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 mechanisms underlying cardiac fibrillation have been investigated for\nover a century, but we are still finding surprising results that change our\nview of this phenomenon. The present study focuses on the transition from\nnormal rhythm to atrial fibrillation associated with a gradual increase in the\npacing rate. While some of our findings are consistent with existing\nexperimental, numerical, and theoretical studies of this problem, one result\nappears to contradict the accepted picture. Specifically we show that, in a\ntwo-dimensional model of paced homogeneous atrial tissue, transition from\ndiscordant alternans to conduction block, wave breakup, reentry, and spiral\nwave chaos is associated with transient growth of finite amplitude disturbances\nrather than a conventional instability. It is mathematically very similar to\nsubcritical, or bypass, transition from laminar fluid flow to turbulence, which\nallows many of the tools developed in the context of fluid turbulence to be\nused for improving our understanding of cardiac arrhythmias.\n",
"title": "Memory effects, transient growth, and wave breakup in a model of paced atrium"
} | null | null | [
"Physics"
]
| null | true | null | 2501 | null | Validated | null | null |
null | {
"abstract": " In this paper, we propose an information-theoretic exploration strategy for\nstochastic, discrete multi-armed bandits that achieves optimal regret. Our\nstrategy is based on the value of information criterion. This criterion\nmeasures the trade-off between policy information and obtainable rewards. High\namounts of policy information are associated with exploration-dominant searches\nof the space and yield high rewards. Low amounts of policy information favor\nthe exploitation of existing knowledge. Information, in this criterion, is\nquantified by a parameter that can be varied during search. We demonstrate that\na simulated-annealing-like update of this parameter, with a sufficiently fast\ncooling schedule, leads to an optimal regret that is logarithmic with respect\nto the number of episodes.\n",
"title": "An Analysis of the Value of Information when Exploring Stochastic, Discrete Multi-Armed Bandits"
} | null | null | null | null | true | null | 2502 | null | Default | null | null |
null | {
"abstract": " We introduce the Probabilistic Generative Adversarial Network (PGAN), a new\nGAN variant based on a new kind of objective function. The central idea is to\nintegrate a probabilistic model (a Gaussian Mixture Model, in our case) into\nthe GAN framework which supports a new kind of loss function (based on\nlikelihood rather than classification loss), and at the same time gives a\nmeaningful measure of the quality of the outputs generated by the network.\nExperiments with MNIST show that the model learns to generate realistic images,\nand at the same time computes likelihoods that are correlated with the quality\nof the generated images. We show that PGAN is better able to cope with\ninstability problems that are usually observed in the GAN training procedure.\nWe investigate this from three aspects: the probability landscape of the\ndiscriminator, gradients of the generator, and the perfect discriminator\nproblem.\n",
"title": "Probabilistic Generative Adversarial Networks"
} | null | null | [
"Computer Science",
"Statistics"
]
| null | true | null | 2503 | null | Validated | null | null |
null | {
"abstract": " The reversible jump Markov chain Monte Carlo (RJMCMC) method offers an\nacross-model simulation approach for Bayesian estimation and model comparison,\nby exploring the sampling space that consists of several models of possibly\nvarying dimensions. A naive implementation of RJMCMC to models like Gibbs\nrandom fields suffers from computational difficulties: the posterior\ndistribution for each model is termed doubly-intractable since computation of\nthe likelihood function is rarely available. Consequently, it is simply\nimpossible to simulate a transition of the Markov chain in the presence of\nlikelihood intractability. A variant of RJMCMC is presented, called noisy\nRJMCMC, where the underlying transition kernel is replaced with an\napproximation based on unbiased estimators. Based on previous theoretical\ndevelopments, convergence guarantees for the noisy RJMCMC algorithm are\nprovided. The experiments show that the noisy RJMCMC algorithm can be much more\nefficient than other exact methods, provided that an estimator with controlled\nMonte Carlo variance is used, a fact which is in agreement with the theoretical\nanalysis.\n",
"title": "Model comparison for Gibbs random fields using noisy reversible jump Markov chain Monte Carlo"
} | null | null | null | null | true | null | 2504 | null | Default | null | null |
null | {
"abstract": " We define compactifications of vector spaces which are functorial with\nrespect to certain linear maps. These \"many-body\" compactifications are\nmanifolds with corners, and the linear maps lift to b-maps in the sense of\nMelrose. We derive a simple criterion under which the lifted maps are in fact\nb-fibrations, and identify how these restrict to boundary hypersurfaces. This\ntheory is an application of a general result on the iterated blow-up of cleanly\nintersecting submanifolds which extends related results in the literature.\n",
"title": "Functorial compactification of linear spaces"
} | null | null | [
"Mathematics"
]
| null | true | null | 2505 | null | Validated | null | null |
null | {
"abstract": " We show that the m-fold connected sum $m\\#\\mathbb{C}\\mathbb{P}^{2n}$ admits\nan almost complex structure if and only if m is odd.\n",
"title": "Almost complex structures on connected sums of complex projective spaces"
} | null | null | null | null | true | null | 2506 | null | Default | null | null |
null | {
"abstract": " We present a microscopic theory of Raman scattering by a two-dimensional\nFermi liquid (FL) with Rashba and Dresselhaus types of spin-orbit coupling, and\nsubject to an in-plane magnetic field (B). In the long-wavelength limit, the\nRaman spectrum probes the collective modes of such a FL: the chiral spin waves.\nThe characteristic features of these modes are a linear-in-q term in the\ndispersion and the dependence of the mode frequency on the directions of both q\nand B. All of these features have been observed in recent Raman experiments on\nCdTe quantum wells.\n",
"title": "Raman Scattering by a Two-Dimensional Fermi Liquid with Spin-Orbit Coupling"
} | null | null | null | null | true | null | 2507 | null | Default | null | null |
null | {
"abstract": " In this work, we study the robust subspace tracking (RST) problem and obtain\none of the first two provable guarantees for it. The goal of RST is to track\nsequentially arriving data vectors that lie in a slowly changing\nlow-dimensional subspace, while being robust to corruption by additive sparse\noutliers. It can also be interpreted as a dynamic (time-varying) extension of\nrobust PCA (RPCA), with the minor difference that RST also requires a short\ntracking delay. We develop a recursive projected compressive sensing algorithm\nthat we call Nearly Optimal RST via ReProCS (ReProCS-NORST) because its\ntracking delay is nearly optimal. We prove that NORST solves both the RST and\nthe dynamic RPCA problems under weakened standard RPCA assumptions, two simple\nextra assumptions (slow subspace change and most outlier magnitudes lower\nbounded), and a few minor assumptions.\nOur guarantee shows that NORST enjoys a near optimal tracking delay of $O(r\n\\log n \\log(1/\\epsilon))$. Its required delay between subspace change times is\nthe same, and its memory complexity is $n$ times this value. Thus both these\nare also nearly optimal. Here $n$ is the ambient space dimension, $r$ is the\nsubspaces' dimension, and $\\epsilon$ is the tracking accuracy. NORST also has\nthe best outlier tolerance compared with all previous RPCA or RST methods, both\ntheoretically and empirically (including for real videos), without requiring\nany model on how the outlier support is generated. This is possible because of\nthe extra assumptions it uses.\n",
"title": "Nearly Optimal Robust Subspace Tracking"
} | null | null | null | null | true | null | 2508 | null | Default | null | null |
null | {
"abstract": " In this paper, we argue for the adoption of a normative definition of\nfairness within the machine learning community. After characterizing this\ndefinition, we review the current literature of Fair ML in light of its\nimplications. We end by suggesting ways to incorporate a broader community and\ngenerate further debate around how to decide what is fair in ML.\n",
"title": "The Authority of \"Fair\" in Machine Learning"
} | null | null | null | null | true | null | 2509 | null | Default | null | null |
null | {
"abstract": " Understanding tie strength in social networks, and the factors that influence\nit, have received much attention in a myriad of disciplines for decades.\nSeveral models incorporating indicators of tie strength have been proposed and\nused to quantify relationships in social networks, and a standard set of\nstructural network metrics have been applied to predominantly online social\nmedia sites to predict tie strength. Here, we introduce the concept of the\n\"social bow tie\" framework, a small subgraph of the network that consists of a\ncollection of nodes and ties that surround a tie of interest, forming a\ntopological structure that resembles a bow tie. We also define several\nintuitive and interpretable metrics that quantify properties of the bow tie. We\nuse random forests and regression models to predict categorical and continuous\nmeasures of tie strength from different properties of the bow tie, including\nnodal attributes. We also investigate what aspects of the bow tie are most\npredictive of tie strength in two distinct social networks: a collection of 75\nrural villages in India and a nationwide call network of European mobile phone\nusers. Our results indicate several of the bow tie metrics are highly\npredictive of tie strength, and we find the more the social circles of two\nindividuals overlap, the stronger their tie, consistent with previous findings.\nHowever, we also find that the more tightly-knit their non-overlapping social\ncircles, the weaker the tie. This new finding complements our current\nunderstanding of what drives the strength of ties in social networks.\n",
