present a first investigation of the impact of the sparse\nvariants of the measurement operator on the image reconstruction quality. We\nfinally analyse the interesting super-resolution potential associated with the\nspread spectrum effect of the w-modulation, and showcase it through\nsimulations. Our C++ code is available online on GitHub.\n",
"title": "The w-effect in interferometric imaging: from a fast sparse measurement operator to super-resolution"
}
| null | null |
[
"Physics"
] | null | true | null |
10853
| null |
Validated
| null | null |
null |
{
"abstract": " Graphs are a prevalent tool in data science, as they model the inherent\nstructure of the data. They have been used successfully in unsupervised and\nsemi-supervised learning. Typically they are constructed either by connecting\nnearest samples, or by learning them from data, solving an optimization\nproblem. While graph learning does achieve a better quality, it also comes with\na higher computational cost. In particular, the current state-of-the-art model\ncost is $\\mathcal{O}(n^2)$ for $n$ samples. In this paper, we show how to scale\nit, obtaining an approximation with leading cost of $\\mathcal{O}(n\\log(n))$,\nwith quality that approaches the exact graph learning model. Our algorithm uses\nknown approximate nearest neighbor techniques to reduce the number of\nvariables, and automatically selects the correct parameters of the model,\nrequiring a single intuitive input: the desired edge density.\n",
"title": "Large Scale Graph Learning from Smooth Signals"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
10854
| null |
Validated
| null | null |
null |
{
"abstract": " Power plant is a complex and nonstationary system for which the traditional\nmachine learning modeling approaches fall short of expectations. The\nensemble-based online learning methods provide an effective way to continuously\nlearn from the dynamic environment and autonomously update models to respond to\nenvironmental changes. This paper proposes such an online ensemble regression\napproach to model power plant performance, which is critically important for\noperation optimization. The experimental results on both simulated and real\ndata show that the proposed method can achieve performance with less than 1%\nmean average percentage error, which meets the general expectations in field\noperations.\n",
"title": "Power Plant Performance Modeling with Concept Drift"
}
| null | null | null | null | true | null |
10855
| null |
Default
| null | null |
null |
{
"abstract": " Studies of affect labeling, i.e. putting your feelings into words, indicate\nthat it can attenuate positive and negative emotions. Here we track the\nevolution of individual emotions for tens of thousands of Twitter users by\nanalyzing the emotional content of their tweets before and after they\nexplicitly report having a strong emotion. Our results reveal how emotions and\ntheir expression evolve at the temporal resolution of one minute. While the\nexpression of positive emotions is preceded by a short but steep increase in\npositive valence and followed by short decay to normal levels, negative\nemotions build up more slowly, followed by a sharp reversal to previous levels,\nmatching earlier findings of the attenuating effects of affect labeling. We\nestimate that positive and negative emotions last approximately 1.25 and 1.5\nhours from onset to evanescence. A separate analysis for male and female\nsubjects is suggestive of possible gender-specific differences in emotional\ndynamics.\n",
"title": "Does putting your emotions into words make you feel better? Measuring the minute-scale dynamics of emotions from online data"
}
| null | null | null | null | true | null |
10856
| null |
Default
| null | null |
null |
{
"abstract": " This paper positively solves an open problem if it is possible to provide a\nHilbert system to Epistemic Logic of Friendship (EFL) by Seligman, Girard and\nLiu. To find a Hilbert system, we first introduce a sound, complete and\ncut-free tree (or nested) sequent calculus for EFL, which is an integrated\ncombination of Seligman's sequent calculus for basic hybrid logic and a tree\nsequent calculus for modal logic. Then we translate a tree sequent into an\nordinary formula to specify a Hilbert system of EFL and finally show that our\nHilbert system is sound and complete for the intended two-dimensional\nsemantics.\n",
"title": "Axiomatizing Epistemic Logic of Friendship via Tree Sequent Calculus"
}
| null | null | null | null | true | null |
10857
| null |
Default
| null | null |
null |
{
"abstract": " Lindel{ö}f's hypothesis, one of the most important open problems in the\nhistory of mathematics, states that for large $t$, Riemann's zeta function\n$\\zeta(\\frac{1}{2}+it)$ is of order $O(t^{\\varepsilon})$ for any\n$\\varepsilon>0$. It is well known that for large $t$, the leading order\nasymptotics of the Riemann zeta function can be expressed in terms of a\ntranscendental exponential sum. The usual approach to the Lindelöf hypothesis\ninvolves the use of ingenious techniques for the estimation of this sum.\nHowever, since such estimates can not yield an asymptotic formula for the above\nsum, it appears that this approach cannot lead to the proof of the Lindelöf\nhypothesis. Here, a completely different approach is introduced: the Riemann\nzeta function is embedded in a classical problem in the theory of complex\nanalysis known as a Riemann-Hilbert problem, and then, the large\n$t$-asymptotics of the associated integral equation is formally computed. This\nyields two different results. First, the formal proof that a certain Riemann\nzeta-type double exponential sum satisfies the asymptotic estimate of the\nLindelöf hypothesis. Second, it is formally shown that the sum of\n$|\\zeta(1/2+it)|^2$ and of a certain sum which depends on $\\epsilon$, satisfies\nfor large $t$ the estimate of the Lindelöf hypothesis. Hence, since the above\nidentity is valid for all $\\epsilon$, this asymptotic identity suggests the\nvalidity of Lindelöf's hypothesis. The completion of the rigorous derivation\nof the above results will be presented in a companion paper.\n",
"title": "A novel approach to the Lindelöf hypothesis"
}
| null | null | null | null | true | null |
10858
| null |
Default
| null | null |
null |
{
"abstract": " Complex systems in a wide variety of areas such as biological modeling, image\nprocessing, and language recognition can be modeled using networks of very\nsimple machines called finite automata. Connecting subsystems modeled using\nfinite automata into a network allows for more computational power. One such\nnetwork, called a cellular automaton, consists of an n-dimensional array for n\n> 1 with a single finite automaton located at each point of the array. One of\nthe oldest problems associated with cellular automata is the firing\nsynchronization problem, originally proposed by John Myhill in 1957. As with\nany long-standing problem, there are a large number of solutions to the firing\nsynchronization problem. Our goal, and the contribution of this work, is to\nsummarize recent solutions to the problem. We focus primarily on solutions to\nthe original problem, that is, the problem where the network is a\none-dimensional array and there is a single initiator located at one of the\nends. We summarize both minimal-time and non-minimal-time solutions, with an\nemphasis on solutions that were published after 1998. We also focus on\nsolutions that minimize the number of states required by the finite automata.\nIn the process we also identify open problems that remain in terms of finding\nminimal-state solutions to the firing synchronization problem.\n",
"title": "An Overview of Recent Solutions to and Lower Bounds for the Firing Synchronization Problem"
