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": " The analysis of mixed data has been raising challenges in statistics and\nmachine learning. One of two most prominent challenges is to develop new\nstatistical techniques and methodologies to effectively handle mixed data by\nmaking the data less heterogeneous with minimum loss of information. The other\nchallenge is that such methods must be able to apply in large-scale tasks when\ndealing with huge amount of mixed data. To tackle these challenges, we\nintroduce parameter sharing and balancing extensions to our recent model, the\nmixed-variate restricted Boltzmann machine (MV.RBM) which can transform\nheterogeneous data into homogeneous representation. We also integrate\nstructured sparsity and distance metric learning into RBM-based models. Our\nproposed methods are applied in various applications including latent patient\nprofile modelling in medical data analysis and representation learning for\nimage retrieval. The experimental results demonstrate the models perform better\nthan baseline methods in medical data and outperform state-of-the-art rivals in\nimage dataset.\n",
"title": "Statistical Latent Space Approach for Mixed Data Modelling and Applications"
} | null | null | null | null | true | null | 1001 | null | Default | null | null |
null | {
"abstract": " The development of spintronic technology with increasingly dense, high-speed,\nand complex devices will be accelerated by accessible microscopy techniques\ncapable of probing magnetic phenomena on picosecond time scales and at deeply\nsub-micron length scales. A recently developed time-resolved magneto-thermal\nmicroscope provides a path towards this goal if it is augmented with a\npicosecond, nanoscale heat source. We theoretically study adiabatic\nnanofocusing and near-field heat induction using conical gold plasmonic\nantennas to generate sub-100 nm thermal gradients for time-resolved\nmagneto-thermal imaging. Finite element calculations of antenna-sample\ninteractions reveal focused electromagnetic loss profiles that are either\npeaked directly under the antenna or are annular, depending on the sample's\nconductivity, the antenna's apex radius, and the tip-sample separation. We find\nthat the thermal gradient is confined to 40 nm to 60 nm full width at half\nmaximum for realistic ranges of sample conductivity and apex radius. To\nmitigate this variation, which is undesirable for microscopy, we investigate\nthe use of a platinum capping layer on top of the sample as a thermal\ntransduction layer to produce heat uniformly across different sample materials.\nAfter determining the optimal capping layer thickness, we simulate the\nevolution of the thermal gradient in the underlying sample layer, and find that\nthe temporal width is below 10 ps. These results lay a theoretical foundation\nfor nanoscale, time-resolved magneto-thermal imaging.\n",
"title": "Near-field coupling of gold plasmonic antennas for sub-100 nm magneto-thermal microscopy"
} | null | null | null | null | true | null | 1002 | null | Default | null | null |
null | {
"abstract": " Motivation: P values derived from the null hypothesis significance testing\nframework are strongly affected by sample size, and are known to be\nirreproducible in underpowered studies, yet no suitable replacement has been\nproposed. Results: Here we present implementations of non-parametric\nstandardized median effect size estimates, dNEF, for high-throughput sequencing\ndatasets. Case studies are shown for transcriptome and tag-sequencing datasets.\nThe dNEF measure is shown to be more repro- ducible and robust than P values\nand requires sample sizes as small as 3 to reproducibly identify differentially\nabundant features. Availability: Source code and binaries freely available at:\nthis https URL, omicplotR, and\nthis https URL.\n",
"title": "A reproducible effect size is more useful than an irreproducible hypothesis test to analyze high throughput sequencing datasets"
} | null | null | null | null | true | null | 1003 | null | Default | null | null |
null | {
"abstract": " We develop high temperature series expansions for the thermodynamic\nproperties of the honeycomb-lattice Kitaev-Heisenberg model. Numerical results\nfor uniform susceptibility, heat capacity and entropy as a function of\ntemperature for different values of the Kitaev coupling $K$ and Heisenberg\nexachange coupling $J$ (with $|J|\\le |K|$) are presented. These expansions show\ngood convergence down to a temperature of a fraction of $K$ and in some cases\ndown to $T=K/10$. In the Kitaev exchange dominated regime, the inverse\nsusceptibility has a nearly linear temperature dependence over a wide\ntemperature range. However, we show that already at temperatures $10$-times the\nCurie-Weiss temperature, the effective Curie-Weiss constant estimated from the\ndata can be off by a factor of 2. We find that the magnitude of the heat\ncapacity maximum at the short-range order peak, is substantially smaller for\nsmall $J/K$ than for $J$ of order or larger than $K$. We suggest that this\nitself represents a simple marker for the relative importance of the Kitaev\nterms in these systems. Somewhat surprisingly, both heat capacity and\nsusceptibility data on Na$_2$IrO$_3$ are consistent with a dominant {\\it\nantiferromagnetic} Kitaev exchange constant of about $300-400$ $K$.\n",
"title": "High temperature thermodynamics of the honeycomb-lattice Kitaev-Heisenberg model: A high temperature series expansion study"
} | null | null | null | null | true | null | 1004 | null | Default | null | null |
null | {
"abstract": " The Baran metric $\\delta_E$ is a Finsler metric on the interior of $E\\subset\n\\R^n$ arising from Pluripotential Theory. We consider the few instances, namely\n$E$ being the ball, the simplex, or the sphere, where $\\delta_E$ is known to be\nRiemaniann and we prove that the eigenfunctions of the associated Laplace\nBeltrami operator (with no boundary conditions) are the orthogonal polynomials\nwith respect to the pluripotential equilibrium measure $\\mu_E$ of $E.$ We\nconjecture that this may hold in a wider generality.\nThe considered differential operators have been already introduced in the\nframework of orthogonal polynomials and studied in connection with certain\nsymmetry groups. In this work instead we highlight the relationships between\northogonal polynomials with respect to $\\mu_E$ and the Riemaniann structure\nnaturally arising from Pluripotential Theory\n",
"title": "Laplace Beltrami operator in the Baran metric and pluripotential equilibrium measure: the ball, the simplex and the sphere"
} | null | null | [
"Mathematics"
]
| null | true | null | 1005 | null | Validated | null | null |
null | {
"abstract": " We consider a condensate of exciton-polaritons in a diluted magnetic\nsemiconductor microcavity. Such system may exhibit magnetic self-trapping in\nthe case of sufficiently strong coupling between polaritons and magnetic ions\nembedded in the semiconductor. We investigate the effect of the nonequilibrium\nnature of exciton-polaritons on the physics of the resulting self-trapped\nmagnetic polarons. We find that multiple polarons can exist at the same time,\nand derive a critical condition for self-trapping which is different to the one\npredicted previously in the equilibrium case. Using the Bogoliubov-de Gennes\napproximation, we calculate the excitation spectrum and provide a physical\nexplanation in terms of the effective magnetic attraction between polaritons,\nmediated by the ion subsystem.\n",
"title": "Magnetic polarons in a nonequilibrium polariton condensate"
} | null | null | [
"Physics"
]
| null | true | null | 1006 | null | Validated | null | null |
null | {
"abstract": " We consider the statistical problem of recovering a hidden \"ground truth\"\nbinary labeling for the vertices of a graph up to low Hamming error from noisy\nedge and vertex measurements. We present new algorithms and a sharp\nfinite-sample analysis for this problem on trees and sparse graphs with poor\nexpansion properties such as hypergrids and ring lattices. Our method\ngeneralizes and improves over that of Globerson et al. (2015), who introduced\nthe problem for two-dimensional grid lattices.\nFor trees we provide a simple, efficient, algorithm that infers the ground\ntruth with optimal Hamming error has optimal sample complexity and implies\nrecovery results for all connected graphs. Here, the presence of side\ninformation is critical to obtain a non-trivial recovery rate. We then show how\nto adapt this algorithm to tree decompositions of edge-subgraphs of certain\ngraph families such as lattices, resulting in optimal recovery error rates that\ncan be obtained efficiently\nThe thrust of our analysis is to 1) use the tree decomposition along with\nedge measurements to produce a small class of viable vertex labelings and 2)\napply an analysis influenced by statistical learning theory to show that we can\ninfer the ground truth from this class using vertex measurements. We show the\npower of our method in several examples including hypergrids, ring lattices,\nand the Newman-Watts model for small world graphs. For two-dimensional grids,\nour results improve over Globerson et al. (2015) by obtaining optimal recovery\nin the constant-height regime.\n",
"title": "Inference in Sparse Graphs with Pairwise Measurements and Side Information"
} | null | null | null | null | true | null | 1007 | null | Default | null | null |
null | {
"abstract": " We consider the problem of estimating an expected outcome from a stochastic\nsimulation model using importance sampling. We propose a two-stage procedure\nthat involves a regression stage and a sampling stage to construct our\nestimator. We introduce a parametric and a nonparametric regression estimator\nin the first stage and study how the allocation between the two stages affects\nthe performance of final estimator. We derive the oracle property for both\napproaches. We analyze the empirical performances of our approaches using two\nsimulated data and a case study on wind turbine reliability evaluation.\n",
"title": "Oracle Importance Sampling for Stochastic Simulation Models"
} | null | null | null | null | true | null | 1008 | null | Default | null | null |
null | {
"abstract": " Generalized cross validation (GCV) is one of the most important approaches\nused to estimate parameters in the context of inverse problems and\nregularization techniques. A notable example is the determination of the\nsmoothness parameter in splines. When the data are generated by a state space\nmodel, like in the spline case, efficient algorithms are available to evaluate\nthe GCV score with complexity that scales linearly in the data set size.\nHowever, these methods are not amenable to on-line applications since they rely\non forward and backward recursions. Hence, if the objective has been evaluated\nat time $t-1$ and new data arrive at time t, then O(t) operations are needed to\nupdate the GCV score. In this paper we instead show that the update cost is\n$O(1)$, thus paving the way to the on-line use of GCV. This result is obtained\nby deriving the novel GCV filter which extends the classical Kalman filter\nequations to efficiently propagate the GCV score over time. We also illustrate\napplications of the new filter in the context of state estimation and on-line\nregularized linear system identification.\n",
"title": "The Generalized Cross Validation Filter"
} | null | null | null | null | true | null | 1009 | null | Default | null | null |
null | {
"abstract": " Biochemical oscillations are prevalent in living organisms. Systems with a\nsmall number of constituents cannot sustain coherent oscillations for an\nindefinite time because of fluctuations in the period of oscillation. We show\nthat the number of coherent oscillations that quantifies the precision of the\noscillator is universally bounded by the thermodynamic force that drives the\nsystem out of equilibrium and by the topology of the underlying biochemical\nnetwork of states. Our results are valid for arbitrary Markov processes, which\nare commonly used to model biochemical reactions. We apply our results to a\nmodel for a single KaiC protein and to an activator-inhibitor model that\nconsists of several molecules. From a mathematical perspective, based on strong\nnumerical evidence, we conjecture a universal constraint relating the imaginary\nand real parts of the first non-trivial eigenvalue of a stochastic matrix.\n",
"title": "Coherence of Biochemical Oscillations is Bounded by Driving Force and Network Topology"
} | null | null | null | null | true | null | 1010 | null | Default | null | null |
null | {
"abstract": " Algorithms are often used to produce decision-making rules that classify or\nevaluate individuals. When these individuals have incentives to be classified a\ncertain way, they may behave strategically to influence their outcomes. We\ndevelop a model for how strategic agents can invest effort in order to change\nthe outcomes they receive, and we give a tight characterization of when such\nagents can be incentivized to invest specified forms of effort into improving\ntheir outcomes as opposed to \"gaming\" the classifier. We show that whenever any\n\"reasonable\" mechanism can do so, a simple linear mechanism suffices.\n",
"title": "How Do Classifiers Induce Agents To Invest Effort Strategically?"