"title": "The Social Bow Tie"
} | null | null | null | null | true | null | 2510 | null | Default | null | null |
null | {
"abstract": " The paper considers non-stationary responses in reduced-order model of\npartially liquid-filled tank under external forcing. The model involves one\ncommon degree of freedom for the tank and the non-sloshing portion of the\nliquid, and the other one -- for the sloshing portion of the liquid. The\ncoupling between these degrees of freedom is nonlinear, with the lowest-order\npotential dictated by symmetry considerations. Since the mass of the sloshing\nliquid in realistic conditions does not exceed 10% of the total mass of the\nsystem, the reduced-order model turns to be formally equivalent to well-studied\noscillatory systems with nonlinear energy sinks (NES). Exploiting this analogy,\nand applying the methodology known from the studies of the systems with the\nNES, we predict a multitude of possible non-stationary responses in the\nconsidered model. These responses conform, at least on the qualitative level,\nto the responses observed in experimental sloshing settings, multi-modal\ntheoretical models and full-scale numeric simulations.\n",
"title": "Response Regimes in Equivalent Mechanical Model of Moderately Nonlinear Liquid Sloshing"
} | null | null | null | null | true | null | 2511 | null | Default | null | null |
null | {
"abstract": " Investigation of social influence dynamics requires mathematical models that\nare \"simple\" enough to admit rigorous analysis, and yet sufficiently \"rich\" to\ncapture salient features of social groups. Thus, the mechanism of iterative\nopinion pooling from (DeGroot, 1974), which can explain the generation of\nconsensus, was elaborated in (Friedkin and Johnsen, 1999) to take into account\nindividuals' ongoing attachments to their initial opinions, or prejudices. The\n\"anchorage\" of individuals to their prejudices may disable reaching consensus\nand cause disagreement in a social influence network. Further elaboration of\nthis model may be achieved by relaxing its restrictive assumption of a\ntime-invariant influence network. During opinion dynamics on an issue, arcs of\ninterpersonal influence may be added or subtracted from the network, and the\ninfluence weights assigned by an individual to his/her neighbors may alter. In\nthis paper, we establish new important properties of the (Friedkin and Johnsen,\n1999) opinion formation model, and also examine its extension to time-varying\nsocial influence networks.\n",
"title": "Opinion evolution in time-varying social influence networks with prejudiced agents"
} | null | null | [
"Computer Science",
"Physics",
"Mathematics"
]
| null | true | null | 2512 | null | Validated | null | null |
null | {
"abstract": " The weighted k-nearest neighbors algorithm is one of the most fundamental\nnon-parametric methods in pattern recognition and machine learning. The\nquestion of setting the optimal number of neighbors as well as the optimal\nweights has received much attention throughout the years, nevertheless this\nproblem seems to have remained unsettled. In this paper we offer a simple\napproach to locally weighted regression/classification, where we make the\nbias-variance tradeoff explicit. Our formulation enables us to phrase a notion\nof optimal weights, and to efficiently find these weights as well as the\noptimal number of neighbors efficiently and adaptively, for each data point\nwhose value we wish to estimate. The applicability of our approach is\ndemonstrated on several datasets, showing superior performance over standard\nlocally weighted methods.\n",
"title": "k*-Nearest Neighbors: From Global to Local"
} | null | null | null | null | true | null | 2513 | null | Default | null | null |
null | {
"abstract": " An important novelty of 5G is its role in transforming the industrial\nproduction into Industry 4.0. Specifically, Ultra-Reliable Low Latency\nCommunications (URLLC) will, in many cases, enable replacement of cables with\nwireless connections and bring freedom in designing and operating\ninterconnected machines, robots, and devices. However, not all industrial links\nwill be of URLLC type; e.g. some applications will require high data rates.\nFurthermore, these industrial networks will be highly heterogeneous, featuring\nvarious communication technologies. We consider network slicing as a mechanism\nto handle the diverse set of requirements to the network. We present methods\nfor slicing deterministic and packet-switched industrial communication\nprotocols at an abstraction level that is decoupled from the specific\nimplementation of the underlying technologies. Finally, we show how network\ncalculus can be used to assess the end-to-end properties of the network slices.\n",
"title": "Network Slicing for Ultra-Reliable Low Latency Communication in Industry 4.0 Scenarios"
} | null | null | null | null | true | null | 2514 | null | Default | null | null |
null | {
"abstract": " We consider a reinforcement learning (RL) setting in which the agent\ninteracts with a sequence of episodic MDPs. At the start of each episode the\nagent has access to some side-information or context that determines the\ndynamics of the MDP for that episode. Our setting is motivated by applications\nin healthcare where baseline measurements of a patient at the start of a\ntreatment episode form the context that may provide information about how the\npatient might respond to treatment decisions. We propose algorithms for\nlearning in such Contextual Markov Decision Processes (CMDPs) under an\nassumption that the unobserved MDP parameters vary smoothly with the observed\ncontext. We also give lower and upper PAC bounds under the smoothness\nassumption. Because our lower bound has an exponential dependence on the\ndimension, we consider a tractable linear setting where the context is used to\ncreate linear combinations of a finite set of MDPs. For the linear setting, we\ngive a PAC learning algorithm based on KWIK learning techniques.\n",
"title": "Markov Decision Processes with Continuous Side Information"
} | null | null | null | null | true | null | 2515 | null | Default | null | null |
null | {
"abstract": " We study a three-wave truncation of a recently proposed damped/forced\nhigh-order nonlinear Schrödinger equation for deep-water gravity waves under\nthe effect of wind and viscosity. The evolution of the norm (wave-action) and\nspectral mean of the full model are well captured by the reduced dynamics.\nThree regimes are found for the wind-viscosity balance: we classify them\naccording to the attractor in the phase-plane of the truncated system and to\nthe shift of the spectral mean. A downshift can coexist with both net forcing\nand damping, i.e., attraction to period-1 or period-2 solutions. Upshift is\nassociated with stronger winds, i.e., to a net forcing where the attractor is\nalways a period-1 solution. The applicability of our classification to\nexperiments in long wave-tanks is verified.\n",
"title": "Nonlinear stage of Benjamin-Feir instability in forced/damped deep water waves"
} | null | null | [
"Physics"
]
| null | true | null | 2516 | null | Validated | null | null |
null | {
"abstract": " With the future likely to see even more pervasive computation, computational\nthinking (problem-solving skills incorporating computing knowledge) is now\nbeing recognized as a fundamental skill needed by all students. Computational\nthinking is conceptualizing as opposed to programming, promotes natural human\nthinking style than algorithmic reasoning, complements and combines\nmathematical and engineering thinking, and it emphasizes ideas, not artifacts.\nIn this paper, we outline a new visual language, called Patch, using which\nstudents are able to express their solutions to eScience computational problems\nin abstract visual tools. Patch is closer to high level procedural languages\nsuch as C++ or Java than Scratch or Snap! but similar to them in ease of use\nand combines simplicity and expressive power in one single platform.\n",
"title": "Computational Thinking in Patch"
} | null | null | null | null | true | null | 2517 | null | Default | null | null |
null | {
"abstract": " Skoda's 1972 result on ideal generation is a crucial ingredient in the\nanalytic approach to the finite generation of the canonical ring and the\nabundance conjecture. Special analytic techniques developed by Skoda, other\nthan applications of the usual vanishing theorems and L2 estimates for the\nd-bar equation, are required for its proof. This note (which is part of a\nlecture given in the 60th birthday conference for Lawrence Ein) gives a\nsimpler, more straightforward proof of Skoda's result, which makes it a natural\nconsequence of the standard techniques in vanishing theorems and solving d-bar\nequation with L2 estimates. The proof involves the following three ingredients:\n(i) one particular Cauchy-Schwarz inequality for tensors with a special factor\nwhich accounts for the exponent of the denominator in the formulation of the\nintegral condition for Skoda's ideal generation, (ii) the nonnegativity of\nNakano curvature of the induced metric of a special co-rank-1 subbundle of a\ntrivial vector bundle twisted by a special scalar weight function, and (iii)\nthe vanishing theorem and solvability of d-bar equation with L2 estimates for\nvector bundles of nonnegative Nakano curvature on a strictly pseudoconvex\ndomain. Our proof gives readily other similar results on ideal generation.\n",
"title": "Skoda's Ideal Generation from Vanishing Theorem for Semipositive Nakano Curvature and Cauchy-Schwarz Inequality for Tensors"
} | null | null | [
"Mathematics"
]
| null | true | null | 2518 | null | Validated | null | null |
null | {
"abstract": " One of the defining properties of deep learning is that models are chosen to\nhave many more parameters than available training data. In light of this\ncapacity for overfitting, it is remarkable that simple algorithms like SGD\nreliably return solutions with low test error. One roadblock to explaining\nthese phenomena in terms of implicit regularization, structural properties of\nthe solution, and/or easiness of the data is that many learning bounds are\nquantitatively vacuous when applied to networks learned by SGD in this \"deep\nlearning\" regime. Logically, in order to explain generalization, we need\nnonvacuous bounds. We return to an idea by Langford and Caruana (2001), who\nused PAC-Bayes bounds to compute nonvacuous numerical bounds on generalization\nerror for stochastic two-layer two-hidden-unit neural networks via a\nsensitivity analysis. By optimizing the PAC-Bayes bound directly, we are able\nto extend their approach and obtain nonvacuous generalization bounds for deep\nstochastic neural network classifiers with millions of parameters trained on\nonly tens of thousands of examples. We connect our findings to recent and old\nwork on flat minima and MDL-based explanations of generalization.\n",