}
| null | null | null | null | true | null |
10859
| null |
Default
| null | null |
null |
{
"abstract": " The practical success of Boolean Satisfiability (SAT) solvers stems from the\nCDCL (Conflict-Driven Clause Learning) approach to SAT solving. However, from a\npropositional proof complexity perspective, CDCL is no more powerful than the\nresolution proof system, for which many hard examples exist. This paper\nproposes a new problem transformation, which enables reducing the decision\nproblem for formulas in conjunctive normal form (CNF) to the problem of solving\nmaximum satisfiability over Horn formulas. Given the new transformation, the\npaper proves a polynomial bound on the number of MaxSAT resolution steps for\npigeonhole formulas. This result is in clear contrast with earlier results on\nthe length of proofs of MaxSAT resolution for pigeonhole formulas. The paper\nalso establishes the same polynomial bound in the case of modern core-guided\nMaxSAT solvers. Experimental results, obtained on CNF formulas known to be hard\nfor CDCL SAT solvers, show that these can be efficiently solved with modern\nMaxSAT solvers.\n",
"title": "On Tackling the Limits of Resolution in SAT Solving"
}
| null | null | null | null | true | null |
10860
| null |
Default
| null | null |
null |
{
"abstract": " Additive regression provides an extension of linear regression by modeling\nthe signal of a response as a sum of functions of covariates of relatively low\ncomplexity. We study penalized estimation in high-dimensional nonparametric\nadditive regression where functional semi-norms are used to induce smoothness\nof component functions and the empirical $L_2$ norm is used to induce sparsity.\nThe functional semi-norms can be of Sobolev or bounded variation types and are\nallowed to be different amongst individual component functions. We establish\nnew oracle inequalities for the predictive performance of such methods under\nthree simple technical conditions: a sub-gaussian condition on the noise, a\ncompatibility condition on the design and the functional classes under\nconsideration, and an entropy condition on the functional classes. For random\ndesigns, the sample compatibility condition can be replaced by its population\nversion under an additional condition to ensure suitable convergence of\nempirical norms. In homogeneous settings where the complexities of the\ncomponent functions are of the same order, our results provide a spectrum of\nexplicit convergence rates, from the so-called slow rate without requiring the\ncompatibility condition to the fast rate under the hard sparsity or certain\n$L_q$ sparsity to allow many small components in the true regression function.\nThese results significantly broadens and sharpens existing ones in the\nliterature.\n",
"title": "Penalized Estimation in Additive Regression with High-Dimensional Data"
}
| null | null | null | null | true | null |
10861
| null |
Default
| null | null |
null |
{
"abstract": " Underwater machine vision has attracted significant attention, but its low\nquality has prevented it from a wide range of applications. Although many\ndifferent algorithms have been developed to solve this problem, real-time\nadaptive methods are frequently deficient. In this paper, based on filtering\nand the use of generative adversarial networks (GANs), two approaches are\nproposed for the aforementioned issue, i.e., a filtering-based restoration\nscheme (FRS) and a GAN-based restoration scheme (GAN-RS). Distinct from\nprevious methods, FRS restores underwater images in the Fourier domain, which\nis composed of a parameter search, filtering, and enhancement. Aiming to\nfurther improve the image quality, GAN-RS can adaptively restore underwater\nmachine vision in real time without the need for pretreatment. In particular,\ninformation in the Lab color space and the dark channel is developed as loss\nfunctions, namely, underwater index loss and dark channel prior loss,\nrespectively. More specifically, learning from the underwater index, the\ndiscriminator is equipped with a carefully crafted underwater branch to predict\nthe underwater probability of an image. A multi-stage loss strategy is then\ndeveloped to guarantee the effective training of GANs. Through extensive\ncomparisons on the image quality and applications, the superiority of the\nproposed approaches is confirmed. Consequently, the GAN-RS is considerably\nfaster and achieves a state-of-the-art performance in terms of the color\ncorrection, contrast stretch, dehazing, and feature restoration of various\nunderwater scenes. The source code will be made available.\n",
"title": "Towards Quality Advancement of Underwater Machine Vision with Generative Adversarial Networks"
}
| null | null |
[
"Computer Science"
] | null | true | null |
10862
| null |
Validated
| null | null |
null |
{
"abstract": " Recent studies have shown that sketches and diagrams play an important role\nin the daily work of software developers. If these visual artifacts are\narchived, they are often detached from the source code they document, because\nthere is no adequate tool support to assist developers in capturing, archiving,\nand retrieving sketches related to certain source code artifacts. This paper\npresents SketchLink, a tool that aims at increasing the value of sketches and\ndiagrams created during software development by supporting developers in these\ntasks. Our prototype implementation provides a web application that employs the\ncamera of smartphones and tablets to capture analog sketches, but can also be\nused on desktop computers to upload, for instance, computer-generated diagrams.\nWe also implemented a plugin for a Java IDE that embeds the links in Javadoc\ncomments and visualizes them in situ in the source code editor as graphical\nicons.\n",
"title": "Linking Sketches and Diagrams to Source Code Artifacts"
}
| null | null | null | null | true | null |
10863
| null |
Default
| null | null |
null |
{
"abstract": " Recently, two influential PNAS papers have shown how our preferences for\n'Hello Kitty' and 'Harley Davidson', obtained through Facebook likes, can\naccurately predict details about our personality, religiosity, political\nattitude and sexual orientation (Konsinski et al. 2013; Youyou et al 2015). In\nthis paper, we make the claim that though the wide variety of Facebook likes\nmight predict such personal traits, even more accurate and generalizable\nresults can be reached through applying a contexts-specific, parsimonious data\nstrategy. We built this claim by predicting present day voter intention based\nsolely on likes directed toward posts from political actors. Combining the\nonline and offline, we join a subsample of surveyed respondents to their public\nFacebook activity and apply machine learning classifiers to explore the link\nbetween their political liking behaviour and actual voting intention. Through\nthis work, we show how even a single well-chosen Facebook like, can reveal as\nmuch about our political voter intention as hundreds of random likes. Further,\nby including the entire political like history of the respondents, our model\nreaches prediction accuracies above previous multiparty studies (60-70%). We\nconclude the paper by discussing how a parsimonious data strategy applied, with\nsome limitations, allow us to generalize our findings to the 1,4 million Danes\nwith at least one political like and even to other political multiparty\nsystems.\n",
"title": "Parsimonious Data: How a single Facebook like predicts voting behaviour in multiparty systems"
}
| null | null |
[
"Computer Science"
] | null | true | null |
10864
| null |
Validated
| null | null |
null |
{