} | null | null | null | null | true | null | 1011 | null | Default | null | null |
null | {
"abstract": " In this work we present a technique to use natural language to help\nreinforcement learning generalize to unseen environments. This technique uses\nneural machine translation, specifically the use of encoder-decoder networks,\nto learn associations between natural language behavior descriptions and\nstate-action information. We then use this learned model to guide agent\nexploration using a modified version of policy shaping to make it more\neffective at learning in unseen environments. We evaluate this technique using\nthe popular arcade game, Frogger, under ideal and non-ideal conditions. This\nevaluation shows that our modified policy shaping algorithm improves over a\nQ-learning agent as well as a baseline version of policy shaping.\n",
"title": "Guiding Reinforcement Learning Exploration Using Natural Language"
} | null | null | null | null | true | null | 1012 | null | Default | null | null |
null | {
"abstract": " In light of the classic impossibility results of Arrow and Gibbard and\nSatterthwaite regarding voting with ordinal rules, there has been recent\ninterest in characterizing how well common voting rules approximate the social\noptimum. In order to quantify the quality of approximation, it is natural to\nconsider the candidates and voters as embedded within a common metric space,\nand to ask how much further the chosen candidate is from the population as\ncompared to the socially optimal one. We use this metric preference model to\nexplore a fundamental and timely question: does the social welfare of a\npopulation improve when candidates are representative of the population? If so,\nthen by how much, and how does the answer depend on the complexity of the\nmetric space?\nWe restrict attention to the most fundamental and common social choice\nsetting: a population of voters, two independently drawn candidates, and a\nmajority rule election. When candidates are not representative of the\npopulation, it is known that the candidate selected by the majority rule can be\nthrice as far from the population as the socially optimal one. We examine how\nthis ratio improves when candidates are drawn independently from the population\nof voters. Our results are two-fold: When the metric is a line, the ratio\nimproves from $3$ to $4-2\\sqrt{2}$, roughly $1.1716$; this bound is tight. When\nthe metric is arbitrary, we show a lower bound of $1.5$ and a constant upper\nbound strictly better than $2$ on the approximation ratio of the majority rule.\nThe positive result depends in part on the assumption that candidates are\nindependent and identically distributed. However, we show that independence\nalone is not enough to achieve the upper bound: even when candidates are drawn\nindependently, if the population of candidates can be different from the\nvoters, then an upper bound of $2$ on the approximation is tight.\n",
"title": "Of the People: Voting Is More Effective with Representative Candidates"
} | null | null | null | null | true | null | 1013 | null | Default | null | null |
null | {
"abstract": " In this paper, we investigate a coverage extension scheme based on orthogonal\nrandom precoding (ORP) for the downlink of massive multiple-input\nmultiple-output (MIMO) systems. In this scheme, a precoding matrix consisting\nof orthogonal vectors is employed at the transmitter to enhance the maximum\nsignal-to-interference-plus-noise ratio (SINR) of the user. To analyze and\noptimize the ORP scheme in terms of cell coverage, we derive the analytical\nexpressions of the downlink coverage probability for two receiver structures,\nnamely, the single-antenna (SA) receiver and multiple-antenna receiver with\nantenna selection (AS). The simulation results show that the analytical\nexpressions accurately capture the coverage behaviors of the systems employing\nthe ORP scheme. It is also shown that the optimal coverage performance is\nachieved when a single precoding vector is used under the condition that the\nthreshold of the signal-to-noise ratio of the coverage is greater than one. The\nperformance of the ORP scheme is further analyzed when different random\nprecoder groups are utilized over multiple time slots to exploit precoding\ndiversity. The numerical results show that the proposed ORP scheme over\nmultiple time slots provides a substantial coverage gain over the space-time\ncoding scheme despite its low feedback overhead.\n",
"title": "Cell Coverage Extension with Orthogonal Random Precoding for Massive MIMO Systems"
} | null | null | null | null | true | null | 1014 | null | Default | null | null |
null | {
"abstract": " We introduce a new paradigm that is important for community detection in the\nrealm of network analysis. Networks contain a set of strong, dominant\ncommunities, which interfere with the detection of weak, natural community\nstructure. When most of the members of the weak communities also belong to\nstronger communities, they are extremely hard to be uncovered. We call the weak\ncommunities the hidden community structure.\nWe present a novel approach called HICODE (HIdden COmmunity DEtection) that\nidentifies the hidden community structure as well as the dominant community\nstructure. By weakening the strength of the dominant structure, one can uncover\nthe hidden structure beneath. Likewise, by reducing the strength of the hidden\nstructure, one can more accurately identify the dominant structure. In this\nway, HICODE tackles both tasks simultaneously.\nExtensive experiments on real-world networks demonstrate that HICODE\noutperforms several state-of-the-art community detection methods in uncovering\nboth the dominant and the hidden structure. In the Facebook university social\nnetworks, we find multiple non-redundant sets of communities that are strongly\nassociated with residential hall, year of registration or career position of\nthe faculties or students, while the state-of-the-art algorithms mainly locate\nthe dominant ground truth category. In the Due to the difficulty of labeling\nall ground truth communities in real-world datasets, HICODE provides a\npromising approach to pinpoint the existing latent communities and uncover\ncommunities for which there is no ground truth. Finding this unknown structure\nis an extremely important community detection problem.\n",
"title": "Hidden Community Detection in Social Networks"
} | null | null | null | null | true | null | 1015 | null | Default | null | null |
null | {
"abstract": " We reevaluate the Zemach, recoil and polarizability corrections to the\nhyperfine splitting in muonic hydrogen expressing them through the low-energy\nproton structure constants and obtain the precise values of the Zemach radius\nand two-photon exchange (TPE) contribution. The uncertainty of TPE correction\nto S energy levels in muonic hydrogen of 105 ppm exceeds the ppm accuracy level\nof the forthcoming 1S hyperfine splitting measurements at PSI, J-PARC and\nRIKEN-RAL.\n",
"title": "Two-photon exchange correction to the hyperfine splitting in muonic hydrogen"
} | null | null | null | null | true | null | 1016 | null | Default | null | null |
null | {
"abstract": " Ising models describe the joint probability distribution of a vector of\nbinary feature variables. Typically, not all the variables interact with each\nother and one is interested in learning the presumably sparse network structure\nof the interacting variables. However, in the presence of latent variables, the\nconventional method of learning a sparse model might fail. This is because the\nlatent variables induce indirect interactions of the observed variables. In the\ncase of only a few latent conditional Gaussian variables these spurious\ninteractions contribute an additional low-rank component to the interaction\nparameters of the observed Ising model. Therefore, we propose to learn a sparse\n+ low-rank decomposition of the parameters of an Ising model using a convex\nregularized likelihood problem. We show that the same problem can be obtained\nas the dual of a maximum-entropy problem with a new type of relaxation, where\nthe sample means collectively need to match the expected values only up to a\ngiven tolerance. The solution to the convex optimization problem has\nconsistency properties in the high-dimensional setting, where the number of\nobserved binary variables and the number of latent conditional Gaussian\nvariables are allowed to grow with the number of training samples.\n",
"title": "Ising Models with Latent Conditional Gaussian Variables"
} | null | null | null | null | true | null | 1017 | null | Default | null | null |
null | {
"abstract": " Direct experimental investigations of the low-energy electronic structure of\nthe Na$_2$IrO$_3$ iridate insulator are sparse and draw two conflicting\npictures. One relies on flat bands and a clear gap, the other involves\ndispersive states approaching the Fermi level, pointing to surface metallicity.\nHere, by a combination of angle-resolved photoemission, photoemission electron\nmicroscopy, and x-ray absorption, we show that the correct picture is more\ncomplex and involves an anomalous band, arising from charge transfer from Na\natoms to Ir-derived states. Bulk quasiparticles do exist, but in one of the two\npossible surface terminations the charge transfer is smaller and they remain\nelusive.\n",
"title": "Quasiparticles and charge transfer at the two surfaces of the honeycomb iridate Na$_2$IrO$_3$"
} | null | null | null | null | true | null | 1018 | null | Default | null | null |
null | {
"abstract": " We establish a fundamental property of bivariate Pareto records for\nindependent observations uniformly distributed in the unit square. We prove\nthat the asymptotic conditional distribution of the number of records broken by\nan observation given that the observation sets a record is Geometric with\nparameter 1/2.\n",
"title": "Breaking Bivariate Records"
} | null | null | null | null | true | null | 1019 | null | Default | null | null |
null | {
"abstract": " This work compares several node (and network) criticality measures\nquantifying to which extend each node is critical with respect to the\ncommunication flow between nodes of the network, and introduces a new measure\nbased on the Bag-of-Paths (BoP) framework. Network disconnection simulation\nexperiments show that the new BoP measure outperforms all the other measures on\na sample of Erdos-Renyi and Albert-Barabasi graphs. Furthermore, a faster\n(still O(n^3)), approximate, BoP criticality relying on the Sherman-Morrison\nrank-one update of a matrix is introduced for tackling larger networks. This\napproximate measure shows similar performances as the original, exact, one.\n",
"title": "A Bag-of-Paths Node Criticality Measure"
} | null | null | [
"Computer Science",
"Physics"
]
| null | true | null | 1020 | null | Validated | null | null |
null | {
"abstract": " We consider a programming language based on the lamplighter group that uses\nonly composition and iteration as control structures. We derive generating\nfunctions and counting formulas for this language and special subsets of it,\nestablishing lower and upper bounds on the growth rate of semantically distinct\nprograms. Finally, we show how to sample random programs and analyze the\ndistribution of runtimes induced by such sampling.\n",
"title": "Generation and analysis of lamplighter programs"
} | null | null | null | null | true | null | 1021 | null | Default | null | null |
null | {
"abstract": " We propose an extended variant of the reformulation and decomposition\nalgorithm for solving a special class of mixed-integer bilevel linear programs\n(MIBLPs) where continuous and integer variables are involved in both upper- and\nlower-level problems. In particular, we consider MIBLPs with upper-level\nconstraints that involve lower-level variables. We assume that the inducible\nregion is nonempty and all variables are bounded. By using the reformulation\nand decomposition scheme, an MIBLP is first converted into its equivalent\nsingle-level formulation, then computed by a column-and-constraint generation\nbased decomposition algorithm. The solution procedure is enhanced by a\nprojection strategy that does not require the relatively complete response\nproperty. To ensure its performance, we prove that our new method converges to\nthe global optimal solution in a finite number of iterations. A large-scale\ncomputational study on random instances and instances of hierarchical supply\nchain planning are presented to demonstrate the effectiveness of the algorithm.\n",
"title": "A Projection-Based Reformulation and Decomposition Algorithm for Global Optimization of a Class of Mixed Integer Bilevel Linear Programs"
} | null | null | null | null | true | null | 1022 | null | Default | null | null |
null | {
"abstract": " Continuing the study of preduals of spaces $\\mathcal{L}(H,Y)$ of bounded,\nlinear maps, we consider the situation that $H$ is a Hilbert space. We\nestablish a natural correspondence between isometric preduals of\n$\\mathcal{L}(H,Y)$ and isometric preduals of $Y$.\nThe main ingredient is a Tomiyama-type result which shows that every\ncontractive projection that complements $\\mathcal{L}(H,Y)$ in its bidual is\nautomatically a right $\\mathcal{L}(H)$-module map.\nAs an application, we show that isometric preduals of\n$\\mathcal{L}(\\mathcal{S}_1)$, the algebra of operators on the space of\ntrace-class operators, correspond to isometric preduals of $\\mathcal{S}_1$\nitself (and there is an abundance of them). On the other hand, the compact\noperators are the unique predual of $\\mathcal{S}_1$ making its multiplication\nseparately weak* continuous.\n",
"title": "Preduals for spaces of operators involving Hilbert spaces and trace-class operators"
} | null | null | [
"Mathematics"
]
| null | true | null | 1023 | null | Validated | null | null |
null | {
"abstract": " EPG graphs, introduced by Golumbic et al. in 2009, are edge-intersection\ngraphs of paths on an orthogonal grid. The class $B_k$-EPG is the subclass of\nEPG graphs where the path on the grid associated to each vertex has at most $k$\nbends. Epstein et al. showed in 2013 that computing a maximum clique in\n$B_1$-EPG graphs is polynomial. As remarked in [Heldt et al., 2014], when the\nnumber of bends is at least $4$, the class contains $2$-interval graphs for\nwhich computing a maximum clique is an NP-hard problem. The complexity status\nof the Maximum Clique problem remains open for $B_2$ and $B_3$-EPG graphs. In\nthis paper, we show that we can compute a maximum clique in polynomial time in\n$B_2$-EPG graphs given a representation of the graph.\nMoreover, we show that a simple counting argument provides a\n${2(k+1)}$-approximation for the coloring problem on $B_k$-EPG graphs without\nknowing the representation of the graph. It generalizes a result of [Epstein et\nal, 2013] on $B_1$-EPG graphs (where the representation was needed).\n",