"title": "Computing Nonvacuous Generalization Bounds for Deep (Stochastic) Neural Networks with Many More Parameters than Training Data"
} | null | null | [
"Computer Science"
]
| null | true | null | 2519 | null | Validated | null | null |
null | {
"abstract": " Dielectric microstructures have generated much interest in recent years as a\nmeans of accelerating charged particles when powered by solid state lasers. The\nacceleration gradient (or particle energy gain per unit length) is an important\nfigure of merit. To design structures with high acceleration gradients, we\nexplore the adjoint variable method, a highly efficient technique used to\ncompute the sensitivity of an objective with respect to a large number of\nparameters. With this formalism, the sensitivity of the acceleration gradient\nof a dielectric structure with respect to its entire spatial permittivity\ndistribution is calculated by the use of only two full-field electromagnetic\nsimulations, the original and adjoint. The adjoint simulation corresponds\nphysically to the reciprocal situation of a point charge moving through the\naccelerator gap and radiating. Using this formalism, we perform numerical\noptimizations aimed at maximizing acceleration gradients, which generate\nfabricable structures of greatly improved performance in comparison to\npreviously examined geometries.\n",
"title": "Method for Computationally Efficient Design of Dielectric Laser Accelerators"
} | null | null | null | null | true | null | 2520 | null | Default | null | null |
null | {
"abstract": " Given a zero-dimensional ideal I in a polynomial ring, many computations\nstart by finding univariate polynomials in I. Searching for a univariate\npolynomial in I is a particular case of considering the minimal polynomial of\nan element in P/I. It is well known that minimal polynomials may be computed\nvia elimination, therefore this is considered to be a \"resolved problem\". But\nbeing the key of so many computations, it is worth investigating its meaning,\nits optimization, its applications.\n",
"title": "Computing and Using Minimal Polynomials"
} | null | null | null | null | true | null | 2521 | null | Default | null | null |
null | {
"abstract": " In this paper, we introduce the concept of a virtual machine with\ngraph-organised memory as a versatile backend for both explicit-state and\nabstraction-driven verification of software. Our virtual machine uses the LLVM\nIR as its instruction set, enriched with a small set of hypercalls. We show\nthat the provided hypercalls are sufficient to implement a small operating\nsystem, which can then be linked with applications to provide a\nPOSIX-compatible verification environment. Finally, we demonstrate the\nviability of the approach through a comparison with a more\ntraditionally-designed LLVM model checker.\n",
"title": "DiVM: Model Checking with LLVM and Graph Memory"
} | null | null | [
"Computer Science"
]
| null | true | null | 2522 | null | Validated | null | null |
null | {
"abstract": " We present a self-contained proof of Uhlenbeck's decomposition theorem for\n$\\Omega\\in L^p(\\mathbb{B}^n,so(m)\\otimes\\Lambda^1\\mathbb{R}^n)$ for $p\\in\n(1,n)$ with Sobolev type estimates in the case $p \\in[n/2,n)$ and\nMorrey-Sobolev type estimates in the case $p\\in (1,n/2)$. We also prove an\nanalogous theorem in the case when $\\Omega\\in L^p( \\mathbb{B}^n, TCO_{+}(m)\n\\otimes \\Lambda^1\\mathbb{R}^n)$, which corresponds to Uhlenbeck's theorem with\nconformal gauge group.\n",
"title": "Uhlenbeck's decomposition in Sobolev and Morrey-Sobolev spaces"
} | null | null | null | null | true | null | 2523 | null | Default | null | null |
null | {
"abstract": " We consider the problem of making distributed computations robust to noise,\nin particular to worst-case (adversarial) corruptions of messages. We give a\ngeneral distributed interactive coding scheme which simulates any asynchronous\ndistributed protocol while tolerating an optimal corruption of a $\\Theta(1/n)$\nfraction of all messages while incurring a moderate blowup of $O(n\\log^2 n)$ in\nthe communication complexity.\nOur result is the first fully distributed interactive coding scheme in which\nthe topology of the communication network is not known in advance. Prior work\nrequired either a coordinating node to be connected to all other nodes in the\nnetwork or assumed a synchronous network in which all nodes already know the\ncomplete topology of the network.\n",
"title": "Making Asynchronous Distributed Computations Robust to Noise"
} | null | null | null | null | true | null | 2524 | null | Default | null | null |
null | {
"abstract": " We present a straightforward source-to-source transformation that introduces\njustifications for user-defined constraints into the CHR programming language.\nThen a scheme of two rules suffices to allow for logical retraction (deletion,\nremoval) of constraints during computation. Without the need to recompute from\nscratch, these rules remove not only the constraint but also undo all\nconsequences of the rule applications that involved the constraint. We prove a\nconfluence result concerning the rule scheme and show its correctness. When\nalgorithms are written in CHR, constraints represent both data and operations.\nCHR is already incremental by nature, i.e. constraints can be added at runtime.\nLogical retraction adds decrementality. Hence any algorithm written in CHR with\njustifications will become fully dynamic. Operations can be undone and data can\nbe removed at any point in the computation without compromising the correctness\nof the result. We present two classical examples of dynamic algorithms, written\nin our prototype implementation of CHR with justifications that is available\nonline: maintaining the minimum of a changing set of numbers and shortest paths\nin a graph whose edges change.\n",
"title": "Justifications in Constraint Handling Rules for Logical Retraction in Dynamic Algorithms"
} | null | null | null | null | true | null | 2525 | null | Default | null | null |
null | {
"abstract": " We investigate the open dynamics of an atomic impurity embedded in a\none-dimensional Bose-Hubbard lattice. We derive the reduced evolution equation\nfor the impurity and show that the Bose-Hubbard lattice behaves as a tunable\nengineered environment allowing to simulate both Markovian and non-Markovian\ndynamics in a controlled and experimentally realisable way. We demonstrate that\nthe presence or absence of memory effects is a signature of the nature of the\nexcitations induced by the impurity, being delocalized or localized in the two\nlimiting cases of superfluid and Mott insulator, respectively. Furthermore, our\nfindings show how the excitations supported in the two phases can be\ncharacterized as information carriers.\n",
"title": "Bose-Hubbard lattice as a controllable environment for open quantum systems"
} | null | null | null | null | true | null | 2526 | null | Default | null | null |
null | {
"abstract": " Composition and lattice join (transitive closure of a union) of equivalence\nrelations are operations taking pairs of decidable equivalence relations to\nrelations that are semi-decidable, but not necessarily decidable. This article\naddresses the question, is every semi-decidable equivalence relation obtainable\nin those ways from a pair of decidable equivalence relations? It is shown that\nevery semi-decidable equivalence relation, of which every equivalence class is\ninfinite, is obtainable as both a composition and a lattice join of decidable\nequivalence relations having infinite equivalence classes. An example is\nconstructed of a semi-decidable, but not decidable, equivalence relation having\nfinite equivalence classes that can be obtained from decidable equivalence\nrelations, both by composition and also by lattice join. Another example is\nconstructed, in which such a relation cannot be obtained from decidable\nequivalence relations in either of the two ways.\n",
"title": "Semi-decidable equivalence relations obtained by composition and lattice join of decidable equivalence relations"
} | null | null | null | null | true | null | 2527 | null | Default | null | null |
null | {
"abstract": " We propose a simple risk-limiting audit for elections, ClipAudit. To\ndetermine whether candidate A (the reported winner) actually beat candidate B\nin a plurality election, ClipAudit draws ballots at random, without\nreplacement, until either all cast ballots have been drawn, or until \\[ a - b\n\\ge \\beta \\sqrt{a+b}\n\\] where $a$ is the number of ballots in the sample for the reported winner\nA, and $b$ is the number of ballots in the sample for opponent B, and where\n$\\beta$ is a constant determined a priori as a function of the number $n$ of\nballots cast and the risk-limit $\\alpha$. ClipAudit doesn't depend on the\nunofficial margin (as does Bravo). We show how to extend ClipAudit to contests\nwith multiple winners or losers, or to multiple contests.\n",
"title": "ClipAudit: A Simple Risk-Limiting Post-Election Audit"
} | null | null | null | null | true | null | 2528 | null | Default | null | null |
null | {
"abstract": " In this paper we study the category LCA(2) of certain non-locally compact\nabelian topological groups, and extend the notion of Weil index. As\napplications we deduce some product formulas for curves over local fields and\narithmetic surfaces.\n",
"title": "LCA(2), Weil index, and product formula"
} | null | null | null | null | true | null | 2529 | null | Default | null | null |
null | {
"abstract": " We study how to sample paths of a random walk up to the first time it crosses\na fixed barrier, in the setting where the step sizes are iid with negative mean\nand have a regularly varying right tail. We introduce a desirable property for\na change of measure to be suitable for exact simulation. We study whether the\nchange of measure of Blanchet and Glynn (2008) satisfies this property and show\nthat it does so if and only if the tail index $\\alpha$ of the right tail lies\nin the interval $(1, \\, 3/2)$.\n",