"abstract": " While machine learning is going through an era of celebrated success,\nconcerns have been raised about the vulnerability of its backbone: stochastic\ngradient descent (SGD). Recent approaches have been proposed to ensure the\nrobustness of distributed SGD against adversarial (Byzantine) workers sending\npoisoned gradients during the training phase. Some of these approaches have\nbeen proven Byzantine-resilient: they ensure the convergence of SGD despite the\npresence of a minority of adversarial workers.\nWe show in this paper that convergence is not enough. In high dimension $d\n\\gg 1$, an adver\\-sary can build on the loss function's non-convexity to make\nSGD converge to ineffective models. More precisely, we bring to light that\nexisting Byzantine-resilient schemes leave a margin of poisoning of\n$\\Omega\\left(f(d)\\right)$, where $f(d)$ increases at least like $\\sqrt{d~}$.\nBased on this leeway, we build a simple attack, and experimentally show its\nstrong to utmost effectivity on CIFAR-10 and MNIST.\nWe introduce Bulyan, and prove it significantly reduces the attackers leeway\nto a narrow $O( \\frac{1}{\\sqrt{d~}})$ bound. We empirically show that Bulyan\ndoes not suffer the fragility of existing aggregation rules and, at a\nreasonable cost in terms of required batch size, achieves convergence as if\nonly non-Byzantine gradients had been used to update the model.\n",
"title": "The Hidden Vulnerability of Distributed Learning in Byzantium"
}
| null | null | null | null | true | null |
10865
| null |
Default
| null | null |
null |
{
"abstract": " The development of positioning technologies has resulted in an increasing\namount of mobility data being available. While bringing a lot of convenience to\npeople's life, such availability also raises serious concerns about privacy. In\nthis paper, we concentrate on one of the most sensitive information that can be\ninferred from mobility data, namely social relationships. We propose a novel\nsocial relation inference attack that relies on an advanced feature learning\ntechnique to automatically summarize users' mobility features. Compared to\nexisting approaches, our attack is able to predict any two individuals' social\nrelation, and it does not require the adversary to have any prior knowledge on\nexisting social relations. These advantages significantly increase the\napplicability of our attack and the scope of the privacy assessment. Extensive\nexperiments conducted on a large dataset demonstrate that our inference attack\nis effective, and achieves between 13% to 20% improvement over the best\nstate-of-the-art scheme. We propose three defense mechanisms -- hiding,\nreplacement and generalization -- and evaluate their effectiveness for\nmitigating the social link privacy risks stemming from mobility data sharing.\nOur experimental results show that both hiding and replacement mechanisms\noutperform generalization. Moreover, hiding and replacement achieve a\ncomparable trade-off between utility and privacy, the former preserving better\nutility and the latter providing better privacy.\n",
"title": "walk2friends: Inferring Social Links from Mobility Profiles"
}
| null | null | null | null | true | null |
10866
| null |
Default
| null | null |
null |
{
"abstract": " We consider a basic problem at the interface of two fundamental fields:\nsubmodular optimization and online learning. In the online unconstrained\nsubmodular maximization (online USM) problem, there is a universe\n$[n]=\\{1,2,...,n\\}$ and a sequence of $T$ nonnegative (not necessarily\nmonotone) submodular functions arrive over time. The goal is to design a\ncomputationally efficient online algorithm, which chooses a subset of $[n]$ at\neach time step as a function only of the past, such that the accumulated value\nof the chosen subsets is as close as possible to the maximum total value of a\nfixed subset in hindsight. Our main result is a polynomial-time no-$1/2$-regret\nalgorithm for this problem, meaning that for every sequence of nonnegative\nsubmodular functions, the algorithm's expected total value is at least $1/2$\ntimes that of the best subset in hindsight, up to an error term sublinear in\n$T$. The factor of $1/2$ cannot be improved upon by any polynomial-time online\nalgorithm when the submodular functions are presented as value oracles.\nPrevious work on the offline problem implies that picking a subset uniformly at\nrandom in each time step achieves zero $1/4$-regret.\nA byproduct of our techniques is an explicit subroutine for the two-experts\nproblem that has an unusually strong regret guarantee: the total value of its\nchoices is comparable to twice the total value of either expert on rounds it\ndid not pick that expert. This subroutine may be of independent interest.\n",
"title": "An Optimal Algorithm for Online Unconstrained Submodular Maximization"
}
| null | null | null | null | true | null |
10867
| null |
Default
| null | null |
null |
{
"abstract": " Grasping is a complex process involving knowledge of the object, the\nsurroundings, and of oneself. While humans are able to integrate and process\nall of the sensory information required for performing this task, equipping\nmachines with this capability is an extremely challenging endeavor. In this\npaper, we investigate how deep learning techniques can allow us to translate\nhigh-level concepts such as motor imagery to the problem of robotic grasp\nsynthesis. We explore a paradigm based on generative models for learning\nintegrated object-action representations, and demonstrate its capacity for\ncapturing and generating multimodal, multi-finger grasp configurations on a\nsimulated grasping dataset.\n",
"title": "Modeling Grasp Motor Imagery through Deep Conditional Generative Models"
}
| null | null | null | null | true | null |
10868
| null |
Default
| null | null |
null |
{
"abstract": " State-of-the-art speaker diarization systems utilize knowledge from external\ndata, in the form of a pre-trained distance metric, to effectively determine\nrelative speaker identities to unseen data. However, much of recent focus has\nbeen on choosing the appropriate feature extractor, ranging from pre-trained\n$i-$vectors to representations learned via different sequence modeling\narchitectures (e.g. 1D-CNNs, LSTMs, attention models), while adopting\noff-the-shelf metric learning solutions. In this paper, we argue that,\nregardless of the feature extractor, it is crucial to carefully design a metric\nlearning pipeline, namely the loss function, the sampling strategy and the\ndiscrimnative margin parameter, for building robust diarization systems.\nFurthermore, we propose to adopt a fine-grained validation process to obtain a\ncomprehensive evaluation of the generalization power of metric learning\npipelines. To this end, we measure diarization performance across different\nlanguage speakers, and variations in the number of speakers in a recording.\nUsing empirical studies, we provide interesting insights into the effectiveness\nof different design choices and make recommendations.\n",
"title": "Designing an Effective Metric Learning Pipeline for Speaker Diarization"
}
| null | null |
[
"Computer Science"
] | null | true | null |
10869
| null |
Validated
| null | null |
null |
{