"title": "Computing maximum cliques in $B_2$-EPG graphs"
} | null | null | null | null | true | null | 1024 | null | Default | null | null |
null | {
"abstract": " The Web is an important resource for understanding and diagnosing medical\nconditions. Based on exposure to online content, people may develop undue\nhealth concerns, believing that common and benign symptoms are explained by\nserious illnesses. In this paper, we investigate potential strategies to mine\nqueries and searcher histories for clues that could help search engines choose\nthe most appropriate information to present in response to exploratory medical\nqueries. To do this, we performed a longitudinal study of health search\nbehavior using the logs of a popular search engine. We found that query\nvariations which might appear innocuous (e.g. \"bad headache\" vs \"severe\nheadache\") may hold valuable information about the searcher which could be used\nby search engines to improve performance. Furthermore, we investigated how\nmedically concerned users respond differently to search engine result pages\n(SERPs) and find that their disposition for clicking on concerning pages is\npronounced, potentially leading to a self-reinforcement of concern. Finally, we\nstudied to which degree variations in the SERP impact future search and\nreal-world health-seeking behavior and obtained some surprising results (e.g.,\nviewing concerning pages may lead to a short-term reduction of real-world\nhealth seeking).\n",
"title": "Interactions between Health Searchers and Search Engines"
} | null | null | null | null | true | null | 1025 | null | Default | null | null |
null | {
"abstract": " For an effect algebra $A$, we examine the category of all morphisms from\nfinite Boolean algebras into $A$. This category can be described as a category\nof elements of a presheaf $R(A)$ on the category of finite Boolean algebras. We\nprove that some properties (being an orthoalgebra, the Riesz decomposition\nproperty, being a Boolean algebra) of an effect algebra $A$ can be\ncharacterized by properties of the category of elements of the presheaf $R(A)$.\nWe prove that the tensor product of of effect algebras arises as a left Kan\nextension of the free product of finite Boolean algebras along the inclusion\nfunctor. As a consequence, the tensor product of effect algebras can be\nexpressed by means of the Day convolution of presheaves on finite Boolean\nalgebras.\n",
"title": "Effect algebras as presheaves on finite Boolean algebras"
} | null | null | [
"Mathematics"
]
| null | true | null | 1026 | null | Validated | null | null |
null | {
"abstract": " In a previous work we have detailed the requirements to obtain a maximal\nperformance benefit by implementing fully connected deep neural networks (DNN)\nin form of arrays of resistive devices for deep learning. This concept of\nResistive Processing Unit (RPU) devices we extend here towards convolutional\nneural networks (CNNs). We show how to map the convolutional layers to RPU\narrays such that the parallelism of the hardware can be fully utilized in all\nthree cycles of the backpropagation algorithm. We find that the noise and bound\nlimitations imposed due to analog nature of the computations performed on the\narrays effect the training accuracy of the CNNs. Noise and bound management\ntechniques are presented that mitigate these problems without introducing any\nadditional complexity in the analog circuits and can be addressed by the\ndigital circuits. In addition, we discuss digitally programmable update\nmanagement and device variability reduction techniques that can be used\nselectively for some of the layers in a CNN. We show that combination of all\nthose techniques enables a successful application of the RPU concept for\ntraining CNNs. The techniques discussed here are more general and can be\napplied beyond CNN architectures and therefore enables applicability of RPU\napproach for large class of neural network architectures.\n",
"title": "Training Deep Convolutional Neural Networks with Resistive Cross-Point Devices"
} | null | null | [
"Computer Science",
"Statistics"
]
| null | true | null | 1027 | null | Validated | null | null |
null | {
"abstract": " We study magnetic Taylor-Couette flow in a system having nondimensional radii\n$r_i=1$ and $r_o=2$, and periodic in the axial direction with wavelengths\n$h\\ge100$. The rotation ratio of the inner and outer cylinders is adjusted to\nbe slightly in the Rayleigh-stable regime, where magnetic fields are required\nto destabilize the flow, in this case triggering the axisymmetric helical\nmagnetorotational instability (HMRI). Two choices of imposed magnetic field are\nconsidered, both having the same azimuthal component $B_\\phi=r^{-1}$, but\ndiffering axial components. The first choice has $B_z=0.1$, and yields the\nfamiliar HMRI, consisting of unidirectionally traveling waves. The second\nchoice has $B_z\\approx0.1\\sin(2\\pi z/h)$, and yields HMRI waves that travel in\nopposite directions depending on the sign of $B_z$. The first configuration\ncorresponds to a convective instability, the second to an absolute instability.\nThe two variants behave very similarly regarding both linear onset as well as\nnonlinear equilibration.\n",
"title": "Absolute versus convective helical magnetorotational instabilities in Taylor-Couette flows"
} | null | null | null | null | true | null | 1028 | null | Default | null | null |
null | {
"abstract": " Compared with the two-component Camassa-Holm system, the modified\ntwo-component Camassa-Holm system introduces a regularized density which makes\npossible the existence of solutions of lower regularity, and in particular of\nmultipeakon solutions. In this paper, we derive a new pointwise invariant for\nthe modified two-component Camassa-Holm system. The derivation of the invariant\nuses directly the symmetry of the system, following the classical argument of\nNoether's theorem. The existence of the multipeakon solutions can be directly\ninferred from this pointwise invariant. This derivation shows the strong\nconnection between symmetries and the existence of special solutions. The\nobservation also holds for the scalar Camassa-Holm equation and, for\ncomparison, we have also included the corresponding derivation. Finally, we\ncompute explicitly the solutions obtained for the peakon-antipeakon case. We\nobserve the existence of a periodic solution which has not been reported in the\nliterature previously. This case shows the attractive effect that the\nintroduction of an elastic potential can have on the solutions.\n",
"title": "Symmetries and multipeakon solutions for the modified two-component Camassa-Holm system"
} | null | null | [
"Mathematics"
]
| null | true | null | 1029 | null | Validated | null | null |
null | {
"abstract": " The two dimensional incompressible Navier-Stokes equation on $D_\\delta := [0,\n2\\pi\\delta] \\times [0, 2\\pi]$ with $\\delta \\approx 1$, periodic boundary\nconditions, and viscosity $0 < \\nu \\ll 1$ is considered. Bars and dipoles, two\nexplicitly given quasi-stationary states of the system, evolve on the time\nscale $\\mathcal{O}(e^{-\\nu t})$ and have been shown to play a key role in its\nlong-time evolution. Of particular interest is the role that $\\delta$ plays in\nselecting which of these two states is observed. Recent numerical studies\nsuggest that, after a transient period of rapid decay of the high Fourier\nmodes, the bar state will be selected if $\\delta \\neq 1$, while the dipole will\nbe selected if $\\delta = 1$. Our results support this claim and seek to\nmathematically formalize it. We consider the system in Fourier space, project\nit onto a center manifold consisting of the lowest eight Fourier modes, and use\nthis as a model to study the selection of bars and dipoles. It is shown for\nthis ODE model that the value of $\\delta$ controls the behavior of the\nasymptotic ratio of the low modes, thus determining the likelihood of observing\na bar state or dipole after an initial transient period. Moreover, in our\nmodel, for all $\\delta \\approx 1$, there is an initial time period in which the\nhigh modes decay at the rapid rate $\\mathcal{O}(e^{-t/\\nu})$, while the low\nmodes evolve at the slower $\\mathcal{O}(e^{-\\nu t})$ rate. The results for the\nODE model are proven using energy estimates and invariant manifolds and further\nsupported by formal asymptotic expansions and numerics.\n",
"title": "Selection of quasi-stationary states in the Navier-Stokes equation on the torus"
} | null | null | null | null | true | null | 1030 | null | Default | null | null |
null | {
"abstract": " Training model to generate data has increasingly attracted research attention\nand become important in modern world applications. We propose in this paper a\nnew geometry-based optimization approach to address this problem. Orthogonal to\ncurrent state-of-the-art density-based approaches, most notably VAE and GAN, we\npresent a fresh new idea that borrows the principle of minimal enclosing ball\nto train a generator G\\left(\\bz\\right) in such a way that both training and\ngenerated data, after being mapped to the feature space, are enclosed in the\nsame sphere. We develop theory to guarantee that the mapping is bijective so\nthat its inverse from feature space to data space results in expressive\nnonlinear contours to describe the data manifold, hence ensuring data generated\nare also lying on the data manifold learned from training data. Our model\nenjoys a nice geometric interpretation, hence termed Geometric Enclosing\nNetworks (GEN), and possesses some key advantages over its rivals, namely\nsimple and easy-to-control optimization formulation, avoidance of mode\ncollapsing and efficiently learn data manifold representation in a completely\nunsupervised manner. We conducted extensive experiments on synthesis and\nreal-world datasets to illustrate the behaviors, strength and weakness of our\nproposed GEN, in particular its ability to handle multi-modal data and quality\nof generated data.\n",
"title": "Geometric Enclosing Networks"
} | null | null | null | null | true | null | 1031 | null | Default | null | null |
null | {
"abstract": " We introduce a pliable lasso method for estimation of interaction effects in\nthe Cox proportional hazards model framework. The pliable lasso is a linear\nmodel that includes interactions between covariates X and a set of modifying\nvariables Z and assumes sparsity of the main effects and interaction effects.\nThe hierarchical penalty excludes interaction effects when the corresponding\nmain effects are zero: this avoids overfitting and an explosion of model\ncomplexity. We extend this method to the Cox model for survival data,\nincorporating modifiers that are either fixed or varying in time into the\npartial likelihood. For example, this allows modeling of survival times that\ndiffer based on interactions of genes with age, gender, or other demographic\ninformation. The optimization is done by blockwise coordinate descent on a\nsecond order approximation of the objective.\n",
"title": "A pliable lasso for the Cox model"
} | null | null | null | null | true | null | 1032 | null | Default | null | null |
null | {
"abstract": " We report on the experimental realization of a state-dependent lattice for a\ntwo-orbital fermionic quantum gas with strong interorbital spin exchange. In\nour state-dependent lattice, the ground and metastable excited electronic\nstates of $^{173}$Yb take the roles of itinerant and localized magnetic\nmoments, respectively. Repulsive on-site interactions in conjunction with the\ntunnel mobility lead to spin exchange between mobile and localized particles,\nmodeling the coupling term in the well-known Kondo Hamiltonian. In addition, we\nfind that this exchange process can be tuned resonantly by varying the on-site\nconfinement. We attribute this to a resonant coupling to center-of-mass excited\nbound states of one interorbital scattering channel.\n",
"title": "Localized magnetic moments with tunable spin exchange in a gas of ultracold fermions"
} | null | null | null | null | true | null | 1033 | null | Default | null | null |
null | {
"abstract": " We prove versions of Khintchine's Theorem (1924) for approximations by\nrational numbers whose numerators lie in randomly chosen sets of integers, and\nwe explore the extent to which the monotonicity assumption can be removed.\nRoughly speaking, we show that if the number of available fractions for each\ndenominator grows too fast, then the monotonicity assumption cannot be removed.\nThere are questions in this random setting which may be seen as cognates of the\nDuffin-Schaeffer Conjecture (1941), and are likely to be more accessible. We\npoint out that the direct random analogue of the Duffin-Schaeffer Conjecture,\nlike the Duffin-Schaeffer Conjecture itself, implies Catlin's Conjecture\n(1976). It is not obvious whether the Duffin-Schaeffer Conjecture and its\nrandom version imply one another, and it is not known whether Catlin's\nConjecture implies either of them. The question of whether Catlin implies\nDuffin-Schaeffer has been unsettled for decades.\n",
"title": "Khintchine's Theorem with random fractions"
} | null | null | [
"Mathematics"
]
| null | true | null | 1034 | null | Validated | null | null |
null | {
"abstract": " Neural networks with random hidden nodes have gained increasing interest from\nresearchers and practical applications. This is due to their unique features\nsuch as very fast training and universal approximation property. In these\nnetworks the weights and biases of hidden nodes determining the nonlinear\nfeature mapping are set randomly and are not learned. Appropriate selection of\nthe intervals from which weights and biases are selected is extremely\nimportant. This topic has not yet been sufficiently explored in the literature.\nIn this work a method of generating random weights and biases is proposed. This\nmethod generates the parameters of the hidden nodes in such a way that\nnonlinear fragments of the activation functions are located in the input space\nregions with data and can be used to construct the surface approximating a\nnonlinear target function. The weights and biases are dependent on the input\ndata range and activation function type. The proposed methods allows us to\ncontrol the generalization degree of the model. These all lead to improvement\nin approximation performance of the network. Several experiments show very\npromising results.\n",
"title": "A Method of Generating Random Weights and Biases in Feedforward Neural Networks with Random Hidden Nodes"
} | null | null | [
"Computer Science",
"Statistics"
]
| null | true | null | 1035 | null | Validated | null | null |
null | {