"title": "A Dichotomy for Sampling Barrier-Crossing Events of Random Walks with Regularly Varying Tails"
} | null | null | null | null | true | null | 2530 | null | Default | null | null |
null | {
"abstract": " Cell migration is a fundamental process involved in physiological phenomena\nsuch as the immune response and morphogenesis, but also in pathological\nprocesses, such as the development of tumor metastasis. These functions are\neffectively ensured because cells are active systems that adapt to their\nenvironment. In this work, we consider a migrating cell as an active particle,\nwhere its intracellular activity is responsible for motion. Such system was\nalready modeled in a previous model where the protrusion activity of the cell\nwas described by a stochastic Markovian jump process. The model was proven able\nto capture the diversity in observed trajectories. Here, we add a description\nof the effect of an external chemical attractive signal on the protrusion\ndynamics, that may vary in time. We show that the resulting stochastic model is\na well-posed non-homogeneous Markovian process, and provide cell trajectories\nin different settings, illustrating the effects of the signal on long-term\ntrajectories.\n",
"title": "Crawling migration under chemical signalling: a stochastic particle model"
} | null | null | null | null | true | null | 2531 | null | Default | null | null |
null | {
"abstract": " Recent work has proposed various adversarial losses for training generative\nadversarial networks. Yet, it remains unclear what certain types of functions\nare valid adversarial loss functions, and how these loss functions perform\nagainst one another. In this paper, we aim to gain a deeper understanding of\nadversarial losses by decoupling the effects of their component functions and\nregularization terms. We first derive some necessary and sufficient conditions\nof the component functions such that the adversarial loss is a divergence-like\nmeasure between the data and the model distributions. In order to\nsystematically compare different adversarial losses, we then propose DANTest, a\nnew, simple framework based on discriminative adversarial networks. With this\nframework, we evaluate an extensive set of adversarial losses by combining\ndifferent component functions and regularization approaches. This study leads\nto some new insights into the adversarial losses. For reproducibility, all\nsource code is available at this https URL .\n",
"title": "Towards a Deeper Understanding of Adversarial Losses"
} | null | null | null | null | true | null | 2532 | null | Default | null | null |
null | {
"abstract": " The detection of thousands of extrasolar planets by the transit method\nnaturally raises the question of whether potential extrasolar observers could\ndetect the transits of the Solar System planets. We present a comprehensive\nanalysis of the regions in the sky from where transit events of the Solar\nSystem planets can be detected. We specify how many different Solar System\nplanets can be observed from any given point in the sky, and find the maximum\nnumber to be three. We report the probabilities of a randomly positioned\nexternal observer to be able to observe single and multiple Solar System planet\ntransits; specifically, we find a probability of 2.518% to be able to observe\nat least one transiting planet, 0.229% for at least two transiting planets, and\n0.027% for three transiting planets. We identify 68 known exoplanets that have\na favourable geometric perspective to allow transit detections in the Solar\nSystem and we show how the ongoing K2 mission will extend this list. We use\noccurrence rates of exoplanets to estimate that there are $3.2\\pm1.2$ and\n$6.6^{+1.3}_{-0.8}$ temperate Earth-sized planets orbiting GK and M dwarf stars\nbrighter than $V=13$ and $V=16$ respectively, that are located in the Earth's\ntransit zone.\n",
"title": "Transit Visibility Zones of the Solar System Planets"
} | null | null | [
"Physics"
]
| null | true | null | 2533 | null | Validated | null | null |
null | {
"abstract": " We define nearest-neighbour point processes on graphs with Euclidean edges\nand linear networks. They can be seen as the analogues of renewal processes on\nthe real line. We show that the Delaunay neighbourhood relation on a tree\nsatisfies the Baddeley--M{\\o}ller consistency conditions and provide a\ncharacterisation of Markov functions with respect to this relation. We show\nthat a modified relation defined in terms of the local geometry of the graph\nsatisfies the consistency conditions for all graphs with Euclidean edges.\n",
"title": "Nearest-neighbour Markov point processes on graphs with Euclidean edges"
} | null | null | [
"Mathematics",
"Statistics"
]
| null | true | null | 2534 | null | Validated | null | null |
null | {
"abstract": " One of the challenges in model-based control of stochastic dynamical systems\nis that the state transition dynamics are involved, and it is not easy or\nefficient to make good-quality predictions of the states. Moreover, there are\nnot many representational models for the majority of autonomous systems, as it\nis not easy to build a compact model that captures the entire dynamical\nsubtleties and uncertainties. In this work, we present a hierarchical Bayesian\nlinear regression model with local features to learn the dynamics of a\nmicro-robotic system as well as two simpler examples, consisting of a\nstochastic mass-spring damper and a stochastic double inverted pendulum on a\ncart. The model is hierarchical since we assume non-stationary priors for the\nmodel parameters. These non-stationary priors make the model more flexible by\nimposing priors on the priors of the model. To solve the maximum likelihood\n(ML) problem for this hierarchical model, we use the variational expectation\nmaximization (EM) algorithm, and enhance the procedure by introducing hidden\ntarget variables. The algorithm yields parsimonious model structures, and\nconsistently provides fast and accurate predictions for all our examples\ninvolving large training and test sets. This demonstrates the effectiveness of\nthe method in learning stochastic dynamics, which makes it suitable for future\nuse in a paradigm, such as model-based reinforcement learning, to compute\noptimal control policies in real time.\n",
"title": "A Hierarchical Bayesian Linear Regression Model with Local Features for Stochastic Dynamics Approximation"
} | null | null | null | null | true | null | 2535 | null | Default | null | null |
null | {
"abstract": " Health care is one of the most exciting frontiers in data mining and machine\nlearning. Successful adoption of electronic health records (EHRs) created an\nexplosion in digital clinical data available for analysis, but progress in\nmachine learning for healthcare research has been difficult to measure because\nof the absence of publicly available benchmark data sets. To address this\nproblem, we propose four clinical prediction benchmarks using data derived from\nthe publicly available Medical Information Mart for Intensive Care (MIMIC-III)\ndatabase. These tasks cover a range of clinical problems including modeling\nrisk of mortality, forecasting length of stay, detecting physiologic decline,\nand phenotype classification. We propose strong linear and neural baselines for\nall four tasks and evaluate the effect of deep supervision, multitask training\nand data-specific architectural modifications on the performance of neural\nmodels.\n",
"title": "Multitask Learning and Benchmarking with Clinical Time Series Data"
} | null | null | null | null | true | null | 2536 | null | Default | null | null |
null | {
"abstract": " Let $X$ be a normal, connected and projective variety over an algebraically\nclosed field $k$. It is known that a vector bundle $V$ on $X$ is essentially\nfinite if and only if it is trivialized by a proper surjective morphism $f:Y\\to\nX$. In this paper we introduce a different approach to this problem which\nallows to extend the results to normal, connected and strongly pseudo-proper\nalgebraic stack of finite type over an arbitrary field $k$.\n",
"title": "Essentially Finite Vector Bundles on Normal Pseudo-proper Algebraic Stacks"
} | null | null | null | null | true | null | 2537 | null | Default | null | null |
null | {
"abstract": " A pressure driven flow in contact interface between elastic solids with wavy\nsurfaces is studied. We consider a strong coupling between the solid and the\nfluid problems, which is relevant when the fluid pressure is comparable with\nthe contact pressure. An approximate analytical solution is obtained for this\ncoupled problem. A finite-element monolithically coupled framework is used to\nsolve the problem numerically. A good agreement is obtained between the two\nsolutions within the region of the validity of the analytical one. A power-law\ninterface transmissivity decay is observed near the percolation. Finally, we\nshowed that the external pressure needed to seal the channel is an affine\nfunction of the inlet pressure and does not depend on the outlet pressure.\n",
"title": "Fluid flow across a wavy channel brought in contact"
} | null | null | null | null | true | null | 2538 | null | Default | null | null |
null | {
"abstract": " Species tree reconstruction from genomic data is increasingly performed using\nmethods that account for sources of gene tree discordance such as incomplete\nlineage sorting. One popular method for reconstructing species trees from\nunrooted gene tree topologies is ASTRAL. In this paper, we derive theoretical\nsample complexity results for the number of genes required by ASTRAL to\nguarantee reconstruction of the correct species tree with high probability. We\nalso validate those theoretical bounds in a simulation study. Our results\nindicate that ASTRAL requires $\\mathcal{O}(f^{-2} \\log n)$ gene trees to\nreconstruct the species tree correctly with high probability where n is the\nnumber of species and f is the length of the shortest branch in the species\ntree. Our simulations, which are the first to test ASTRAL explicitly under the\nanomaly zone, show trends consistent with the theoretical bounds and also\nprovide some practical insights on the conditions where ASTRAL works well.\n",
"title": "Species tree estimation using ASTRAL: how many genes are enough?"