"abstract": " Social networks are typical attributed networks with node attributes.\nDifferent from traditional attribute community detection problem aiming at\nobtaining the whole set of communities in the network, we study an\napplication-oriented problem of mining an application-aware community\norganization with respect to specific concerned attributes. The concerned\nattributes are designated based on the requirements of any application by a\nuser in advance. The application-aware community organization w.r.t. concerned\nattributes consists of the communities with feature subspaces containing these\nconcerned attributes. Besides concerned attributes, feature subspace of each\nrequired community may contain some other relevant attributes. All relevant\nattributes of a feature subspace jointly describe and determine the community\nembedded in such subspace. Thus the problem includes two subproblems, i.e., how\nto expand the set of concerned attributes to complete feature subspaces and how\nto mine the communities embedded in the expanded subspaces. Two subproblems are\njointly solved by optimizing a quality function called subspace fitness. An\nalgorithm called ACM is proposed. In order to locate the communities\npotentially belonging to the application-aware community organization, cohesive\nparts of a network backbone composed of nodes with similar concerned attributes\nare detected and set as the community seeds. The set of concerned attributes is\nset as the initial subspace for all community seeds. Then each community seed\nand its attribute subspace are adjusted iteratively to optimize the subspace\nfitness. Extensive experiments on synthetic datasets demonstrate the\neffectiveness and efficiency of our method and applications on real-world\nnetworks show its application values.\n",
"title": "Mining Application-aware Community Organization with Expanded Feature Subspaces from Concerned Attributes in Social Networks"
}
| null | null |
[
"Computer Science",
"Physics"
] | null | true | null |
10870
| null |
Validated
| null | null |
null |
{
"abstract": " $^{13}$C nuclear magnetic resonance measurements were performed for a\nsingle-component molecular material Zn(tmdt)$_{2}$, in which tmdt's form an\narrangement similar to the so-called ${\\kappa}$-type molecular packing in\nquasi-two-dimensional Mott insulators and superconductors. Detailed analysis of\nthe powder spectra uncovered local spin susceptibility in the tmdt ${\\pi}$\norbitals. The obtained shift and relaxation rate revealed the singlet-triplet\nexcitations of the ${\\pi}$ spins, indicating that Zn(tmdt)$_{2}$ is a\nspin-gapped Mott insulator with exceptionally large electron correlations\ncompared to conventional molecular Mott systems.\n",
"title": "A spin-gapped Mott insulator with the dimeric arrangement of twisted molecules Zn(tmdt)$_{2}$"
}
| null | null | null | null | true | null |
10871
| null |
Default
| null | null |
null |
{
"abstract": " Herbertsmithite and Zn-doped barlowite are two compounds for experimental\nrealization of twodimensional gapped kagome spin liquid. Theoretically, it has\nbeen proposed that charge doping a quantum spin liquid gives rise to exotic\nmetallic states, such as high-temperature superconductivity. However, one\nrecent experiment about herbertsmithite with successful Li-doping shows\nsurprisingly the insulating state even under the heavy doped scenario, which\ncan hardly be explained by many-body physics. Using first-principles\ncalculation, we performed a comprehensive study about the Li intercalated\ndoping effect of these two compounds. For the Li-doped herbertsmithite, we\nidentified the optimized Li position at the Cl-(OH)$_3$-Cl pentahedron site\ninstead of previously speculated Cl-(OH)$_3$ tetrahedral site. With the\nincrease of Li doping concentration, the saturation magnetization decreases\nlinearly due to the charge transfer from Li to Cu ions. Moreover, we found that\nLi forms chemical bonds with the nearby (OH)$^-$ and Cl$^-$ ions, which lowers\nthe surrounding chemical potential and traps the electron, as evidenced by the\nlocalized charge distribution, explaining the insulating behavior measured\nexperimentally. Though with different structure from herbertsmithite, Zn-doped\nBarlowite shows the same features upon Li doping. We conclude that Li doping\nthis family of kagome spin liquid cannot realize exotic metallic states, other\nmethods should be further explored, such as element substitution with different\nvalence electrons.\n",
"title": "Li doping kagome spin liquid compounds"
}
| null | null | null | null | true | null |
10872
| null |
Default
| null | null |
null |
{
"abstract": " Non-recurring traffic congestion is caused by temporary disruptions, such as\naccidents, sports games, adverse weather, etc. We use data related to real-time\ntraffic speed, jam factors (a traffic congestion indicator), and events\ncollected over a year from Nashville, TN to train a multi-layered deep neural\nnetwork. The traffic dataset contains over 900 million data records. The\nnetwork is thereafter used to classify the real-time data and identify\nanomalous operations. Compared with traditional approaches of using statistical\nor machine learning techniques, our model reaches an accuracy of 98.73 percent\nwhen identifying traffic congestion caused by football games. Our approach\nfirst encodes the traffic across a region as a scaled image. After that the\nimage data from different timestamps is fused with event- and time-related\ndata. Then a crossover operator is used as a data augmentation method to\ngenerate training datasets with more balanced classes. Finally, we use the\nreceiver operating characteristic (ROC) analysis to tune the sensitivity of the\nclassifier. We present the analysis of the training time and the inference time\nseparately.\n",
"title": "DxNAT - Deep Neural Networks for Explaining Non-Recurring Traffic Congestion"
}
| null | null |
[
"Statistics"
] | null | true | null |
10873
| null |
Validated
| null | null |
null |
{
"abstract": " We derive a closed form description of the convex hull of mixed-integer\nbilinear covering set with bounds on the integer variables. This convex hull\ndescription is completely determined by considering some orthogonal disjunctive\nsets defined in a certain way. Our description does not introduce any new\nvariables. We also derive a linear time separation algorithm for finding the\nfacet defining inequalities of this convex hull. We show the effectiveness of\nthe new inequalities using some examples.\n",
"title": "Facets of a mixed-integer bilinear covering set with bounds on variables"
}
| null | null | null | null | true | null |
10874
| null |
Default
| null | null |
null |
{
"abstract": " V. Nestoridis conjectured that if $\\Omega$ is a simply connected subset of\n$\\mathbb{C}$ that does not contain $0$ and $S(\\Omega)$ is the set of all\nfunctions $f\\in \\mathcal{H}(\\Omega)$ with the property that the set\n$\\left\\{T_N(f)(z)\\coloneqq\\sum_{n=0}^N\\dfrac{f^{(n)}(z)}{n!} (-z)^n : N =\n0,1,2,\\dots \\right\\}$ is dense in $\\mathcal{H}(\\Omega)$, then $S(\\Omega)$ is a\ndense $G_\\delta$ set in $\\mathcal{H}(\\Omega)$. We answer the conjecture in the\naffirmative in the special case where $\\Omega$ is an open disc $D(z_0,r)$ that\ndoes not contain $0$.\n",
"title": "Universal partial sums of Taylor series as functions of the centre of expansion"
}
| null | null | null | null | true | null |
10875
| null |
Default
| null | null |
null |
{
"abstract": " Granular gases as dilute ensembles of particles in random motion are not only\nat the basis of elementary structure-forming processes in the universe and\ninvolved in many industrial and natural phenomena, but also excellent models to\nstudy fundamental statistical dynamics. A vast number of theoretical and\nnumerical investigations have dealt with this apparently simple non-equilibrium\nsystem. The essential difference to molecular gases is the energy dissipation\nin particle collisions, a subtle distinction with immense impact on their\nglobal dynamics. Its most striking manifestation is the so-called granular\ncooling, the gradual loss of mechanical energy in absence of external\nexcitation.\nWe report an experimental study of homogeneous cooling of three-dimensional\n(3D) granular gases in microgravity. Surprisingly, the asymptotic scaling\n$E(t)\\propto t^{-2}$ obtained by Haff's minimal model [J. Fluid Mech. 134, 401\n(1983)] proves to be robust, despite the violation of several of its central\nassumptions. The shape anisotropy of the grains influences the characteristic\ntime of energy loss quantitatively, but not qualitatively. We compare kinetic\nenergies in the individual degrees of freedom, and find a slight predominance\nof the translational motions. In addition, we detect a certain preference of\nthe grains to align with their long axis in flight direction, a feature known\nfrom active matter or animal flocks, and the onset of clustering.\n",