"abstract": " Artificial neural networks have been successfully applied to a variety of\nmachine learning tasks, including image recognition, semantic segmentation, and\nmachine translation. However, few studies fully investigated ensembles of\nartificial neural networks. In this work, we investigated multiple widely used\nensemble methods, including unweighted averaging, majority voting, the Bayes\nOptimal Classifier, and the (discrete) Super Learner, for image recognition\ntasks, with deep neural networks as candidate algorithms. We designed several\nexperiments, with the candidate algorithms being the same network structure\nwith different model checkpoints within a single training process, networks\nwith same structure but trained multiple times stochastically, and networks\nwith different structure. In addition, we further studied the over-confidence\nphenomenon of the neural networks, as well as its impact on the ensemble\nmethods. Across all of our experiments, the Super Learner achieved best\nperformance among all the ensemble methods in this study.\n",
"title": "The Relative Performance of Ensemble Methods with Deep Convolutional Neural Networks for Image Classification"
} | null | null | null | null | true | null | 1036 | null | Default | null | null |
null | {
"abstract": " Suppose the data consist of a set $S$ of points $x_j, 1 \\leq j \\leq J$,\ndistributed in a bounded domain $D \\subset R^N$, where $N$ and $J$ are large\nnumbers. In this paper an algorithm is proposed for checking whether there\nexists a manifold $\\mathbb{M}$ of low dimension near which many of the points\nof $S$ lie and finding such $\\mathbb{M}$ if it exists. There are many dimension\nreduction algorithms, both linear and non-linear. Our algorithm is simple to\nimplement and has some advantages compared with the known algorithms. If there\nis a manifold of low dimension near which most of the data points lie, the\nproposed algorithm will find it. Some numerical results are presented\nillustrating the algorithm and analyzing its performance compared to the\nclassical PCA (principal component analysis) and Isomap.\n",
"title": "Representation of big data by dimension reduction"
} | null | null | null | null | true | null | 1037 | null | Default | null | null |
null | {
"abstract": " Let $k$ be a fixed integer. We determine the complexity of finding a\n$p$-partition $(V_1, \\dots, V_p)$ of the vertex set of a given digraph such\nthat the maximum out-degree of each of the digraphs induced by $V_i$, ($1\\leq\ni\\leq p$) is at least $k$ smaller than the maximum out-degree of $D$. We show\nthat this problem is polynomial-time solvable when $p\\geq 2k$ and ${\\cal\nNP}$-complete otherwise. The result for $k=1$ and $p=2$ answers a question\nposed in \\cite{bangTCS636}. We also determine, for all fixed non-negative\nintegers $k_1,k_2,p$, the complexity of deciding whether a given digraph of\nmaximum out-degree $p$ has a $2$-partition $(V_1,V_2)$ such that the digraph\ninduced by $V_i$ has maximum out-degree at most $k_i$ for $i\\in [2]$. It\nfollows from this characterization that the problem of deciding whether a\ndigraph has a 2-partition $(V_1,V_2)$ such that each vertex $v\\in V_i$ has at\nleast as many neighbours in the set $V_{3-i}$ as in $V_i$, for $i=1,2$ is\n${\\cal NP}$-complete. This solves a problem from \\cite{kreutzerEJC24} on\nmajority colourings.\n",
"title": "Out-degree reducing partitions of digraphs"
} | null | null | null | null | true | null | 1038 | null | Default | null | null |
null | {
"abstract": " These notes are intended to provide a brief primer in plasma physics,\nintroducing common definitions, basic properties, and typical processes found\nin plasmas. These concepts are inherent in contemporary plasma-based\naccelerator schemes, and thus provide a foundation for the more advanced\nexpositions that follow in this volume. No prior knowledge of plasma physics is\nrequired, but the reader is assumed to be familiar with basic electrodynamics\nand fluid mechanics.\n",
"title": "Introduction to Plasma Physics"
} | null | null | null | null | true | null | 1039 | null | Default | null | null |
null | {
"abstract": " In this paper, we extend the Atiyah--Guillemin--Sternberg convexity theorem\nand Delzant's classification of symplectic toric manifolds to presymplectic\nmanifolds. We also define and study the Morita equivalence of presymplectic\ntoric manifolds and of their corresponding framed momentum polytopes, which may\nbe rational or non-rational. Toric orbifolds, quasifolds and non-commutative\ntoric varieties may be viewed as the quotient of our presymplectic toric\nmanifolds by the kernel isotropy foliation of the presymplectic form.\n",
"title": "Presymplectic convexity and (ir)rational polytopes"
} | null | null | null | null | true | null | 1040 | null | Default | null | null |
null | {
"abstract": " This paper is concerned with learning of mixture regression models for\nindividuals that are measured repeatedly. The adjective \"unsupervised\" implies\nthat the number of mixing components is unknown and has to be determined,\nideally by data driven tools. For this purpose, a novel penalized method is\nproposed to simultaneously select the number of mixing components and to\nestimate the mixing proportions and unknown parameters in the models. The\nproposed method is capable of handling both continuous and discrete responses\nby only requiring the first two moment conditions of the model distribution. It\nis shown to be consistent in both selecting the number of components and\nestimating the mixing proportions and unknown regression parameters. Further, a\nmodified EM algorithm is developed to seamlessly integrate model selection and\nestimation. Simulation studies are conducted to evaluate the finite sample\nperformance of the proposed procedure. And it is further illustrated via an\nanalysis of a primary biliary cirrhosis data set.\n",
"title": "Unsupervised Learning of Mixture Regression Models for Longitudinal Data"
} | null | null | null | null | true | null | 1041 | null | Default | null | null |
null | {
"abstract": " By the certain macroscopic perturbations in condensed matter anomalous\nelectron wells can be formed due to a local reduction of electromagnetic zero\npoint energy. These wells are narrow, of the width $\\sim 10^{-11}cm$, and with\nthe depth $\\sim 1MeV$. Such anomalous states, from the formal standpoint of\nquantum mechanics, correspond to a singular solution of a wave equation\nproduced by the non-physical $\\delta(\\vec R)$ source. The resolution, on the\nlevel of the Standard Model, of the tiny region around the formal singularity\nshows that the state is physical. The creation of those states in an atomic\nsystem is of the formal probability $\\exp(-1000)$. The probability becomes not\nsmall under a perturbation which rapidly varies in space, on the scale\n$10^{-11}cm$. In condensed matter such perturbation may relate to acoustic\nshock waves. In this process the short scale is the length of the standing de\nBroglie wave of a reflected lattice atom. Under electron transitions in the\nanomalous well (anomalous atom) $keV$ X-rays are expected to be emitted. A\nmacroscopic amount of anomalous atoms, of the size $10^{-11}cm$ each, can be\nformed in a solid resulting in ${\\it collapsed}$ ${\\it matter}$ with $10^9$\ntimes enhanced density.\n",
"title": "Anomalous electron states"
} | null | null | null | null | true | null | 1042 | null | Default | null | null |
null | {
"abstract": " Light traveling through the vacuum interacts with virtual particles similarly\nto the way that light traveling through a dielectric interacts with ordinary\nmatter. And just as the permittivity of a dielectric can be calculated, the\npermittivity $\\epsilon_0$ of the vacuum can be calculated, yielding an equation\nfor the fine-structure constant $\\alpha$. The most important contributions to\nthe value of $\\alpha$ arise from interactions in the vacuum of photons with\nvirtual, bound states of charged lepton-antilepton pairs. Considering only\nthese contributions, the fully screened $\\alpha \\cong 1/(8^2\\sqrt{3\\pi/2})\n\\cong 1/139$.\n",
"title": "Theoretical calculation of the fine-structure constant and the permittivity of the vacuum"
} | null | null | null | null | true | null | 1043 | null | Default | null | null |
null | {
"abstract": " Many asteroid databases with lightcurve brightness measurements (e.g. WISE,\nPan-STARRS1) contain enormous amounts of data for asteroid shape and spin\nmodelling. While lightcurve inversion is not plausible for individual targets\nwith scarce data, it is possible for large populations with thousands of\nasteroids, where the distributions of the shape and spin characteristics of the\npopulations are obtainable.\nWe aim to introduce a software implementation of a method that computes the\njoint shape elongation p and spin latitude beta distributions for a population,\nwith the brightness observations given in an asteroid database. Other main\ngoals are to include a method for performing validity checks of the algorithm,\nand a tool for a statistical comparison of populations.\nThe LEADER software package read the brightness measurement data for a\nuser-defined subpopulation from a given database. The observations were used to\ncompute estimates of the brightness variations of the population members. A\ncumulative distribution function (CDF) was constructed of these estimates. A\nsuperposition of known analytical basis functions yielded this CDF as a\nfunction of the (shape, spin) distribution. The joint distribution can be\nreconstructed by solving a linear constrained inverse problem. To test the\nvalidity of the method, the algorithm can be run with synthetic asteroid\nmodels, where the shape and spin characteristics are known, and by using the\ngeometries taken from the examined database.\nLEADER is a fast and robust software package for solving shape and spin\ndistributions for large populations. There are major differences in the quality\nand coverage of measurements depending on the database used, so synthetic\nsimulations are always necessary before a database can be reliably used. We\nshow examples of differences in the results when switching to another database.\n",
"title": "LEADER: fast estimates of asteroid shape elongation and spin latitude distributions from scarce photometry"
} | null | null | null | null | true | null | 1044 | null | Default | null | null |
null | {
"abstract": " We present the calibrated-projection MATLAB package implementing the method\nto construct confidence intervals proposed by Kaido, Molinari and Stoye (2017).\nThis manual provides details on how to use the package for inference on\nprojections of partially identified parameters. It also explains how to use the\nMATLAB functions we developed to compute confidence intervals on solutions of\nnonlinear optimization problems with estimated constraints.\n",
"title": "Calibrated Projection in MATLAB: Users' Manual"
} | null | null | null | null | true | null | 1045 | null | Default | null | null |
null | {
"abstract": " We use an atomic fountain clock to measure quantum scattering phase shifts\nprecisely through a series of narrow, low-field Feshbach resonances at average\ncollision energies below $1\\,\\mu$K. Our low spread in collision energy yields\nphase variations of order $\\pm \\pi/2$ for target atoms in several $F,m_F$\nstates. We compare them to a theoretical model and establish the accuracy of\nthe measurements and the theoretical uncertainties from the fitted potential.\nWe find overall excellent agreement, with small statistically significant\ndifferences that remain unexplained.\n",
"title": "Atomic Clock Measurements of Quantum Scattering Phase Shifts Spanning Feshbach Resonances at Ultralow Fields"
} | null | null | null | null | true | null | 1046 | null | Default | null | null |
null | {
"abstract": " A quantitative understanding of how sensory signals are transformed into\nmotor outputs places useful constraints on brain function and helps reveal the\nbrain's underlying computations. We investigate how the nematode C. elegans\nresponds to time-varying mechanosensory signals using a high-throughput\noptogenetic assay and automated behavior quantification. In the prevailing\npicture of the touch circuit, the animal's behavior is determined by which\nneurons are stimulated and by the stimulus amplitude. In contrast, we find that\nthe behavioral response is tuned to temporal properties of mechanosensory\nsignals, like its integral and derivative, that extend over many seconds.\nMechanosensory signals, even in the same neurons, can be tailored to elicit\ndifferent behavioral responses. Moreover, we find that the animal's response\nalso depends on its behavioral context. Most dramatically, the animal ignores\nall tested mechanosensory stimuli during turns. Finally, we present a\nlinear-nonlinear model that predicts the animal's behavioral response to\nstimulus.\n",
"title": "Temporal processing and context dependency in C. elegans mechanosensation"
} | null | null | null | null | true | null | 1047 | null | Default | null | null |
null | {
"abstract": " It has been argued in [EPL {\\bf 90} (2010) 50004], entitled {\\it Essential\ndiscreteness in generalized thermostatistics with non-logarithmic entropy},\nthat \"continuous Hamiltonian systems with long-range interactions and the\nso-called q-Gaussian momentum distributions are seen to be outside the scope of\nnon-extensive statistical mechanics\". The arguments are clever and appealing.\nWe show here that, however, some mathematical subtleties render them\nunconvincing\n",
"title": "On the putative essential discreteness of q-generalized entropies"
} | null | null | null | null | true | null | 1048 | null | Default | null | null |
null | {
"abstract": " We estimate the spin distribution of primordial black holes based on the\nrecent study of the critical phenomena in the gravitational collapse of a\nrotating radiation fluid. We find that primordial black holes are mostly slowly\nrotating.\n",
"title": "Spin Distribution of Primordial Black Holes"
} | null | null | null | null | true | null | 1049 | null | Default | null | null |
null | {