} | null | null | null | null | true | null | 2539 | null | Default | null | null |
null | {
"abstract": " In this paper we establish the characterization of the weighted BMO via two\nweight commutators in the settings of the Neumann Laplacian $\\Delta_{N_+}$ on\nthe upper half space $\\mathbb{R}^n_+$ and the reflection Neumann Laplacian\n$\\Delta_N$ on $\\mathbb{R}^n$ with respect to the weights associated to\n$\\Delta_{N_+}$ and $\\Delta_{N}$ respectively. This in turn yields a weak\nfactorization for the corresponding weighted Hardy spaces, where in particular,\nthe weighted class associated to $\\Delta_{N}$ is strictly larger than the\nMuckenhoupt weighted class and contains non-doubling weights. In our study, we\nalso make contributions to the classical Muckenhoupt--Wheeden weighted Hardy\nspace (BMO space respectively) by showing that it can be characterized via area\nfunction (Carleson measure respectively) involving the semigroup generated by\nthe Laplacian on $\\mathbb{R}^n$ and that the duality of these weighted Hardy\nand BMO spaces holds for Muckenhoupt $A^p$ weights with $p\\in (1,2]$ while the\npreviously known related results cover only $p\\in (1,{n+1\\over n}]$. We also\npoint out that this two weight commutator theorem might not be true in the\nsetting of general operators $L$, and in particular we show that it is not true\nwhen $L$ is the Dirichlet Laplacian $\\Delta_{D_+}$ on $\\mathbb{R}^n_+$.\n",
"title": "Two weight Commutators in the Dirichlet and Neumann Laplacian settings"
} | null | null | null | null | true | null | 2540 | null | Default | null | null |
null | {
"abstract": " We represent Matérn functions in terms of Schoenberg's integrals which\nensure the positive definiteness and prove the systems of translates of\nMatérn functions form Riesz sequences in $L^2(\\R^n)$ or Sobolev spaces. Our\napproach is based on a new class of integral transforms that generalize Fourier\ntransforms for radial functions. We also consider inverse multi-quadrics and\nobtain similar results.\n",
"title": "Schoenberg Representations and Gramian Matrices of Matérn Functions"
} | null | null | [
"Mathematics"
]
| null | true | null | 2541 | null | Validated | null | null |
null | {
"abstract": " The Hasse-Witt matrix of a hypersurface in ${\\mathbb P}^n$ over a finite\nfield of characteristic $p$ gives essentially complete mod $p$ information\nabout the zeta function of the hypersurface. But if the degree $d$ of the\nhypersurface is $\\leq n$, the zeta function is trivial mod $p$ and the\nHasse-Witt matrix is zero-by-zero. We generalize a classical formula for the\nHasse-Witt matrix to obtain a matrix that gives a nontrivial congruence for the\nzeta function for all $d$. We also describe the differential equations\nsatisfied by this matrix and prove that it is generically invertible.\n",
"title": "A generalization of the Hasse-Witt matrix of a hypersurface"
} | null | null | null | null | true | null | 2542 | null | Default | null | null |
null | {
"abstract": " We propose to study the problem of few-shot learning with the prism of\ninference on a partially observed graphical model, constructed from a\ncollection of input images whose label can be either observed or not. By\nassimilating generic message-passing inference algorithms with their\nneural-network counterparts, we define a graph neural network architecture that\ngeneralizes several of the recently proposed few-shot learning models. Besides\nproviding improved numerical performance, our framework is easily extended to\nvariants of few-shot learning, such as semi-supervised or active learning,\ndemonstrating the ability of graph-based models to operate well on 'relational'\ntasks.\n",
"title": "Few-Shot Learning with Graph Neural Networks"
} | null | null | null | null | true | null | 2543 | null | Default | null | null |
null | {
"abstract": " We report on the precise measurement of the atomic mass of a single proton\nwith a purpose-built Penning-trap system. With a precision of 32\nparts-per-trillion our result not only improves on the current CODATA\nliterature value by a factor of three, but also disagrees with it at a level of\nabout 3 standard deviations.\n",
"title": "High-precision measurement of the proton's atomic mass"
} | null | null | null | null | true | null | 2544 | null | Default | null | null |
null | {
"abstract": " Distribution of cold gas in the post-reionization era provides an important\nlink between distribution of galaxies and the process of star formation.\nRedshifted 21 cm radiation from the Hyperfine transition of neutral Hydrogen\nallows us to probe the neutral component of cold gas, most of which is to be\nfound in the interstellar medium of galaxies. Existing and upcoming radio\ntelescopes can probe the large scale distribution of neutral Hydrogen via HI\nintensity mapping. In this paper we use an estimate of the HI power spectrum\nderived using an ansatz to compute the expected signal from the large scale HI\ndistribution at z ~ 3. We find that the scale dependence of bias at small\nscales makes a significant difference to the expected signal even at large\nangular scales. We compare the predicted signal strength with the sensitivity\nof radio telescopes that can observe such radiation and calculate the\nobservation time required for detecting neutral Hydrogen at these redshifts. We\nfind that OWFA (Ooty Wide Field Array) offers the best possibility to detect\nneutral Hydrogen at z ~ 3 before the SKA (Square Kilometer Array) becomes\noperational. We find that the OWFA should be able to make a 3 sigma or a more\nsignificant detection in 2000 hours of observations at several angular scales.\nCalculations done using the Fisher matrix approach indicate that a 5 sigma\ndetection of the binned HI power spectrum via measurement of the amplitude of\nthe HI power spectrum is possible in 1000 hours (Sarkar, Bharadwaj and Ali,\n2017).\n",
"title": "Prospects of detecting HI using redshifted 21 cm radiation at z ~ 3"
} | null | null | [
"Physics"
]
| null | true | null | 2545 | null | Validated | null | null |
null | {
"abstract": " Assume that $T$ is a self-adjoint operator on a Hilbert space $\\mathcal{H}$\nand that the spectrum of $T$ is confined in the union $\\bigcup_{j\\in\nJ}\\Delta_j$, $J\\subseteq\\mathbb{Z}$, of segments $\\Delta_j=[\\alpha_j,\n\\beta_j]\\subset\\mathbb{R}$ such that $\\alpha_{j+1}>\\beta_j$ and $$ \\inf_{j}\n\\left(\\alpha_{j+1}-\\beta_j\\right) = d > 0. $$ If $B$ is a bounded (in general\nnon-self-adjoint) perturbation of $T$ with $\\|B\\|=:b<d/2$ then the spectrum of\nthe perturbed operator $A=T+B$ lies in the union $\\bigcup_{j\\in J}\nU_{b}(\\Delta_j)$ of the mutually disjoint closed $b$-neighborhoods\n$U_{b}(\\Delta_j)$ of the segments $\\Delta_j$ in $\\mathbb{C}$. Let $Q_j$ be the\nRiesz projection onto the invariant subspace of $A$ corresponding to the part\nof the spectrum of $A$ lying in $U_{b}\\left(\\Delta_j\\right)$, $j\\in J$. Our\nmain result is as follows: The subspaces $\\mathcal{L}_j=Q_j(\\mathcal H)$, $j\\in\nJ$, form an unconditional basis in the whole space $\\mathcal H$.\n",
"title": "Unconditional bases of subspaces related to non-self-adjoint perturbations of self-adjoint operators"
} | null | null | [
"Mathematics"
]
| null | true | null | 2546 | null | Validated | null | null |
null | {
"abstract": " We formulate and propose an algorithm (MultiRank) for the ranking of nodes\nand layers in large multiplex networks. MultiRank takes into account the full\nmultiplex network structure of the data and exploits the dual nature of the\nnetwork in terms of nodes and layers. The proposed centrality of the layers\n(influences) and the centrality of the nodes are determined by a coupled set of\nequations. The basic idea consists in assigning more centrality to nodes that\nreceive links from highly influential layers and from already central nodes.\nThe layers are more influential if highly central nodes are active in them. The\nalgorithm applies to directed/undirected as well as to weighted/unweighted\nmultiplex networks. We discuss the application of MultiRank to three major\nexamples of multiplex network datasets: the European Air Transportation\nMultiplex Network, the Pierre Auger Multiplex Collaboration Network and the FAO\nMultiplex Trade Network.\n",
"title": "Centralities of Nodes and Influences of Layers in Large Multiplex Networks"
} | null | null | null | null | true | null | 2547 | null | Default | null | null |
null | {