"title": "Free Cooling of a Granular Gas in Three Dimensions"
}
| null | null |
[
"Physics"
] | null | true | null |
10876
| null |
Validated
| null | null |
null |
{
"abstract": " The SCUBA-2 Ambitious Sky Survey (SASSy) is composed of shallow 850-$\\umu$m\nimaging using the Sub-millimetre Common-User Bolometer Array 2 (SCUBA-2) on the\nJames Clerk Maxwell Telescope. Here we describe the extraction of a catalogue\nof beam-sized sources from a roughly $120\\,{\\rm deg}^2$ region of the Galactic\nplane mapped uniformly (to an rms level of about 40\\,mJy), covering longitude\n120\\degr\\,$<$\\,\\textit{l}\\,$<$\\,140\\degr\\ and latitude\n$\\abs{\\textit{b}}$\\,$<$\\,2.9\\degr. We used a matched-filtering approach to\nincrease the signal-to-noise (S/N) ratio in these noisy maps and tested the\nefficiency of our extraction procedure through estimates of the false discovery\nrate, as well as by adding artificial sources to the real images. The primary\ncatalogue contains a total of 189 sources at 850\\,$\\umu$m, down to a S/N\nthreshold of approximately 4.6. Additionally, we list 136 sources detected down\nto ${\\rm S/N}=4.3$, but recognise that as we go lower in S/N, the reliability\nof the catalogue rapidly diminishes. We perform follow-up observations of some\nof our lower significance sources through small targeted SCUBA-2 images, and\nlist 265 sources detected in these maps down to ${\\rm S/N}=5$. This illustrates\nthe real power of SASSy: inspecting the shallow maps for regions of 850-$\\umu$m\nemission and then using deeper targeted images to efficiently find fainter\nsources. We also perform a comparison of the SASSy sources with the Planck\nCatalogue of Compact Sources and the \\textit{IRAS} Point Source Catalogue, to\ndetermine which sources discovered in this field might be new, and hence\npotentially cold regions at an early stage of star formation.\n",
"title": "The SCUBA-2 Ambitious Sky Survey: a catalogue of beam-sized sources in the Galactic longitude range 120 to 140"
}
| null | null | null | null | true | null |
10877
| null |
Default
| null | null |
null |
{
"abstract": " In this paper, we propose a novel unfitted finite element method for the\nsimulation of multiple body contact. The computational mesh is generated\nindependently of the geometry of the interacting solids, which can be\narbitrarily complex. The key novelty of the approach is the combination of\nelements of the CutFEM technology, namely the enrichment of the solution field\nvia the definition of overlapping fictitious domains with a dedicated\npenalty-type regularisation of discrete operators, and the LaTIn hybrid-mixed\nformulation of complex interface conditions. Furthermore, the novel P1-P1\ndiscretisation scheme that we propose for the unfitted LaTIn solver is shown to\nbe stable, robust and optimally convergent with mesh refinement. Finally, the\npaper introduces a high-performance 3D level-set/CutFEM framework for the\nversatile and robust solution of contact problems involving multiple bodies of\ncomplex geometries, with more than two bodies interacting at a single point.\n",
"title": "A stable and optimally convergent LaTIn-Cut Finite Element Method for multiple unilateral contact problems"
}
| null | null | null | null | true | null |
10878
| null |
Default
| null | null |
null |
{
"abstract": " We employ a recently developed computational many-body technique to study for\nthe first time the half-filled Anderson-Hubbard model at finite temperature and\narbitrary correlation ($U$) and disorder ($V$) strengths. Interestingly, the\nnarrow zero temperature metallic range induced by disorder from the Mott\ninsulator expands with increasing temperature in a manner resembling a quantum\ncritical point. Our study of the resistivity temperature scaling $T^{\\alpha}$\nfor this metal reveals non Fermi liquid characteristics. Moreover, a continuous\ndependence of $\\alpha$ on $U$ and $V$ from linear to nearly quadratic was\nobserved. We argue that these exotic results arise from a systematic change\nwith $U$ and $V$ of the \"effective\" disorder, a combination of quenched\ndisorder and intrinsic localized spins.\n",
"title": "Non Fermi liquid behavior and continuously tunable resistivity exponents in the Anderson-Hubbard model at finite temperature"
}
| null | null |
[
"Physics"
] | null | true | null |
10879
| null |
Validated
| null | null |
null |
{
"abstract": " We study fairness in collaborative-filtering recommender systems, which are\nsensitive to discrimination that exists in historical data. Biased data can\nlead collaborative-filtering methods to make unfair predictions for users from\nminority groups. We identify the insufficiency of existing fairness metrics and\npropose four new metrics that address different forms of unfairness. These\nfairness metrics can be optimized by adding fairness terms to the learning\nobjective. Experiments on synthetic and real data show that our new metrics can\nbetter measure fairness than the baseline, and that the fairness objectives\neffectively help reduce unfairness.\n",
"title": "Beyond Parity: Fairness Objectives for Collaborative Filtering"
}
| null | null | null | null | true | null |
10880
| null |
Default
| null | null |
null |
{
"abstract": " We consider the setup of nonparametric 'blind regression' for estimating the\nentries of a large $m \\times n$ matrix, when provided with a small, random\nfraction of noisy measurements. We assume that all rows $u \\in [m]$ and columns\n$i \\in [n]$ of the matrix are associated to latent features $x_1(u)$ and\n$x_2(i)$ respectively, and the $(u,i)$-th entry of the matrix, $A(u, i)$ is\nequal to $f(x_1(u), x_2(i))$ for a latent function $f$. Given noisy\nobservations of a small, random subset of the matrix entries, our goal is to\nestimate the unobserved entries of the matrix as well as to \"de-noise\" the\nobserved entries.\nAs the main result of this work, we introduce a neighbor-based estimation\nalgorithm inspired by the classical Taylor's series expansion. We establish its\nconsistency when the underlying latent function $f$ is Lipschitz, the latent\nfeatures belong to a compact domain, and the fraction of observed entries in\nthe matrix is at least $\\max \\left(m^{-1 + \\delta}, n^{-1/2 + \\delta} \\right)$,\nfor any $\\delta > 0$. As an important byproduct, our analysis sheds light into\nthe performance of the classical collaborative filtering (CF) algorithm for\nmatrix completion, which has been widely utilized in practice. Experiments with\nthe MovieLens and Netflix datasets suggest that our algorithm provides a\nprincipled improvement over basic CF and is competitive with matrix\nfactorization methods.\nOur algorithm has a natural extension to tensor completion. For a $t$-order\nbalanced tensor with total of $N$ entries, we prove that our approach provides\na consistent estimator when at least $N^{-\\frac{\\lfloor 2t/3 \\rfloor}{2t}+\n\\delta}$ fraction of entries are observed, for any $\\delta > 0$. When applied\nto the setting of image in-painting (a tensor of order 3), we find that our\napproach is competitive with respect to state-of-art tensor completion\nalgorithms across benchmark images.\n",