"abstract": " Deep convolutional neural networks (CNN) based solutions are the current\nstate- of-the-art for computer vision tasks. Due to the large size of these\nmodels, they are typically run on clusters of CPUs or GPUs. However, power\nrequirements and cost budgets can be a major hindrance in adoption of CNN for\nIoT applications. Recent research highlights that CNN contain significant\nredundancy in their structure and can be quantized to lower bit-width\nparameters and activations, while maintaining acceptable accuracy. Low\nbit-width and especially single bit-width (binary) CNN are particularly\nsuitable for mobile applications based on FPGA implementation, due to the\nbitwise logic operations involved in binarized CNN. Moreover, the transition to\nlower bit-widths opens new avenues for performance optimizations and model\nimprovement. In this paper, we present an automatic flow from trained\nTensorFlow models to FPGA system on chip implementation of binarized CNN. This\nflow involves quantization of model parameters and activations, generation of\nnetwork and model in embedded-C, followed by automatic generation of the FPGA\naccelerator for binary convolutions. The automated flow is demonstrated through\nimplementation of binarized \"YOLOV2\" on the low cost, low power Cyclone- V FPGA\ndevice. Experiments on object detection using binarized YOLOV2 demonstrate\nsignificant performance benefit in terms of model size and inference speed on\nFPGA as compared to CPU and mobile CPU platforms. Furthermore, the entire\nautomated flow from trained models to FPGA synthesis can be completed within\none hour.\n",
"title": "Automated flow for compressing convolution neural networks for efficient edge-computation with FPGA"
} | null | null | null | null | true | null | 1050 | null | Default | null | null |
null | {
"abstract": " Objective: to establish an algorithmic framework and a benchmark dataset for\ncomparing methods of pulse rate estimation using imaging photoplethysmography\n(iPPG). Approach: first we reveal essential steps of pulse rate estimation from\nfacial video and review methods applied at each of the steps. Then we\ninvestigate performance of these methods for DEAP dataset\nwww.eecs.qmul.ac.uk/mmv/datasets/deap/ containing facial videos and reference\ncontact photoplethysmograms. Main results: best assessment precision is\nachieved when pulse rate is estimated using continuous wavelet transform from\niPPG extracted by the POS method (overall mean absolute error below 2 heart\nbeats per minute). Significance: we provide a generic framework for theoretical\ncomparison of methods for pulse rate estimation from iPPG and report results\nfor the most popular methods on a publicly available dataset that can be used\nas a benchmark.\n",
"title": "Pulse rate estimation using imaging photoplethysmography: generic framework and comparison of methods on a publicly available dataset"
} | null | null | null | null | true | null | 1051 | null | Default | null | null |
null | {
"abstract": " Convolutional neural networks have recently demonstrated high-quality\nreconstruction for single-image super-resolution. In this paper, we propose the\nLaplacian Pyramid Super-Resolution Network (LapSRN) to progressively\nreconstruct the sub-band residuals of high-resolution images. At each pyramid\nlevel, our model takes coarse-resolution feature maps as input, predicts the\nhigh-frequency residuals, and uses transposed convolutions for upsampling to\nthe finer level. Our method does not require the bicubic interpolation as the\npre-processing step and thus dramatically reduces the computational complexity.\nWe train the proposed LapSRN with deep supervision using a robust Charbonnier\nloss function and achieve high-quality reconstruction. Furthermore, our network\ngenerates multi-scale predictions in one feed-forward pass through the\nprogressive reconstruction, thereby facilitates resource-aware applications.\nExtensive quantitative and qualitative evaluations on benchmark datasets show\nthat the proposed algorithm performs favorably against the state-of-the-art\nmethods in terms of speed and accuracy.\n",
"title": "Deep Laplacian Pyramid Networks for Fast and Accurate Super-Resolution"
} | null | null | [
"Computer Science"
]
| null | true | null | 1052 | null | Validated | null | null |
null | {
"abstract": " Compute the coarsest simulation preorder included in an initial preorder is\nused to reduce the resources needed to analyze a given transition system. This\ntechnique is applied on many models like Kripke structures, labeled graphs,\nlabeled transition systems or even word and tree automata. Let (Q,\n$\\rightarrow$) be a given transition system and Rinit be an initial preorder\nover Q. Until now, algorithms to compute Rsim , the coarsest simulation\nincluded in Rinit , are either memory efficient or time efficient but not both.\nIn this paper we propose the foundation for a series of efficient simulation\nalgorithms with the introduction of the notion of maximal transitions and the\nnotion of stability of a preorder with respect to a coarser one. As an\nillustration we solve an open problem by providing the first algorithm with the\nbest published time complexity, O(|Psim |.|$\\rightarrow$|), and a bit space\ncomplexity in O(|Psim |^2. log(|Psim |) + |Q|. log(|Q|)), with Psim the\npartition induced by Rsim.\n",
"title": "Foundation for a series of efficient simulation algorithms"
} | null | null | null | null | true | null | 1053 | null | Default | null | null |
null | {
"abstract": " A principle on the macroscopic motion of systems in thermodynamic\nequilibrium, rarely discussed in texts, is reviewed: Very small but still\nmacroscopic parts of a fully isolated system in thermal equilibrium move as if\npoints of a rigid body, macroscopic energy being dissipated to increase\ninternal energy, and increase entropy along. It appears particularly important\nin Space physics, when dissipation involves long-range fields at\nElectromagnetism and Gravitation, rather than short-range contact forces. It is\nshown how new physics, Special Relativity as regards Electromagnetism, first\nNewtonian theory then General Relativity as regards Gravitation, determine\ndifferent dissipative processes involved in the approach to that equilibrium.\n",
"title": "A Review of Macroscopic Motion in Thermodynamic Equilibrium"
} | null | null | null | null | true | null | 1054 | null | Default | null | null |
null | {
"abstract": " CaFe2As2 exhibits collapsed tetragonal (cT) structure and varied exotic\nbehavior under pressure at low temperatures that led to debate on linking the\nstructural changes to its exceptional electronic properties like\nsuperconductivity, magnetism, etc. Here, we investigate the electronic\nstructure of CaFe2As2 forming in different structures employing density\nfunctional theory. The results indicate better stability of the cT phase with\nenhancement in hybridization induced effects and shift of the energy bands\ntowards lower energies. The Fermi surface centered around $\\Gamma$ point\ngradually vanishes with the increase in pressure. Consequently, the nesting\nbetween the hole and electron Fermi surfaces associated to the spin density\nwave state disappears indicating a pathway to achieve the proximity to quantum\nfluctuations. The magnetic moment at the Fe sites diminishes in the cT phase\nconsistent with the magnetic susceptibility results. Notably, the hybridization\nof Ca 4s states (Ca-layer may be treated as a charge reservoir layer akin to\nthose in cuprate superconductors) is significantly enhanced in the cT phase\nrevealing its relevance in its interesting electronic properties.\n",
"title": "Emergent electronic structure of CaFe2As2"
} | null | null | null | null | true | null | 1055 | null | Default | null | null |
null | {
"abstract": " We study a mathematical model of cell populations dynamics proposed by M.\nRotenberg and investigated by M. Boulanouar. Here, a cell is characterized by\nher maturity and speed of maturation. The growth of cell populations is\ndescribed by a partial differential equation with a boundary condition. In the\nfirst part of the paper we exploit semigroup theory approach and apply Lord\nKelvin's method of images in order to give a new proof that the model is well\nposed. Next, we use a semi-explicit formula for the semigroup related to the\nmodel obtained by the method of images in order to give growth estimates for\nthe semigroup. The main part of the paper is devoted to the asymptotic\nbehaviour of the semigroup. We formulate conditions for the asymptotic\nstability of the semigroup in the case in which the average number of viable\ndaughters per mitosis equals one. To this end we use methods developed by K.\nPichór and R. Rudnicki.\n",
"title": "Lord Kelvin's method of images approach to the Rotenberg model and its asymptotics"
} | null | null | [
"Mathematics"
]
| null | true | null | 1056 | null | Validated | null | null |
null | {
"abstract": " The extraction system of CSNS mainly consists of two kinds of magnets: eight\nkickers and one lambertson magnet. In this paper, firstly, the magnetic test\nresults of the eight kickers were introduced and then the filed uniformity and\nmagnetizing relationship of the kickers were given. Secondly, during the beam\ncommissioning in the future, in order to obtain more accurate magnetizing\nrelationship, a new method to measure the magnetizing coefficients of the\nkickers by the real extraction beam was given and the data analysis would also\nbe processed.\n",
"title": "Study of the Magnetizing Relationship of the Kickers for CSNS"
} | null | null | null | null | true | null | 1057 | null | Default | null | null |
null | {
"abstract": " Many real-world analytics problems involve two significant challenges:\nprediction and optimization. Due to the typically complex nature of each\nchallenge, the standard paradigm is to predict, then optimize. By and large,\nmachine learning tools are intended to minimize prediction error and do not\naccount for how the predictions will be used in a downstream optimization\nproblem. In contrast, we propose a new and very general framework, called Smart\n\"Predict, then Optimize\" (SPO), which directly leverages the optimization\nproblem structure, i.e., its objective and constraints, for designing\nsuccessful analytics tools. A key component of our framework is the SPO loss\nfunction, which measures the quality of a prediction by comparing the objective\nvalues of the solutions generated using the predicted and observed parameters,\nrespectively. Training a model with respect to the SPO loss is computationally\nchallenging, and therefore we also develop a surrogate loss function, called\nthe SPO+ loss, which upper bounds the SPO loss, has desirable convexity\nproperties, and is statistically consistent under mild conditions. We also\npropose a stochastic gradient descent algorithm which allows for situations in\nwhich the number of training samples is large, model regularization is desired,\nand/or the optimization problem of interest is nonlinear or integer. Finally,\nwe perform computational experiments to empirically verify the success of our\nSPO framework in comparison to the standard predict-then-optimize approach.\n",
"title": "Smart \"Predict, then Optimize\""
} | null | null | null | null | true | null | 1058 | null | Default | null | null |
null | {
"abstract": " Novel data acquisition schemes have been an emerging need for scanning\nmicroscopy based imaging techniques to reduce the time in data acquisition and\nto minimize probing radiation in sample exposure. Varies sparse sampling\nschemes have been studied and are ideally suited for such applications where\nthe images can be reconstructed from a sparse set of measurements. Dynamic\nsparse sampling methods, particularly supervised learning based iterative\nsampling algorithms, have shown promising results for sampling pixel locations\non the edges or boundaries during imaging. However, dynamic sampling for\nimaging skeleton-like objects such as metal dendrites remains difficult. Here,\nwe address a new unsupervised learning approach using Hierarchical Gaussian\nMixture Mod- els (HGMM) to dynamically sample metal dendrites. This technique\nis very useful if the users are interested in fast imaging the primary and\nsecondary arms of metal dendrites in solidification process in materials\nscience.\n",
"title": "U-SLADS: Unsupervised Learning Approach for Dynamic Dendrite Sampling"
} | null | null | null | null | true | null | 1059 | null | Default | null | null |
null | {
"abstract": " We consider a registration-based approach for localizing sensor networks from\nrange measurements. This is based on the assumption that one can find\noverlapping cliques spanning the network. That is, for each sensor, one can\nidentify geometric neighbors for which all inter-sensor ranges are known. Such\ncliques can be efficiently localized using multidimensional scaling. However,\nsince each clique is localized in some local coordinate system, we are required\nto register them in a global coordinate system. In other words, our approach is\nbased on transforming the localization problem into a problem of registration.\nIn this context, the main contributions are as follows. First, we describe an\nefficient method for partitioning the network into overlapping cliques. Second,\nwe study the problem of registering the localized cliques, and formulate a\nnecessary rigidity condition for uniquely recovering the global sensor\ncoordinates. In particular, we present a method for efficiently testing\nrigidity, and a proposal for augmenting the partitioned network to enforce\nrigidity. A recently proposed semidefinite relaxation of global registration is\nused for registering the cliques. We present simulation results on random and\nstructured sensor networks to demonstrate that the proposed method compares\nfavourably with state-of-the-art methods in terms of run-time, accuracy, and\nscalability.\n",
"title": "On a registration-based approach to sensor network localization"
} | null | null | null | null | true | null | 1060 | null | Default | null | null |
null | {
"abstract": " How might a smooth probability distribution be estimated, with accurately\nquantified uncertainty, from a limited amount of sampled data? Here we describe\na field-theoretic approach that addresses this problem remarkably well in one\ndimension, providing an exact nonparametric Bayesian posterior without relying\non tunable parameters or large-data approximations. Strong non-Gaussian\nconstraints, which require a non-perturbative treatment, are found to play a\nmajor role in reducing distribution uncertainty. A software implementation of\nthis method is provided.\n",
"title": "Density estimation on small datasets"
} | null | null | [
"Computer Science",
"Quantitative Biology"
]
| null | true | null | 1061 | null | Validated | null | null |
null | {
"abstract": " In the context of commutative differential graded algebras over $\\mathbb Q$,\nwe show that an iteration of \"odd spherical fibration\" creates a \"total space\"\ncommutative differential graded algebra with only odd degree cohomology. Then\nwe show for such a commutative differential graded algebra that, for any of its\n\"fibrations\" with \"fiber\" of finite cohomological dimension, the induced map on\ncohomology is injective.\n",
"title": "Generalized Euler classes, differential forms and commutative DGAs"
} | null | null | null | null | true | null | 1062 | null | Default | null | null |