"abstract": " Given a combinatorial design $\\mathcal{D}$ with block set $\\mathcal{B}$, the\nblock-intersection graph (BIG) of $\\mathcal{D}$ is the graph that has\n$\\mathcal{B}$ as its vertex set, where two vertices $B_{1} \\in \\mathcal{B}$ and\n$B_{2} \\in \\mathcal{B} $ are adjacent if and only if $|B_{1} \\cap B_{2}| > 0$.\nThe $i$-block-intersection graph ($i$-BIG) of $\\mathcal{D}$ is the graph that\nhas $\\mathcal{B}$ as its vertex set, where two vertices $B_{1} \\in \\mathcal{B}$\nand $B_{2} \\in \\mathcal{B}$ are adjacent if and only if $|B_{1} \\cap B_{2}| =\ni$. In this paper several constructions are obtained that start with twofold\ntriple systems (TTSs) with Hamiltonian $2$-BIGs and result in larger TTSs that\nalso have Hamiltonian $2$-BIGs. These constructions collectively enable us to\ndetermine the complete spectrum of TTSs with Hamiltonian $2$-BIGs (equivalently\nTTSs with cyclic $2$-intersecting Gray codes) as well as the complete spectrum\nfor TTSs with $2$-BIGs that have Hamilton paths (i.e., for TTSs with\n$2$-intersecting Gray codes).\nIn order to prove these spectrum results, we sometimes require ingredient\nTTSs that have large partial parallel classes; we prove lower bounds on the\nsizes of partial parallel clasess in arbitrary TTSs, and then construct larger\nTTSs with both cyclic $2$-intersecting Gray codes and parallel classes.\n",
"title": "Twofold triple systems with cyclic 2-intersecting Gray codes"
} | null | null | null | null | true | null | 2548 | null | Default | null | null |
null | {
"abstract": " The growing literature on affect among software developers mostly reports on\nthe linkage between happiness, software quality, and developer productivity.\nUnderstanding the positive side of happiness -- positive emotions and moods --\nis an attractive and important endeavor. Scholars in industrial and\norganizational psychology have suggested that also studying the negative side\n-- unhappiness -- could lead to cost-effective ways of enhancing working\nconditions, job performance, and to limiting the occurrence of psychological\ndisorders. Our comprehension of the consequences of (un)happiness among\ndevelopers is still too shallow, and is mainly expressed in terms of\ndevelopment productivity and software quality. In this paper, we attempt to\nuncover the experienced consequences of unhappiness among software developers.\nUsing qualitative data analysis of the responses given by 181 questionnaire\nparticipants, we identified 49 consequences of unhappiness while doing software\ndevelopment. We found detrimental consequences on developers' mental\nwell-being, the software development process, and the produced artifacts. Our\nclassification scheme, available as open data, will spawn new happiness\nresearch opportunities of cause-effect type, and it can act as a guideline for\npractitioners for identifying damaging effects of unhappiness and for fostering\nhappiness on the job.\n",
"title": "Consequences of Unhappiness While Developing Software"
} | null | null | [
"Computer Science"
]
| null | true | null | 2549 | null | Validated | null | null |
null | {
"abstract": " We propose a typesafe abstraction to tensors (i.e. multidimensional arrays)\nexploiting the type-level programming capabilities of Scala through\nheterogeneous lists (HList), and showcase typesafe abstractions of common\ntensor operations and various neural layers such as convolution or recurrent\nneural networks. This abstraction could lay the foundation of future typesafe\ndeep learning frameworks that runs on Scala/JVM.\n",
"title": "Typesafe Abstractions for Tensor Operations"
} | null | null | null | null | true | null | 2550 | null | Default | null | null |
null | {
"abstract": " Exploiting sparsity enables hardware systems to run neural networks faster\nand more energy-efficiently. However, most prior sparsity-centric optimization\ntechniques only accelerate the forward pass of neural networks and usually\nrequire an even longer training process with iterative pruning and retraining.\nWe observe that artificially inducing sparsity in the gradients of the gates in\nan LSTM cell has little impact on the training quality. Further, we can enforce\nstructured sparsity in the gate gradients to make the LSTM backward pass up to\n45% faster than the state-of-the-art dense approach and 168% faster than the\nstate-of-the-art sparsifying method on modern GPUs. Though the structured\nsparsifying method can impact the accuracy of a model, this performance gap can\nbe eliminated by mixing our sparse training method and the standard dense\ntraining method. Experimental results show that the mixed method can achieve\ncomparable results in a shorter time span than using purely dense training.\n",
"title": "Structurally Sparsified Backward Propagation for Faster Long Short-Term Memory Training"
} | null | null | null | null | true | null | 2551 | null | Default | null | null |
null | {
"abstract": " Density-based clustering techniques are used in a wide range of data mining\napplications. One of their most attractive features con- sists in not making\nuse of prior knowledge of the number of clusters that a dataset contains along\nwith their shape. In this paper we propose a new algorithm named Linear DBSCAN\n(Lin-DBSCAN), a simple approach to clustering inspired by the density model\nintroduced with the well known algorithm DBSCAN. Designed to minimize the\ncomputational cost of density based clustering on geospatial data, Lin-DBSCAN\nfeatures a linear time complexity that makes it suitable for real-time\napplications on low-resource devices. Lin-DBSCAN uses a discrete version of the\ndensity model of DBSCAN that takes ad- vantage of a grid-based scan and merge\napproach. The name of the algorithm stems exactly from its main features\noutlined above. The algorithm was tested with well known data sets.\nExperimental results prove the efficiency and the validity of this approach\nover DBSCAN in the context of spatial data clustering, enabling the use of a\ndensity-based clustering technique on large datasets with low computational\ncost.\n",
"title": "Linear density-based clustering with a discrete density model"
} | null | null | null | null | true | null | 2552 | null | Default | null | null |
null | {
"abstract": " We propose a fast proximal Newton-type algorithm for minimizing regularized\nfinite sums that returns an $\\epsilon$-suboptimal point in\n$\\tilde{\\mathcal{O}}(d(n + \\sqrt{\\kappa d})\\log(\\frac{1}{\\epsilon}))$ FLOPS,\nwhere $n$ is number of samples, $d$ is feature dimension, and $\\kappa$ is the\ncondition number. As long as $n > d$, the proposed method is more efficient\nthan state-of-the-art accelerated stochastic first-order methods for non-smooth\nregularizers which requires $\\tilde{\\mathcal{O}}(d(n + \\sqrt{\\kappa\nn})\\log(\\frac{1}{\\epsilon}))$ FLOPS. The key idea is to form the subsampled\nNewton subproblem in a way that preserves the finite sum structure of the\nobjective, thereby allowing us to leverage recent developments in stochastic\nfirst-order methods to solve the subproblem. Experimental results verify that\nthe proposed algorithm outperforms previous algorithms for $\\ell_1$-regularized\nlogistic regression on real datasets.\n",
"title": "An inexact subsampled proximal Newton-type method for large-scale machine learning"
} | null | null | null | null | true | null | 2553 | null | Default | null | null |
null | {
"abstract": " We use grey forecast model to predict the future energy consumption of four\nstates in the U.S, and make some improvments to the model.\n",
"title": "Future Energy Consumption Prediction Based on Grey Forecast Model"
} | null | null | [
"Statistics"
]
| null | true | null | 2554 | null | Validated | null | null |
null | {
"abstract": " Text password has long been the dominant user authentication technique and is\nused by large numbers of Internet services. If they follow recommended\npractice, users are faced with the almost insuperable problem of generating and\nmanaging a large number of site-unique and strong (i.e. non-guessable)\npasswords. One way of addressing this problem is through the use of a password\ngenerator, i.e. a client-side scheme which generates (and regenerates)\nsite-specific strong passwords on demand, with the minimum of user input. This\npaper provides a detailed specification and analysis of AutoPass, a password\ngenerator scheme previously outlined as part of a general analysis of such\nschemes. AutoPass has been designed to address issues identified in previously\nproposed password generators, and incorporates novel techniques to address\nthese issues. Unlike almost all previously proposed schemes, AutoPass enables\nthe generation of passwords that meet important real-world requirements,\nincluding forced password changes, use of pre-specified passwords, and\ngeneration of passwords meeting site-specific requirements.\n",