"title": "Blind Regression via Nearest Neighbors under Latent Variable Models"
}
| null | null | null | null | true | null |
10881
| null |
Default
| null | null |
null |
{
"abstract": " A general formulation of optimization problems in which various candidate\nsolutions may use different feature-sets is presented, encompassing supervised\nclassification, automated program learning and other cases. A novel\ncharacterization of the concept of a \"good quality feature\" for such an\noptimization problem is provided; and a proposal regarding the integration of\nquality based feature selection into metalearning is suggested, wherein the\nquality of a feature for a problem is estimated using knowledge about related\nfeatures in the context of related problems. Results are presented regarding\nextensive testing of this \"feature metalearning\" approach on supervised text\nclassification problems; it is demonstrated that, in this context, feature\nmetalearning can provide significant and sometimes dramatic speedup over\nstandard feature selection heuristics.\n",
"title": "Metalearning for Feature Selection"
}
| null | null | null | null | true | null |
10882
| null |
Default
| null | null |
null |
{
"abstract": " Portable computing devices, which include tablets, smart phones and various\ntypes of wearable sensors, experienced a rapid development in recent years. One\nof the most critical limitations for these devices is the power consumption as\nthey use batteries as the power supply. However, the bottleneck of the power\nsaving schemes in both hardware design and software algorithm is the huge\nvariability in power consumption. The variability is caused by a myriad of\nfactors, including the manufacturing process, the ambient environment\n(temperature, humidity), the aging effects and etc. As the technology node\nscaled down to 28nm and even lower, the variability becomes more severe. As a\nresult, a platform for variability characterization seems to be very necessary\nand helpful.\n",
"title": "Variability-Aware Design for Energy Efficient Computational Artificial Intelligence Platform"
}
| null | null | null | null | true | null |
10883
| null |
Default
| null | null |
null |
{
"abstract": " Spinal cord stimulation has enabled humans with motor complete spinal cord\ninjury (SCI) to independently stand and recover some lost autonomic function.\nQuantifying the quality of bipedal standing under spinal stimulation is\nimportant for spinal rehabilitation therapies and for new strategies that seek\nto combine spinal stimulation and rehabilitative robots (such as exoskeletons)\nin real time feedback. To study the potential for automated electromyography\n(EMG) analysis in SCI, we evaluated the standing quality of paralyzed patients\nundergoing electrical spinal cord stimulation using both video and\nmulti-channel surface EMG recordings during spinal stimulation therapy\nsessions. The quality of standing under different stimulation settings was\nquantified manually by experienced clinicians. By correlating features of the\nrecorded EMG activity with the expert evaluations, we show that multi-channel\nEMG recording can provide accurate, fast, and robust estimation for the quality\nof bipedal standing in spinally stimulated SCI patients. Moreover, our analysis\nshows that the total number of EMG channels needed to effectively predict\nstanding quality can be reduced while maintaining high estimation accuracy,\nwhich provides more flexibility for rehabilitation robotic systems to\nincorporate EMG recordings.\n",
"title": "Quantifying Performance of Bipedal Standing with Multi-channel EMG"
}
| null | null | null | null | true | null |
10884
| null |
Default
| null | null |
null |
{
"abstract": " General description of an on-line procedure of calibration for IGRT (Image\nGuided Radiotherapy) is given. The algorithm allows to improve targeting cancer\nby estimating its position in space and suggests appropriate correction of the\nposition of the patient. The description is given in the Geometric Algebra\nlanguage which significantly simplifies calculations and clarifies\npresentation.\n",
"title": "A framework for on-line calibration of LINAC devices"
}
| null | null | null | null | true | null |
10885
| null |
Default
| null | null |
null |
{
"abstract": " Media seems to have become more partisan, often providing a biased coverage\nof news catering to the interest of specific groups. It is therefore essential\nto identify credible information content that provides an objective narrative\nof an event. News communities such as digg, reddit, or newstrust offer\nrecommendations, reviews, quality ratings, and further insights on journalistic\nworks. However, there is a complex interaction between different factors in\nsuch online communities: fairness and style of reporting, language clarity and\nobjectivity, topical perspectives (like political viewpoint), expertise and\nbias of community members, and more. This paper presents a model to\nsystematically analyze the different interactions in a news community between\nusers, news, and sources. We develop a probabilistic graphical model that\nleverages this joint interaction to identify 1) highly credible news articles,\n2) trustworthy news sources, and 3) expert users who perform the role of\n\"citizen journalists\" in the community. Our method extends CRF models to\nincorporate real-valued ratings, as some communities have very fine-grained\nscales that cannot be easily discretized without losing information. To the\nbest of our knowledge, this paper is the first full-fledged analysis of\ncredibility, trust, and expertise in news communities.\n",
"title": "People on Media: Jointly Identifying Credible News and Trustworthy Citizen Journalists in Online Communities"
}
| null | null | null | null | true | null |
10886
| null |
Default
| null | null |
null |
{
"abstract": " Currently, we are in an environment where the fraction of automated vehicles\nis negligibly small. We anticipate that this fraction will increase in coming\ndecades before if ever, we have a fully automated transportation system.\nMotivated by this we address the problem of provable safety of mixed traffic\nconsisting of both intelligent vehicles (IVs) as well as human-driven vehicles\n(HVs). An important issue that arises is that such mixed systems may well have\nlesser throughput than all human traffic systems if the automated vehicles are\nexpected to remain provably safe with respect to human traffic. This\nnecessitates the consideration of strategies such as platooning of automated\nvehicles in order to increase the throughput. In this paper, we address the\ndesign of provably safe systems consisting of a mix of automated and\nhuman-driven vehicles including the use of platooning by automated vehicles.\nWe design motion planing policies and coordination rules for participants in\nthis novel mixed system. HVs are considered as nearsighted and modeled with\nrelatively loose constraints, while IVs are considered as capable of following\nmuch tighter constraints. HVs are expected to follow reasonable and simple\nrules. IVs are designed to move under a model predictive control (MPC) based\nmotion plans and coordination protocols. Our contribution of this paper is in\nshowing how to integrate these two types of models safely into a mixed system.\nSystem safety is proved in single lane scenarios, as well as in multi-lane\nsituations allowing lane changes.\n",
"title": "Towards Provably Safe Mixed Transportation Systems with Human-driven and Automated Vehicles"
}
| null | null |
[
"Computer Science"
] | null | true | null |
10887
| null |
Validated
| null | null |
null |
{