null | {
"abstract": " Both the human brain and artificial learning agents operating in real-world\nor comparably complex environments are faced with the challenge of online model\nselection. In principle this challenge can be overcome: hierarchical Bayesian\ninference provides a principled method for model selection and it converges on\nthe same posterior for both off-line (i.e. batch) and online learning. However,\nmaintaining a parameter posterior for each model in parallel has in general an\neven higher memory cost than storing the entire data set and is consequently\nclearly unfeasible. Alternatively, maintaining only a limited set of models in\nmemory could limit memory requirements. However, sufficient statistics for one\nmodel will usually be insufficient for fitting a different kind of model,\nmeaning that the agent loses information with each model change. We propose\nthat episodic memory can circumvent the challenge of limited memory-capacity\nonline model selection by retaining a selected subset of data points. We design\na method to compute the quantities necessary for model selection even when the\ndata is discarded and only statistics of one (or few) learnt models are\navailable. We demonstrate on a simple model that a limited-sized episodic\nmemory buffer, when the content is optimised to retain data with statistics not\nmatching the current representation, can resolve the fundamental challenge of\nonline model selection.\n",
"title": "Episodic memory for continual model learning"
} | null | null | [
"Computer Science",
"Statistics"
]
| null | true | null | 1063 | null | Validated | null | null |
null | {
"abstract": " Fifth Generation (5G) telecommunication system is going to deliver a flexible\nradio access network (RAN). Security functions such as authorization,\nauthentication and accounting (AAA) are expected to be distributed from central\nclouds to edge clouds. We propose a novel architectural security solution that\napplies to 5G networks. It is called Trust Zone (TZ) that is designed as an\nenhancement of the 5G AAA in the edge cloud. TZ also provides an autonomous and\ndecentralized security policy for different tenants under variable network\nconditions. TZ also initiates an ability of disaster cognition and extends the\nsecurity functionalities to a set of flexible and highly available emergency\nservices in the edge cloud.\n",
"title": "Security Trust Zone in 5G Networks"
} | null | null | null | null | true | null | 1064 | null | Default | null | null |
null | {
"abstract": " Consider reconstructing a signal $x$ by minimizing a weighted sum of a convex\ndifferentiable negative log-likelihood (NLL) (data-fidelity) term and a convex\nregularization term that imposes a convex-set constraint on $x$ and enforces\nits sparsity using $\\ell_1$-norm analysis regularization. We compute upper\nbounds on the regularization tuning constant beyond which the regularization\nterm overwhelmingly dominates the NLL term so that the set of minimum points of\nthe objective function does not change. Necessary and sufficient conditions for\nirrelevance of sparse signal regularization and a condition for the existence\nof finite upper bounds are established. We formulate an optimization problem\nfor finding these bounds when the regularization term can be globally minimized\nby a feasible $x$ and also develop an alternating direction method of\nmultipliers (ADMM) type method for their computation. Simulation examples show\nthat the derived and empirical bounds match.\n",
"title": "Upper-Bounding the Regularization Constant for Convex Sparse Signal Reconstruction"
} | null | null | [
"Mathematics",
"Statistics"
]
| null | true | null | 1065 | null | Validated | null | null |
null | {
"abstract": " This document is a response to a report from the University of Melbourne on\nthe privacy of the Opal dataset release. The Opal dataset was released by\nData61 (CSIRO) in conjunction with the Transport for New South Wales (TfNSW).\nThe data consists of two separate weeks of \"tap-on/tap-off\" data of individuals\nwho used any of the four different modes of public transport from TfNSW: buses,\nlight rail, train and ferries. These taps are recorded through the smart\nticketing system, known as Opal, available in the state of New South Wales,\nAustralia.\n",
"title": "On the Privacy of the Opal Data Release: A Response"
} | null | null | [
"Computer Science"
]
| null | true | null | 1066 | null | Validated | null | null |
null | {
"abstract": " The stochastic Gross-Pitaevskii equation is used as a model to describe\nBose-Einstein condensation at positive temperature. The equation is a complex\nGinzburg Landau equation with a trapping potential and an additive space-time\nwhite noise. Two important questions for this system are the global existence\nof solutions in the support of the Gibbs measure, and the convergence of those\nsolutions to the equilibrium for large time. In this paper, we give a proof of\nthese two results in one space dimension. In order to prove the convergence to\nequilibrium, we use the associated purely dissipative equation as an auxiliary\nequation, for which the convergence may be obtained using standard techniques.\nGlobal existence is obtained for all initial data, and not almost surely with\nrespect to the invariant measure.\n",
"title": "Long time behavior of Gross-Pitaevskii equation at positive temperature"
} | null | null | null | null | true | null | 1067 | null | Default | null | null |
null | {
"abstract": " Let $\\theta, \\theta'$ be irrational numbers and $A, B$ be matrices in\n$SL_2(\\mathbb{Z})$ of infinite order. We compute the $K$-theory of the crossed\nproduct $\\mathcal{A}_{\\theta}\\rtimes_A \\mathbb{Z}$ and show that\n$\\mathcal{A}_{\\theta} \\rtimes_A\\mathbb{Z}$ and $\\mathcal{A}_{\\theta'} \\rtimes_B\n\\mathbb{Z}$ are $*$-isomorphic if and only if $\\theta = \\pm\\theta'\n\\pmod{\\mathbb{Z}}$ and $I-A^{-1}$ is matrix equivalent to $I-B^{-1}$. Combining\nthis result and an explicit construction of equivariant bimodules, we show that\n$\\mathcal{A}_{\\theta} \\rtimes_A\\mathbb{Z}$ and $\\mathcal{A}_{\\theta'} \\rtimes_B\n\\mathbb{Z}$ are Morita equivalent if and only if $\\theta$ and $\\theta'$ are in\nthe same $GL_2(\\mathbb{Z})$ orbit and $I-A^{-1}$ is matrix equivalent to\n$I-B^{-1}$. Finally, we determine the Morita equivalence class of\n$\\mathcal{A}_{\\theta} \\rtimes F$ for any finite subgroup $F$ of\n$SL_2(\\mathbb{Z})$.\n",
"title": "Isomorphism and Morita equivalence classes for crossed products of irrational rotation algebras by cyclic subgroups of $SL_2(\\mathbb{Z})$"
} | null | null | null | null | true | null | 1068 | null | Default | null | null |
null | {
"abstract": " This paper proposes a new convex model predictive control strategy for\ndynamic optimal power flow between battery energy storage systems distributed\nin an AC microgrid. The proposed control strategy uses a new problem\nformulation, based on a linear d-q reference frame voltage-current model and\nlinearised power flow approximations. This allows the optimal power flows to be\nsolved as a convex optimisation problem, for which fast and robust solvers\nexist. The proposed method does not assume real and reactive power flows are\ndecoupled, allowing line losses, voltage constraints and converter current\nconstraints to be addressed. In addition, non-linear variations in the charge\nand discharge efficiencies of lithium ion batteries are analysed and included\nin the control strategy. Real-time digital simulations were carried out for an\nislanded microgrid based on the IEEE 13 bus prototypical feeder, with\ndistributed battery energy storage systems and intermittent photovoltaic\ngeneration. It is shown that the proposed control strategy approaches the\nperformance of a strategy based on non-convex optimisation, while reducing the\nrequired computation time by a factor of 1000, making it suitable for a\nreal-time model predictive control implementation.\n",
"title": "Model Predictive Control for Distributed Microgrid Battery Energy Storage Systems"
} | null | null | null | null | true | null | 1069 | null | Default | null | null |
null | {
"abstract": " We unveil the geometric nature of the multiplet of fundamental fermions in\nthe Standard Model of fundamental particles as a noncommutative analogue of de\nRham forms on the internal finite quantum space.\n",
"title": "On noncommutative geometry of the Standard Model: fermion multiplet as internal forms"
} | null | null | null | null | true | null | 1070 | null | Default | null | null |
null | {
"abstract": " We present a selective review of statistical modeling of dynamic networks. We\nfocus on models with latent variables, specifically, the latent space models\nand the latent class models (or stochastic blockmodels), which investigate both\nthe observed features and the unobserved structure of networks. We begin with\nan overview of the static models, and then we introduce the dynamic extensions.\nFor each dynamic model, we also discuss its applications that have been studied\nin the literature, with the data source listed in Appendix. Based on the\nreview, we summarize a list of open problems and challenges in dynamic network\nmodeling with latent variables.\n",
"title": "A Review of Dynamic Network Models with Latent Variables"
} | null | null | [
"Statistics"
]
| null | true | null | 1071 | null | Validated | null | null |
null | {
"abstract": " Pipelines combining SQL-style business intelligence (BI) queries and linear\nalgebra (LA) are becoming increasingly common in industry. As a result, there\nis a growing need to unify these workloads in a single framework.\nUnfortunately, existing solutions either sacrifice the inherent benefits of\nexclusively using a relational database (e.g. logical and physical\nindependence) or incur orders of magnitude performance gaps compared to\nspecialized engines (or both). In this work we study applying a new type of\nquery processing architecture to standard BI and LA benchmarks. To do this we\npresent a new in-memory query processing engine called LevelHeaded. LevelHeaded\nuses worst-case optimal joins as its core execution mechanism for both BI and\nLA queries. With LevelHeaded, we show how crucial optimizations for BI and LA\nqueries can be captured in a worst-case optimal query architecture. Using these\noptimizations, LevelHeaded outperforms other relational database engines\n(LogicBlox, MonetDB, and HyPer) by orders of magnitude on standard LA\nbenchmarks, while performing on average within 31% of the best-of-breed BI\n(HyPer) and LA (Intel MKL) solutions on their own benchmarks. Our results show\nthat such a single query processing architecture is capable of delivering\ncompetitive performance on both BI and LA queries.\n",
"title": "LevelHeaded: Making Worst-Case Optimal Joins Work in the Common Case"
} | null | null | null | null | true | null | 1072 | null | Default | null | null |
null | {
"abstract": " In this paper, we propose a simple but effective method for training neural\nnetworks with a limited amount of training data. Our approach inherits the idea\nof knowledge distillation that transfers knowledge from a deep or wide\nreference model to a shallow or narrow target model. The proposed method\nemploys this idea to mimic predictions of reference estimators that are more\nrobust against overfitting than the network we want to train. Different from\nalmost all the previous work for knowledge distillation that requires a large\namount of labeled training data, the proposed method requires only a small\namount of training data. Instead, we introduce pseudo training examples that\nare optimized as a part of model parameters. Experimental results for several\nbenchmark datasets demonstrate that the proposed method outperformed all the\nother baselines, such as naive training of the target model and standard\nknowledge distillation.\n",
"title": "Few-shot learning of neural networks from scratch by pseudo example optimization"
} | null | null | null | null | true | null | 1073 | null | Default | null | null |
null | {
"abstract": " In this paper, we investigate the umbral representation of the Fubini\npolynomials $F_{x}^{n}:=F_{n}(x)$ to derive some properties involving these\npolynomials. For any prime number $p$ and any polynomial $f$ with integer\ncoefficients, we show $(f(F_{x}))^{p}\\equiv f(F_{x})$ and we give other curious\ncongruences.\n",
"title": "Identities and congruences involving the Fubini polynomials"
} | null | null | null | null | true | null | 1074 | null | Default | null | null |
null | {
"abstract": " In this paper, a brief review of delay population models and their\napplications in ecology is provided. The inclusion of diffusion and nonlocality\nterms in delay models has given more capabilities to these models enabling them\nto capture several ecological phenomena such as the Allee effect, waves of\ninvasive species and spatio-temporal competitions of interacting species.\nMoreover, recent advances in the studies of traveling and stationary wave\nsolutions of delay models are outlined. In particular, the existence of\nstationary and traveling wave solutions of delay models, stability of wave\nsolutions, formation of wavefronts in the special domain, and possible outcomes\nof delay models are discussed.\n",
"title": "Introduction to Delay Models and Their Wave Solutions"
} | null | null | null | null | true | null | 1075 | null | Default | null | null |
null | {
"abstract": " I show that propositional intuitionistic logic is complete with respect to an\nadaptation of Dummett's pragmatist justification procedure. In particular,\ngiven a pragmatist justification of an argument, I show how to obtain a natural\ndeduction derivation of the conclusion of the argument from, at most, the same\nassumptions.\n",
"title": "On Dummett's Pragmatist Justification Procedure"
} | null | null | null | null | true | null | 1076 | null | Default | null | null |
null | {
"abstract": " We present a newly discovered correlation between the wind outflow velocity\nand the X-ray luminosity in the luminous ($L_{\\rm bol}\\sim10^{47}\\,\\rm\nerg\\,s^{-1}$) nearby ($z=0.184$) quasar PDS\\,456. All the contemporary\nXMM-Newton, NuSTAR and Suzaku observations from 2001--2014 were revisited and\nwe find that the centroid energy of the blueshifted Fe\\,K absorption profile\nincreases with luminosity. This translates into a correlation between the wind\noutflow velocity and the hard X-ray luminosity (between 7--30\\,keV) where we\nfind that $v_{\\rm w}/c \\propto L_{7-30}^{\\gamma}$ where $\\gamma=0.22\\pm0.04$.\nWe also show that this is consistent with a wind that is predominately\nradiatively driven, possibly resulting from the high Eddington ratio of\nPDS\\,456.\n",
"title": "Evidence for a radiatively driven disc-wind in PDS 456?"