"title": "AutoPass: An Automatic Password Generator"
} | null | null | null | null | true | null | 2555 | null | Default | null | null |
null | {
"abstract": " Randomized experiments have been critical tools of decision making for\ndecades. However, subjects can show significant heterogeneity in response to\ntreatments in many important applications. Therefore it is not enough to simply\nknow which treatment is optimal for the entire population. What we need is a\nmodel that correctly customize treatment assignment base on subject\ncharacteristics. The problem of constructing such models from randomized\nexperiments data is known as Uplift Modeling in the literature. Many algorithms\nhave been proposed for uplift modeling and some have generated promising\nresults on various data sets. Yet little is known about the theoretical\nproperties of these algorithms. In this paper, we propose a new tree-based\nensemble algorithm for uplift modeling. Experiments show that our algorithm can\nachieve competitive results on both synthetic and industry-provided data. In\naddition, by properly tuning the \"node size\" parameter, our algorithm is proved\nto be consistent under mild regularity conditions. This is the first consistent\nalgorithm for uplift modeling that we are aware of.\n",
"title": "A Practically Competitive and Provably Consistent Algorithm for Uplift Modeling"
} | null | null | null | null | true | null | 2556 | null | Default | null | null |
null | {
"abstract": " The planets of the Solar System divide neatly between those with atmospheres\nand those without when arranged by insolation ($I$) and escape velocity\n($v_{\\mathrm{esc}}$). The dividing line goes as $I \\propto v_{\\mathrm{esc}}^4$.\nExoplanets with reported masses and radii are shown to crowd against the\nextrapolation of the Solar System trend, making a metaphorical cosmic shoreline\nthat unites all the planets. The $I \\propto v_{\\mathrm{esc}}^4$ relation may\nimplicate thermal escape. We therefore address the general behavior of\nhydrodynamic thermal escape models ranging from Pluto to highly-irradiated\nExtrasolar Giant Planets (EGPs). Energy-limited escape is harder to test\nbecause copious XUV radiation is mostly a feature of young stars, and hence\nrequires extrapolating to historic XUV fluences ($I_{\\mathrm{xuv}}$) using\nproxies and power laws. An energy-limited shoreline should scale as\n$I_{\\mathrm{xuv}} \\propto v_{\\mathrm{esc}}^3\\sqrt{\\rho}$, which differs\ndistinctly from the apparent $I_{\\mathrm{xuv}} \\propto v_{\\mathrm{esc}}^4$\nrelation. Energy-limited escape does provide good quantitative agreement to the\nhighly irradiated EGPs. Diffusion-limited escape implies that no planet can\nlose more than 1% of its mass as H$_2$. Impact erosion, to the extent that\nimpact velocities $v_{\\mathrm{imp}}$ can be estimated for exoplanets, fits to a\n$v_{\\mathrm{imp}} \\approx 4\\,-\\,5\\, v_{\\mathrm{esc}}$ shoreline. The\nproportionality constant is consistent with what the collision of comet\nShoemaker-Levy 9 showed us we should expect of modest impacts in deep\natmospheres. With respect to the shoreline, Proxima Centauri b is on the\nmetaphorical beach. Known hazards include its rapid energetic accretion, high\nimpact velocities, its early life on the wrong side of the runaway greenhouse,\nand Proxima Centauri's XUV radiation. In its favor is a vast phase space of\nunknown unknowns.\n",
"title": "The cosmic shoreline: the evidence that escape determines which planets have atmospheres, and what this may mean for Proxima Centauri b"
} | null | null | null | null | true | null | 2557 | null | Default | null | null |
null | {
"abstract": " When plated onto substrates, cell morphology and even stem cell\ndifferentiation are influenced by the stiffness of their environment. Stiffer\nsubstrates give strongly spread (eventually polarized) cells with strong focal\nadhesions, and stress fibers; very soft substrates give a less developed\ncytoskeleton, and much lower cell spreading. The kinetics of this process of\ncell spreading is studied extensively, and important universal relationships\nare established on how the cell area grows with time. Here we study the\npopulation dynamics of spreading cells, investigating the characteristic\nprocesses involved in cell response to the substrate. We show that unlike the\nindividual cell morphology, this population dynamics does not depend on the\nsubstrate stiffness. Instead, a strong activation temperature dependence is\nobserved. Different cell lines on different substrates all have long-time\nstatistics controlled by the thermal activation over a single energy barrier\ndG=19 kcal/mol, while the early-time kinetics follows a power law $t^5$. This\nimplies that the rate of spreading depends on an internal process of\nadhesion-mechanosensing complex assembly and activation: the operational\ncomplex must have 5 component proteins, and the last process in the sequence\n(which we believe is the activation of focal adhesion kinase) is controlled by\nthe binding energy dG.\n",
"title": "Universal kinetics for engagement of mechanosensing pathways in cell adhesion"
} | null | null | null | null | true | null | 2558 | null | Default | null | null |
null | {
"abstract": " Agent-Based Computing is a diverse research domain concerned with the\nbuilding of intelligent software based on the concept of \"agents\". In this\npaper, we use Scientometric analysis to analyze all sub-domains of agent-based\ncomputing. Our data consists of 1,064 journal articles indexed in the ISI web\nof knowledge published during a twenty year period: 1990-2010. These were\nretrieved using a topic search with various keywords commonly used in\nsub-domains of agent-based computing. In our proposed approach, we have\nemployed a combination of two applications for analysis, namely Network\nWorkbench and CiteSpace - wherein Network Workbench allowed for the analysis of\ncomplex network aspects of the domain, detailed visualization-based analysis of\nthe bibliographic data was performed using CiteSpace. Our results include the\nidentification of the largest cluster based on keywords, the timeline of\npublication of index terms, the core journals and key subject categories. We\nalso identify the core authors, top countries of origin of the manuscripts\nalong with core research institutes. Finally, our results have interestingly\nrevealed the strong presence of agent-based computing in a number of\nnon-computing related scientific domains including Life Sciences, Ecological\nSciences and Social Sciences.\n",
"title": "Agent-based computing from multi-agent systems to agent-based Models: a visual survey"
} | null | null | [
"Computer Science",
"Physics"
]
| null | true | null | 2559 | null | Validated | null | null |
null | {
"abstract": " As novel topological phases in correlated electron systems, we have found two\nexamples of non-ferromagnetic states that exhibit a large anomalous Hall\neffect. One is the chiral spin liquid compound Pr$_{2}$Ir$_{2}$O$_{7}$, which\nexhibits a spontaneous Hall effect in a spin liquid state due to spin ice\ncorrelation. The other is the chiral antiferromagnets Mn$_{3}$Sn and Mn$_{3}$Ge\nthat exhibit a large anomalous Hall effect at room temperature. The latter\nshows a sign change of the anomalous Hall effect by a small change in the\nmagnetic field by a few 100 G, which should be useful for various applications.\nWe will discuss that the magnetic Weyl metal states are the origin for such a\nlarge anomalous Hall effect observed in both the spin liquid and\nantiferromagnet that possess almost no magnetization.\n",
"title": "Large Spontaneous Hall Effects in Chiral Topological Magnets"
} | null | null | null | null | true | null | 2560 | null | Default | null | null |
null | {
"abstract": " Motivated by the current interest in the understanding of the Mott insulators\naway from half filling, observed in many perovskite oxides, we study the Mott\nmetal-insulator transition (MIT) in the doped Hubbard-Holstein model using the\nHatree-Fock mean field theory. The Hubbard-Holstein model is the simplest model\ncontaining both the Coulomb and the electron-lattice interactions, which are\nimportant ingredients in the physics of the perovskite oxides. In contrast to\nthe half-filled Hubbard model, which always results in a single phase (either\nmetallic or insulating), our results show that away from half-filling, a mixed\nphase of metallic and insulating regions occur. As the dopant concentration is\nincreased, the metallic part progressively grows in volume, until it exceeds\nthe percolation threshold, leading to percolative conduction. This happens\nabove a critical dopant concentration $\\delta_c$, which, depending on the\nstrength of the electron-lattice interaction, can be a significant fraction of\nunity. This means that the material could be insulating even for a substantial\namount of doping, in contrast to the expectation that doped holes would destroy\nthe insulating behavior of the half-filled Hubbard model. Our theory provides a\nframework for the understanding of the density-driven metal-insulator\ntransition observed in many complex oxides.\n",