"abstract": " We consider the problem of streaming kernel regression, when the observations\narrive sequentially and the goal is to recover the underlying mean function,\nassumed to belong to an RKHS. The variance of the noise is not assumed to be\nknown. In this context, we tackle the problem of tuning the regularization\nparameter adaptively at each time step, while maintaining tight confidence\nbounds estimates on the value of the mean function at each point. To this end,\nwe first generalize existing results for finite-dimensional linear regression\nwith fixed regularization and known variance to the kernel setup with a\nregularization parameter allowed to be a measurable function of past\nobservations. Then, using appropriate self-normalized inequalities we build\nupper and lower bound estimates for the variance, leading to Bersntein-like\nconcentration bounds. The later is used in order to define the adaptive\nregularization. The bounds resulting from our technique are valid uniformly\nover all observation points and all time steps, and are compared against the\nliterature with numerical experiments. Finally, the potential of these tools is\nillustrated by an application to kernelized bandits, where we revisit the\nKernel UCB and Kernel Thompson Sampling procedures, and show the benefits of\nthe novel adaptive kernel tuning strategy.\n",
"title": "Streaming kernel regression with provably adaptive mean, variance, and regularization"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
10888
| null |
Validated
| null | null |
null |
{
"abstract": " Stochastic differential equations (SDEs) are increasingly used in\nlongitudinal data analysis, compartmental models, growth modelling, and other\napplications in a number of disciplines. Parameter estimation, however,\ncurrently requires specialized software packages that can be difficult to use\nand understand. This work develops and demonstrates an approach for estimating\nreducible SDEs using standard nonlinear least squares or mixed-effects\nsoftware. Reducible SDEs are obtained through a change of variables in linear\nSDEs, and are sufficiently flexible for modelling many situations. The approach\nis based on extending a known technique that converts maximum likelihood\nestimation for a Gaussian model with a nonlinear transformation of the\ndependent variable into an equivalent least-squares problem. A similar idea can\nbe used for Bayesian maximum a posteriori estimation. It is shown how to obtain\nparameter estimates for reducible SDEs containing both process and observation\nnoise, including hierarchical models with either fixed or random group\nparameters. Code and examples in R are given. Univariate SDEs are discussed in\ndetail, with extensions to the multivariate case outlined more briefly. The use\nof well tested and familiar standard software should make SDE modelling more\ntransparent and accessible. Keywords: stochastic processes; longitudinal data;\ngrowth curves; compartmental models; mixed-effects; R\n",
"title": "Estimating reducible stochastic differential equations by conversion to a least-squares problem"
}
| null | null | null | null | true | null |
10889
| null |
Default
| null | null |
null |
{
"abstract": " Learning an encoding of feature vectors in terms of an over-complete\ndictionary or a information geometric (Fisher vectors) construct is wide-spread\nin statistical signal processing and computer vision. In content based\ninformation retrieval using deep-learning classifiers, such encodings are\nlearnt on the flattened last layer, without adherence to the multi-linear\nstructure of the underlying feature tensor. We illustrate a variety of feature\nencodings incl. sparse dictionary coding and Fisher vectors along with\nproposing that a structured tensor factorization scheme enables us to perform\nretrieval that can be at par, in terms of average precision, with Fisher vector\nencoded image signatures. In short, we illustrate how structural constraints\nincrease retrieval fidelity.\n",
"title": "Deep Tensor Encoding"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
10890
| null |
Validated
| null | null |
null |
{
"abstract": " We propose hMDAP, a hybrid framework for large-scale data analytical\nprocessing on Spark, to support multi-paradigm process (incl. OLAP, machine\nlearning, and graph analysis etc.) in distributed environments. The framework\nfeatures a three-layer data process module and a business process module which\ncontrols the former. We will demonstrate the strength of hMDAP by using traffic\nscenarios in a real world.\n",
"title": "hMDAP: A Hybrid Framework for Multi-paradigm Data Analytical Processing on Spark"
}
| null | null | null | null | true | null |
10891
| null |
Default
| null | null |
null |
{
"abstract": " A novel approach for unsupervised domain adaptation for neural networks is\nproposed. It relies on metric-based regularization of the learning process. The\nmetric-based regularization aims at domain-invariant latent feature\nrepresentations by means of maximizing the similarity between domain-specific\nactivation distributions. The proposed metric results from modifying an\nintegral probability metric such that it becomes less translation-sensitive on\na polynomial function space. The metric has an intuitive interpretation in the\ndual space as the sum of differences of higher order central moments of the\ncorresponding activation distributions. Under appropriate assumptions on the\ninput distributions, error minimization is proven for the continuous case. As\ndemonstrated by an analysis of standard benchmark experiments for sentiment\nanalysis, object recognition and digit recognition, the outlined approach is\nrobust regarding parameter changes and achieves higher classification\naccuracies than comparable approaches. The source code is available at\nthis https URL.\n",
"title": "Robust Unsupervised Domain Adaptation for Neural Networks via Moment Alignment"
}
| null | null | null | null | true | null |
10892
| null |
Default
| null | null |
null |
{
"abstract": " We present in detail the convolutional neural network used in our previous\nwork to detect cosmic strings in cosmic microwave background (CMB) temperature\nanisotropy maps. By training this neural network on numerically generated CMB\ntemperature maps, with and without cosmic strings, the network can produce\nprediction maps that locate the position of the cosmic strings and provide a\nprobabilistic estimate of the value of the string tension $G\\mu$. Supplying\nnoiseless simulations of CMB maps with arcmin resolution to the network\nresulted in the accurate determination both of string locations and string\ntension for sky maps having strings with string tension as low as\n$G\\mu=5\\times10^{-9}$. The code is publicly available online. Though we trained\nthe network with a long straight string toy model, we show the network performs\nwell with realistic Nambu-Goto simulations.\n",
"title": "A Convolutional Neural Network For Cosmic String Detection in CMB Temperature Maps"
}
| null | null | null | null | true | null |
10893
| null |
Default
| null | null |
null |
{