} | null | null | [
"Physics"
]
| null | true | null | 1077 | null | Validated | null | null |
null | {
"abstract": " Plumbene, similar to silicene, has a buckled honeycomb structure with a large\nband gap ($\\sim 400$ meV). All previous studies have shown that it is a normal\ninsulator. Here, we perform first-principles calculations and employ a\nsixteen-band tight-binding model with nearest-neighbor and\nnext-nearest-neighbor hopping terms to investigate electronic structures and\ntopological properties of the plumbene monolayer. We find that it can become a\ntopological insulator with a large bulk gap ($\\sim 200$ meV) through electron\ndoping, and the nontrivial state is very robust with respect to external\nstrain. Plumbene can be an ideal candidate for realizing the quantum spin Hall\neffect at room temperature. By investigating effects of external electric and\nmagnetic fields on electronic structures and transport properties of plumbene,\nwe present two rich phase diagrams with and without electron doping, and\npropose a theoretical design for a four-state spin-valley filter.\n",
"title": "From a normal insulator to a topological insulator in plumbene"
} | null | null | null | null | true | null | 1078 | null | Default | null | null |
null | {
"abstract": " Providing a background discrimination tool is crucial for enhancing the\nsensitivity of next-generation experiments searching for neutrinoless double-\nbeta decay. The development of high-sensitivity (< 20 eV RMS) cryogenic light\ndetectors allows simultaneous read-out of the light and heat signals and\nenables background suppression through particle identification. The Cryogenic\nwide- Area Light Detector with Excellent Resolution (CALDER) R&D already proved\nthe potential of this technique using the phonon-mediated Kinetic Inductance\nDetectors (KIDs) approach. The first array prototype with 4 Aluminum KIDs on a\n2 $\\times$ 2 cm2 Silicon substrate showed a baseline resolution of 154 $\\pm$ 7\neV RMS. Improving the design and the readout of the resonator, the next CALDER\nprototype featured an energy resolution of 82 $\\pm$ 4 eV, by sampling the same\nsubstrate with a single Aluminum KID.\n",
"title": "High-sensitivity Kinetic Inductance Detectors for CALDER"
} | null | null | null | null | true | null | 1079 | null | Default | null | null |
null | {
"abstract": " We obtain upper bounds on the composition length of a finite permutation\ngroup in terms of the degree and the number of orbits, and analogous bounds for\nprimitive, quasiprimitive and semiprimitive groups. Similarly, we obtain upper\nbounds on the composition length of a finite completely reducible linear group\nin terms of some of its parameters. In almost all cases we show that the bounds\nare sharp, and describe the extremal examples.\n",
"title": "Bounding the composition length of primitive permutation groups and completely reducible linear groups"
} | null | null | null | null | true | null | 1080 | null | Default | null | null |
null | {
"abstract": " In this article we present a Bernstein inequality for sums of random\nvariables which are defined on a spatial lattice structure. The inequality can\nbe used to derive concentration inequalities. It can be useful to obtain\nconsistency properties for nonparametric estimators of conditional expectation\nfunctions.\n",
"title": "A Bernstein Inequality For Spatial Lattice Processes"
} | null | null | null | null | true | null | 1081 | null | Default | null | null |
null | {
"abstract": " We explore different approaches to integrating a simple convolutional neural\nnetwork (CNN) with the Lucene search engine in a multi-stage ranking\narchitecture. Our models are trained using the PyTorch deep learning toolkit,\nwhich is implemented in C/C++ with a Python frontend. One obvious integration\nstrategy is to expose the neural network directly as a service. For this, we\nuse Apache Thrift, a software framework for building scalable cross-language\nservices. In exploring alternative architectures, we observe that once trained,\nthe feedforward evaluation of neural networks is quite straightforward.\nTherefore, we can extract the parameters of a trained CNN from PyTorch and\nimport the model into Java, taking advantage of the Java Deeplearning4J library\nfor feedforward evaluation. This has the advantage that the entire end-to-end\nsystem can be implemented in Java. As a third approach, we can extract the\nneural network from PyTorch and \"compile\" it into a C++ program that exposes a\nThrift service. We evaluate these alternatives in terms of performance (latency\nand throughput) as well as ease of integration. Experiments show that\nfeedforward evaluation of the convolutional neural network is significantly\nslower in Java, while the performance of the compiled C++ network does not\nconsistently beat the PyTorch implementation.\n",
"title": "An Exploration of Approaches to Integrating Neural Reranking Models in Multi-Stage Ranking Architectures"
} | null | null | null | null | true | null | 1082 | null | Default | null | null |
null | {
"abstract": " We study two dispersive regimes in the dynamics of $N$ two-level atoms\ninteracting with a bosonic mode for long interaction times. Firstly, we analyze\nthe dispersive multiqubit quantum Rabi model for the regime in which the qubit\nfrequencies are equal and smaller than the mode frequency, and for values of\nthe coupling strength similar or larger than the mode frequency, namely, the\ndeep strong coupling regime. Secondly, we address an interaction that is\ndependent on the photon number, where the coupling strength is comparable to\nthe geometric mean of the qubit and mode frequencies. We show that the\nassociated dynamics is analytically tractable and provide useful frameworks\nwith which to analyze the system behavior. In the deep strong coupling regime,\nwe unveil the structure of unexpected resonances for specific values of the\ncoupling, present for $N\\ge2$, and in the photon-number-dependent regime we\ndemonstrate that all the nontrivial dynamical behavior occurs in the atomic\ndegrees of freedom for a given Fock state. We verify these assertions with\nnumerical simulations of the qubit population and photon-statistic dynamics.\n",
"title": "Dispersive Regimes of the Dicke Model"
} | null | null | null | null | true | null | 1083 | null | Default | null | null |
null | {
"abstract": " We design and implement the first private and anonymous decentralized\ncrowdsourcing system ZebraLancer. It realizes the fair exchange (i.e. security\nagainst malicious workers and dishonest requesters) without using any\nthird-party arbiter. More importantly, it overcomes two fundamental challenges\nof decentralization, i.e. data leakage and identity breach.\nFirst, our outsource-then-prove methodology resolves the critical tension\nbetween blockchain transparency and data confidentiality without sacrificing\nthe fairness of exchange. ZebraLancer ensures: a requester will not pay more\nthan what data deserve, according to a policy announced when her task is\npublished through the blockchain; each worker indeed gets a payment based on\nthe policy, if submits data to the blockchain; the above properties are\nrealized not only without a central arbiter, but also without leaking the data\nto blockchain network.\nFurthermore, the blockchain transparency might allow one to infer private\ninformation of workers/requesters through their participation history.\nZebraLancer solves the problem by allowing anonymous participations without\nsurrendering user accountability. Specifically, workers cannot misuse anonymity\nto submit multiple times to reap rewards, and an anonymous requester cannot\nmaliciously submit colluded answers to herself to repudiate payments. The idea\nbehind is a subtle linkability: if one authenticates twice in a task, everybody\ncan tell, or else staying anonymous. To realize such delicate linkability, we\nput forth a novel cryptographic notion, the common-prefix-linkable anonymous\nauthentication.\nFinally, we implement our protocol for a common image annotation task and\ndeploy it in a test net of Ethereum. The experiment results show the\napplicability of our protocol and highlight subtleties of tailoring the\nprotocol to be compatible with the existing real-world open blockchain.\n",
"title": "ZebraLancer: Crowdsource Knowledge atop Open Blockchain, Privately and Anonymously"
} | null | null | [
"Computer Science"
]
| null | true | null | 1084 | null | Validated | null | null |
null | {
"abstract": " In this paper, we will show an unprecedented method to accelerate training\nand improve performance, which called random gradient (RG). This method can be\neasier to the training of any model without extra calculation cost, we use\nImage classification, Semantic segmentation, and GANs to confirm this method\ncan improve speed which is training model in computer vision. The central idea\nis using the loss multiplied by a random number to random reduce the\nback-propagation gradient. We can use this method to produce a better result in\nPascal VOC, Cifar, Cityscapes datasets.\n",
"title": "Fast, Better Training Trick -- Random Gradient"
} | null | null | null | null | true | null | 1085 | null | Default | null | null |
null | {
"abstract": " Background: As most of the software development organizations are\nmale-dominated, female developers encountering various negative workplace\nexperiences reported feeling like they \"do not belong\". Exposures to\ndiscriminatory expletives or negative critiques from their male colleagues may\nfurther exacerbate those feelings. Aims: The primary goal of this study is to\nidentify the differences in expressions of sentiments between male and female\ndevelopers during various software engineering tasks. Method: On this goal, we\nmined the code review repositories of six popular open source projects. We used\na semi-automated approach leveraging the name as well as multiple social\nnetworks to identify the gender of a developer. Using SentiSE, a customized and\nstate-of-the-art sentiment analysis tool for the software engineering domain,\nwe classify each communication as negative, positive, or neutral. We also\ncompute the frequencies of sentiment words, emoticons, and expletives used by\neach developer. Results: Our results suggest that the likelihood of using\nsentiment words, emoticons, and expletives during code reviews varies based on\nthe gender of a developer, as females are significantly less likely to express\nsentiments than males. Although female developers were more neutral to their\nmale colleagues than to another female, male developers from three out of the\nsix projects were not only writing more frequent negative comments but also\nwithholding positive encouragements from their female counterparts. Conclusion:\nOur results provide empirical evidence of another factor behind the negative\nwork place experiences encountered by the female developers that may be\ncontributing to the diminishing number of females in the SE industry.\n",
"title": "Expressions of Sentiments During Code Reviews: Male vs. Female"
} | null | null | [
"Computer Science"
]
| null | true | null | 1086 | null | Validated | null | null |
null | {
"abstract": " In this paper our aim is to present the completely monotonicity and convexity\nproperties for the Wright function. As consequences of these results, we\npresent some functional inequalities. Moreover, we derive the monotonicity and\nlog-convexity results for the generalized Wright functions. As applications, we\npresent several new inequalities (like Turán type inequalities) and we prove\nsome geometric properties for four--parametric Mittag--Leffler functions.\n",
"title": "Monotonicity patterns and functional inequalities for classical and generalized Wright functions"
} | null | null | [
"Mathematics"
]
| null | true | null | 1087 | null | Validated | null | null |
null | {
"abstract": " Despite the effectiveness of convolutional neural networks (CNNs) especially\nin image classification tasks, the effect of convolution features on learned\nrepresentations is still limited. It mostly focuses on the salient object of\nthe images, but ignores the variation information on clutter and local. In this\npaper, we propose a special framework, which is the multiple VLAD encoding\nmethod with the CNNs features for image classification. Furthermore, in order\nto improve the performance of the VLAD coding method, we explore the\nmultiplicity of VLAD encoding with the extension of three kinds of encoding\nalgorithms, which are the VLAD-SA method, the VLAD-LSA and the VLAD-LLC method.\nFinally, we equip the spatial pyramid patch (SPM) on VLAD encoding to add the\nspatial information of CNNs feature. In particular, the power of SPM leads our\nframework to yield better performance compared to the existing method.\n",
"title": "Multiple VLAD encoding of CNNs for image classification"
} | null | null | null | null | true | null | 1088 | null | Default | null | null |
null | {
"abstract": " Real and complex Clifford bundles and Dirac operators defined on them are\nconsidered. By using the index theorems of Dirac operators, table of\ntopological invariants is constructed from the Clifford chessboard. Through the\nrelations between K-theory groups, Grothendieck groups and symmetric spaces,\nthe periodic table of topological insulators and superconductors is obtained.\nThis gives the result that the periodic table of real and complex topological\nphases is originated from the Clifford chessboard and index theorems.\n",
"title": "Index of Dirac operators and classification of topological insulators"
} | null | null | null | null | true | null | 1089 | null | Default | null | null |
null | {
"abstract": " The Next Generation Transit Survey (NGTS), operating in Paranal since 2016,\nis a wide-field survey to detect Neptunes and super-Earths transiting bright\nstars, which are suitable for precise radial velocity follow-up and\ncharacterisation. Thereby, its sub-mmag photometric precision and ability to\nidentify false positives are crucial. Particularly, variable background objects\nblended in the photometric aperture frequently mimic Neptune-sized transits and\nare costly in follow-up time. These objects can best be identified with the\ncentroiding technique: if the photometric flux is lost off-centre during an\neclipse, the flux centroid shifts towards the centre of the target star.\nAlthough this method has successfully been employed by the Kepler mission, it\nhas previously not been implemented from the ground. We present a\nfully-automated centroid vetting algorithm developed for NGTS, enabled by our\nhigh-precision auto-guiding. Our method allows detecting centroid shifts with\nan average precision of 0.75 milli-pixel, and down to 0.25 milli-pixel for\nspecific targets, for a pixel size of 4.97 arcsec. The algorithm is now part of\nthe NGTS candidate vetting pipeline and automatically employed for all detected\nsignals. Further, we develop a joint Bayesian fitting model for all photometric\nand centroid data, allowing to disentangle which object (target or background)\nis causing the signal, and what its astrophysical parameters are. We\ndemonstrate our method on two NGTS objects of interest. These achievements make\nNGTS the first ground-based wide-field transit survey ever to successfully\napply the centroiding technique for automated candidate vetting, enabling the\nproduction of a robust candidate list before follow-up.\n",
"title": "Centroid vetting of transiting planet candidates from the Next Generation Transit Survey"
} | null | null | null | null | true | null | 1090 | null | Default | null | null |
null | {
"abstract": " We present the evolution of the Cosmic Spectral Energy Distribution (CSED)\nfrom $z = 1 - 0$. Our CSEDs originate from stacking individual spectral energy\ndistribution fits based on panchromatic photometry from the Galaxy and Mass\nAssembly (GAMA) and COSMOS datasets in ten redshift intervals with completeness\ncorrections applied. Below $z = 0.45$, we have credible SED fits from 100 nm to\n1 mm. Due to the relatively low sensitivity of the far-infrared data, our\nfar-infrared CSEDs contain a mix of predicted and measured fluxes above $z =\n0.45$. Our results include appropriate errors to highlight the impact of these\ncorrections. We show that the bolometric energy output of the Universe has\ndeclined by a factor of roughly four -- from $5.1 \\pm 1.0$ at $z \\sim 1$ to\n$1.3 \\pm 0.3 \\times 10^{35}~h_{70}$~W~Mpc$^{-3}$ at the current epoch. We show\nthat this decrease is robust to cosmic variance, SED modelling and other\nvarious types of error. Our CSEDs are also consistent with an increase in the\nmean age of stellar populations. We also show that dust attenuation has\ndecreased over the same period, with the photon escape fraction at 150~nm\nincreasing from $16 \\pm 3$ at $z \\sim 1$ to $24 \\pm 5$ per cent at the current\nepoch, equivalent to a decrease in $A_\\mathrm{FUV}$ of 0.4~mag. Our CSEDs\naccount for $68 \\pm 12$ and $61 \\pm 13$ per cent of the cosmic optical and\ninfrared backgrounds respectively as defined from integrated galaxy counts and\nare consistent with previous estimates of the cosmic infrared background with\nredshift.\n",
"title": "Galaxy And Mass Assembly: the evolution of the cosmic spectral energy distribution from z = 1 to z = 0"
} | null | null | null | null | true | null | 1091 | null | Default | null | null |
null | {
"abstract": " Let $f$ be a Hecke cusp form of weight $k$ for the full modular group, and\nlet $\\{\\lambda_f(n)\\}_{n\\geq 1}$ be the sequence of its normalized Fourier\ncoefficients. Motivated by the problem of the first sign change of\n$\\lambda_f(n)$, we investigate the range of $x$ (in terms of $k$) for which\nthere are cancellations in the sum $S_f(x)=\\sum_{n\\leq x} \\lambda_f(n)$. We\nfirst show that $S_f(x)=o(x\\log x)$ implies that $\\lambda_f(n)<0$ for some\n$n\\leq x$. We also prove that $S_f(x)=o(x\\log x)$ in the range $\\log x/\\log\\log\nk\\to \\infty$ assuming the Riemann hypothesis for $L(s, f)$, and furthermore\nthat this range is best possible unconditionally. More precisely, we establish\nthe existence of many Hecke cusp forms $f$ of large weight $k$, for which\n$S_f(x)\\gg_A x\\log x$, when $x=(\\log k)^A.$ Our results are $GL_2$ analogues of\nwork of Granville and Soundararajan for character sums, and could also be\ngeneralized to other families of automorphic forms.\n",
"title": "Large sums of Hecke eigenvalues of holomorphic cusp forms"
} | null | null | [
"Mathematics"
]
| null | true | null | 1092 | null | Validated | null | null |
null | {
"abstract": " Recent studies have highlighted the vulnerability of deep neural networks\n(DNNs) to adversarial examples - a visually indistinguishable adversarial image\ncan easily be crafted to cause a well-trained model to misclassify. Existing\nmethods for crafting adversarial examples are based on $L_2$ and $L_\\infty$\ndistortion metrics. However, despite the fact that $L_1$ distortion accounts\nfor the total variation and encourages sparsity in the perturbation, little has\nbeen developed for crafting $L_1$-based adversarial examples. In this paper, we\nformulate the process of attacking DNNs via adversarial examples as an\nelastic-net regularized optimization problem. Our elastic-net attacks to DNNs\n(EAD) feature $L_1$-oriented adversarial examples and include the\nstate-of-the-art $L_2$ attack as a special case. Experimental results on MNIST,\nCIFAR10 and ImageNet show that EAD can yield a distinct set of adversarial\nexamples with small $L_1$ distortion and attains similar attack performance to\nthe state-of-the-art methods in different attack scenarios. More importantly,\nEAD leads to improved attack transferability and complements adversarial\ntraining for DNNs, suggesting novel insights on leveraging $L_1$ distortion in\nadversarial machine learning and security implications of DNNs.\n",
"title": "EAD: Elastic-Net Attacks to Deep Neural Networks via Adversarial Examples"
} | null | null | null | null | true | null | 1093 | null | Default | null | null |
null | {
"abstract": " Maximizing product use is a central goal of many businesses, which makes\nretention and monetization two central analytics metrics in games. Player\nretention may refer to various duration variables quantifying product use:\ntotal playtime or session playtime are popular research targets, and active\nplaytime is well-suited for subscription games. Such research often has the\ngoal of increasing player retention or conversely decreasing player churn.\nSurvival analysis is a framework of powerful tools well suited for retention\ntype data. This paper contributes new methods to game analytics on how playtime\ncan be analyzed using survival analysis without covariates. Survival and hazard\nestimates provide both a visual and an analytic interpretation of the playtime\nphenomena as a funnel type nonparametric estimate. Metrics based on the\nsurvival curve can be used to aggregate this playtime information into a single\nstatistic. Comparison of survival curves between cohorts provides a scientific\nAB-test. All these methods work on censored data and enable computation of\nconfidence intervals. This is especially important in time and sample limited\ndata which occurs during game development. Throughout this paper, we illustrate\nthe application of these methods to real world game development problems on the\nHipster Sheep mobile game.\n",
"title": "Playtime Measurement with Survival Analysis"
} | null | null | null | null | true | null | 1094 | null | Default | null | null |
null | {
"abstract": " In this paper, we enumerate Newton polygons asymptotically. The number of\nNewton polygons is computable by a simple recurrence equation, but unexpectedly\nthe asymptotic formula of its logarithm contains growing oscillatory terms. As\nthe terms come from non-trivial zeros of the Riemann zeta function, an\nestimation of the amplitude of the oscillating part is equivalent to the\nRiemann hypothesis.\n",
"title": "Asymptotic formula of the number of Newton polygons"
} | null | null | null | null | true | null | 1095 | null | Default | null | null |
null | {
"abstract": " By applying invariant-based inverse engineering in the small-oscillations\nregime, we design the time dependence of the control parameters of an overhead\ncrane (trolley displacement and rope length), to transport a load between two\npositions at different heights with minimal final energy excitation for a\nmicrocanonical ensemble of initial conditions. The analogies between ion\ntransport in multisegmented traps or neutral atom transport in moving optical\nlattices and load manipulation by cranes opens a route for a useful transfer of\ntechniques among very different fields.\n",
"title": "Invariant-based inverse engineering of crane control parameters"
} | null | null | null | null | true | null | 1096 | null | Default | null | null |
null | {
"abstract": " In this paper, the authors consider leaf spaces of singular Riemannian\nfoliations $\\mathcal{F}$ on compact manifolds $M$ and the associated\n$\\mathcal{F}$-basic spectrum on $M$, $spec_B(M, \\mathcal{F}),$ counted with\nmultiplicities. Recently, a notion of smooth isometry $\\varphi:\nM_1/\\mathcal{F}_1\\rightarrow M_2/\\mathcal{F}_2$ between the leaf spaces of such\nsingular Riemannian foliations $(M_1,\\mathcal{F}_1)$ and $(M_2,\\mathcal{F}_2)$\nhas appeared in the literature. In this paper, the authors provide an example\nto show that the existence a smooth isometry of leaf spaces as above is not\nsufficient to guarantee the equality of $spec_B(M_1,\\mathcal{F}_1)$ and\n$spec_B(M_2,\\mathcal{F}_2).$ The authors then prove that if some additional\nconditions involving the geometry of the leaves are satisfied, then the\nequality of $spec_B(M_1,\\mathcal{F}_1)$ and $spec_B(M_2,\\mathcal{F}_2)$ is\nguaranteed. Consequences and applications to orbifold spectral theory,\nisometric group actions, and their reductions are also explored.\n",
"title": "Leaf Space Isometries of Singular Riemannian Foliations and Their Spectral Properties"
} | null | null | [
"Mathematics"
]
| null | true | null | 1097 | null | Validated | null | null |
null | {
"abstract": " We discuss a backward Monte-Carlo technique for muon transport problem, with\nemphasis on its application in muography. Backward Monte-Carlo allows exclusive\nsampling of a final state by reversing the simulation flow. In practice it can\nbe made analogous to an adjoint Monte-Carlo, though it is more versatile for\nmuon transport. A backward Monte-Carlo was implemented as a dedicated muon\ntransport library: PUMAS. It is shown for case studies relevant for muography\nimaging that the implementations of forward and backward Monte-Carlo schemes\nagree to better than 1%.\n",
"title": "Backward Monte-Carlo applied to muon transport"
} | null | null | null | null | true | null | 1098 | null | Default | null | null |
null | {
"abstract": " Noise is an inherent part of neuronal dynamics, and thus of the brain. It can\nbe observed in neuronal activity at different spatiotemporal scales, including\nin neuronal membrane potentials, local field potentials,\nelectroencephalography, and magnetoencephalography. A central research topic in\ncontemporary neuroscience is to elucidate the functional role of noise in\nneuronal information processing. Experimental studies have shown that a\nsuitable level of noise may enhance the detection of weak neuronal signals by\nmeans of stochastic resonance. In response, theoretical research, based on the\ntheory of stochastic processes, nonlinear dynamics, and statistical physics,\nhas made great strides in elucidating the mechanism and the many benefits of\nstochastic resonance in neuronal systems. In this perspective, we review recent\nresearch dedicated to neuronal stochastic resonance in biophysical mathematical\nmodels. We also explore the regulation of neuronal stochastic resonance, and we\noutline important open questions and directions for future research. A deeper\nunderstanding of neuronal stochastic resonance may afford us new insights into\nthe highly impressive information processing in the brain.\n",
"title": "Functional importance of noise in neuronal information processing"
} | null | null | null | null | true | null | 1099 | null | Default | null | null |
null | {
"abstract": " Policy evaluation is a crucial step in many reinforcement-learning\nprocedures, which estimates a value function that predicts states' long-term\nvalue under a given policy. In this paper, we focus on policy evaluation with\nlinear function approximation over a fixed dataset. We first transform the\nempirical policy evaluation problem into a (quadratic) convex-concave saddle\npoint problem, and then present a primal-dual batch gradient method, as well as\ntwo stochastic variance reduction methods for solving the problem. These\nalgorithms scale linearly in both sample size and feature dimension. Moreover,\nthey achieve linear convergence even when the saddle-point problem has only\nstrong concavity in the dual variables but no strong convexity in the primal\nvariables. Numerical experiments on benchmark problems demonstrate the\neffectiveness of our methods.\n",
"title": "Stochastic Variance Reduction Methods for Policy Evaluation"
} | null | null | null | null | true | null | 1100 | null | Default | null | null |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.