"abstract": " Improving the quality of end-of-life care for hospitalized patients is a\npriority for healthcare organizations. Studies have shown that physicians tend\nto over-estimate prognoses, which in combination with treatment inertia results\nin a mismatch between patients wishes and actual care at the end of life. We\ndescribe a method to address this problem using Deep Learning and Electronic\nHealth Record (EHR) data, which is currently being piloted, with Institutional\nReview Board approval, at an academic medical center. The EHR data of admitted\npatients are automatically evaluated by an algorithm, which brings patients who\nare likely to benefit from palliative care services to the attention of the\nPalliative Care team. The algorithm is a Deep Neural Network trained on the EHR\ndata from previous years, to predict all-cause 3-12 month mortality of patients\nas a proxy for patients that could benefit from palliative care. Our\npredictions enable the Palliative Care team to take a proactive approach in\nreaching out to such patients, rather than relying on referrals from treating\nphysicians, or conduct time consuming chart reviews of all patients. We also\npresent a novel interpretation technique which we use to provide explanations\nof the model's predictions.\n",
"title": "Improving Palliative Care with Deep Learning"
}
| null | null | null | null | true | null |
10894
| null |
Default
| null | null |
null |
{
"abstract": " Convolutional sparse coding (CSC) improves sparse coding by learning a\nshift-invariant dictionary from the data. However, existing CSC algorithms\noperate in the batch mode and are expensive, in terms of both space and time,\non large datasets. In this paper, we alleviate these problems by using online\nlearning. The key is a reformulation of the CSC objective so that convolution\ncan be handled easily in the frequency domain and much smaller history matrices\nare needed. We use the alternating direction method of multipliers (ADMM) to\nsolve the resulting optimization problem and the ADMM subproblems have\nefficient closed-form solutions. Theoretical analysis shows that the learned\ndictionary converges to a stationary point of the optimization problem.\nExtensive experiments show that convergence of the proposed method is much\nfaster and its reconstruction performance is also better. Moreover, while\nexisting CSC algorithms can only run on a small number of images, the proposed\nmethod can handle at least ten times more images.\n",
"title": "Scalable Online Convolutional Sparse Coding"
}
| null | null | null | null | true | null |
10895
| null |
Default
| null | null |
null |
{
"abstract": " It is challenging to recognize facial action unit (AU) from spontaneous\nfacial displays, especially when they are accompanied by speech. The major\nreason is that the information is extracted from a single source, i.e., the\nvisual channel, in the current practice. However, facial activity is highly\ncorrelated with voice in natural human communications.\nInstead of solely improving visual observations, this paper presents a novel\naudiovisual fusion framework, which makes the best use of visual and acoustic\ncues in recognizing speech-related facial AUs. In particular, a dynamic\nBayesian network (DBN) is employed to explicitly model the semantic and dynamic\nphysiological relationships between AUs and phonemes as well as measurement\nuncertainty. A pilot audiovisual AU-coded database has been collected to\nevaluate the proposed framework, which consists of a \"clean\" subset containing\nfrontal faces under well controlled circumstances and a challenging subset with\nlarge head movements and occlusions. Experiments on this database have\ndemonstrated that the proposed framework yields significant improvement in\nrecognizing speech-related AUs compared to the state-of-the-art visual-based\nmethods especially for those AUs whose visual observations are impaired during\nspeech, and more importantly also outperforms feature-level fusion methods by\nexplicitly modeling and exploiting physiological relationships between AUs and\nphonemes.\n",
"title": "Improving Speech Related Facial Action Unit Recognition by Audiovisual Information Fusion"
}
| null | null | null | null | true | null |
10896
| null |
Default
| null | null |
null |
{
"abstract": " We propose an effective method for creating interpretable control agents, by\n\\textit{re-purposing} the function of a biological neural circuit model, to\ngovern simulated and real world reinforcement learning (RL) test-beds. Inspired\nby the structure of the nervous system of the soil-worm, \\emph{C. elegans}, we\nintroduce \\emph{Neuronal Circuit Policies} (NCPs) as a novel recurrent neural\nnetwork instance with liquid time-constants, universal approximation\ncapabilities and interpretable dynamics. We theoretically show that they can\napproximate any finite simulation time of a given continuous n-dimensional\ndynamical system, with $n$ output units and some hidden units. We model\ninstances of the policies and learn their synaptic and neuronal parameters to\ncontrol standard RL tasks and demonstrate its application for autonomous\nparking of a real rover robot on a pre-defined trajectory. For reconfiguration\nof the \\emph{purpose} of the neural circuit, we adopt a search-based RL\nalgorithm. We show that our neuronal circuit policies perform as good as deep\nneural network policies with the advantage of realizing interpretable dynamics\nat the cell-level. We theoretically find bounds for the time-varying dynamics\nof the circuits, and introduce a novel way to reason about networks' dynamics.\n",
"title": "Re-purposing Compact Neuronal Circuit Policies to Govern Reinforcement Learning Tasks"
}
| null | null | null | null | true | null |
10897
| null |
Default
| null | null |
null |
{
"abstract": " We present a simple, yet useful result about the expected value of the\ndeterminant of random sum of rank-one matrices. Computing such expectations in\ngeneral may involve a sum over exponentially many terms. Nevertheless, we show\nthat an interesting and useful class of such expectations that arise in, e.g.,\nD-optimal estimation and random graphs can be computed efficiently via\ncomputing a single determinant.\n",
"title": "On the Expected Value of the Determinant of Random Sum of Rank-One Matrices"
}
| null | null | null | null | true | null |
10898
| null |
Default
| null | null |
null |
{
"abstract": " A new approach using a hyperbolic-equation system (HES) is proposed to solve\nfor the electron fluids in quasi-neutral plasmas. The HES approach avoids\ntreatments of cross-diffusion terms which cause numerical instabilities in\nconventional approaches using an elliptic equation (EE). A test calculation\nreveals that the HES approach can robustly solve problems of strong magnetic\nconfinement by using an upwind method. The computation time of the HES approach\nis compared with that of the EE approach in terms of the size of the problem\nand the strength of magnetic confinement. The results indicate that the HES\napproach can be used to solve problems in a simple structured mesh without\nincreasing computational time compared to the EE approach and that it features\nfast convergence in conditions of strong magnetic confinement.\n",
"title": "A hyperbolic-equation system approach for magnetized electron fluids in quasi-neutral plasmas"
}
| null | null | null | null | true | null |
10899
| null |
Default
| null | null |
null |
{
"abstract": " A general and easy-to-code numerical method based on radial basis functions\n(RBFs) collocation is proposed for the solution of delay differential equations\n(DDEs). It relies on the interpolation properties of infinitely smooth RBFs,\nwhich allow for a large accuracy over a scattered and relatively small\ndiscretization support. Hardy's multiquadric is chosen as RBF and combined with\nthe Residual Subsampling Algorithm of Driscoll and Heryudono for support\nadaptivity. The performance of the method is very satisfactory, as demonstrated\nover a cross-section of benchmark DDEs, and by comparison with existing\ngeneral-purpose and specialized numerical schemes for DDEs.\n",
"title": "Solving delay differential equations through RBF collocation"
}
| null | null | null | null | true | null |
10900
| null |
Default
| null | null |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.