"title": "Mott metal-insulator transition in the Doped Hubbard-Holstein model"
} | null | null | null | null | true | null | 2561 | null | Default | null | null |
null | {
"abstract": " Opinion mining and sentiment analysis in social media is a research issue\nhaving a great interest in the scientific community. However, before begin this\nanalysis, we are faced with a set of problems. In particular, the problem of\nthe richness of languages and dialects within these media. To address this\nproblem, we propose in this paper an approach of construction and\nimplementation of Syntactic analyzer named ASDA. This tool represents a parser\nfor the Algerian dialect that label the terms of a given corpus. Thus, we\nconstruct a labeling table containing for each term its stem, different\nprefixes and suffixes, allowing us to determine the different grammatical parts\na sort of POS tagging. This labeling will serve us later in the semantic\nprocessing of the Algerian dialect, like the automatic translation of this\ndialect or sentiment analysis\n",
"title": "ASDA : Analyseur Syntaxique du Dialecte Alg{é}rien dans un but d'analyse s{é}mantique"
} | null | null | null | null | true | null | 2562 | null | Default | null | null |
null | {
"abstract": " Developing a dialogue agent that is capable of making autonomous decisions\nand communicating by natural language is one of the long-term goals of machine\nlearning research. Traditional approaches either rely on hand-crafting a small\nstate-action set for applying reinforcement learning that is not scalable or\nconstructing deterministic models for learning dialogue sentences that fail to\ncapture natural conversational variability. In this paper, we propose a Latent\nIntention Dialogue Model (LIDM) that employs a discrete latent variable to\nlearn underlying dialogue intentions in the framework of neural variational\ninference. In a goal-oriented dialogue scenario, these latent intentions can be\ninterpreted as actions guiding the generation of machine responses, which can\nbe further refined autonomously by reinforcement learning. The experimental\nevaluation of LIDM shows that the model out-performs published benchmarks for\nboth corpus-based and human evaluation, demonstrating the effectiveness of\ndiscrete latent variable models for learning goal-oriented dialogues.\n",
"title": "Latent Intention Dialogue Models"
} | null | null | null | null | true | null | 2563 | null | Default | null | null |
null | {
"abstract": " A weak-strong uniqueness result is proved for measure-valued solutions to the\nsystem of conservation laws arising in elastodynamics. The main novelty brought\nforward by the present work is that the underlying stored-energy function of\nthe material is assumed strongly quasiconvex. The proof employs tools from the\ncalculus of variations to establish general convexity-type bounds on\nquasiconvex functions and recasts them in order to adapt the relative entropy\nmethod to quasiconvex elastodynamics.\n",
"title": "Quasiconvex elastodynamics: weak-strong uniqueness for measure-valued solutions"
} | null | null | null | null | true | null | 2564 | null | Default | null | null |
null | {
"abstract": " Gradient descent and coordinate descent are well understood in terms of their\nasymptotic behavior, but less so in a transient regime often used for\napproximations in machine learning. We investigate how proper initialization\ncan have a profound effect on finding near-optimal solutions quickly. We show\nthat a certain property of a data set, namely the boundedness of the\ncorrelations between eigenfeatures and the response variable, can lead to\nfaster initial progress than expected by commonplace analysis. Convex\noptimization problems can tacitly benefit from that, but this automatism does\nnot apply to their dual formulation. We analyze this phenomenon and devise\nprovably good initialization strategies for dual optimization as well as\nheuristics for the non-convex case, relevant for deep learning. We find our\npredictions and methods to be experimentally well-supported.\n",
"title": "Accelerated Dual Learning by Homotopic Initialization"
} | null | null | null | null | true | null | 2565 | null | Default | null | null |
null | {
"abstract": " Inverse reinforcement learning (IRL) aims to explain observed strategic\nbehavior by fitting reinforcement learning models to behavioral data. However,\ntraditional IRL methods are only applicable when the observations are in the\nform of state-action paths. This assumption may not hold in many real-world\nmodeling settings, where only partial or summarized observations are available.\nIn general, we may assume that there is a summarizing function $\\sigma$, which\nacts as a filter between us and the true state-action paths that constitute the\ndemonstration. Some initial approaches to extending IRL to such situations have\nbeen presented, but with very specific assumptions about the structure of\n$\\sigma$, such as that only certain state observations are missing. This paper\ninstead focuses on the most general case of the problem, where no assumptions\nare made about the summarizing function, except that it can be evaluated. We\ndemonstrate that inference is still possible. The paper presents exact and\napproximate inference algorithms that allow full posterior inference, which is\nparticularly important for assessing parameter uncertainty in this challenging\ninference situation. Empirical scalability is demonstrated to reasonably sized\nproblems, and practical applicability is demonstrated by estimating the\nposterior for a cognitive science RL model based on an observed user's task\ncompletion time only.\n",
"title": "Inverse Reinforcement Learning from Summary Data"
} | null | null | null | null | true | null | 2566 | null | Default | null | null |
null | {
"abstract": " On one hand, consider the problem of finding global solutions to a polynomial\noptimization problem and, on the other hand, consider the problem of\ninterpolating a set of points with a complex exponential function. This paper\nproposes a single algorithm to address both problems. It draws on the notion of\nhyponormality in operator theory. Concerning optimization, it seems to be the\nfirst algorithm that is capable of extracting global solutions from a\npolynomial optimization problem where the variables and data are complex\nnumbers. It also applies to real polynomial optimization, a special case of\ncomplex polynomial optimization, and thus extends the work of Henrion and\nLasserre implemented in GloptiPoly. Concerning interpolation, the algorithm\nprovides an alternative to Prony's method based on the Autonne-Takagi\nfactorization and it avoids solving a Vandermonde system. The algorithm and its\nproof are based exclusively on linear algebra. They are devoid of notions from\nalgebraic geometry, contrary to existing methods for interpolation. The\nalgorithm is tested on a series of examples, each illustrating a different\nfacet of the approach. One of the examples demonstrates that hyponormality can\nbe enforced numerically to strenghten a convex relaxation and to force its\nsolution to have rank one.\n",
"title": "Algorithm for Optimization and Interpolation based on Hyponormality"
} | null | null | null | null | true | null | 2567 | null | Default | null | null |
null | {
"abstract": " The step of expert taxa recognition currently slows down the response time of\nmany bioassessments. Shifting to quicker and cheaper state-of-the-art machine\nlearning approaches is still met with expert scepticism towards the ability and\nlogic of machines. In our study, we investigate both the differences in\naccuracy and in the identification logic of taxonomic experts and machines. We\npropose a systematic approach utilizing deep Convolutional Neural Nets with the\ntransfer learning paradigm and extensively evaluate it over a multi-label and\nmulti-pose taxonomic dataset specifically created for this comparison. We also\nstudy the prediction accuracy on different ranks of taxonomic hierarchy in\ndetail. Our results revealed that human experts using actual specimens yield\nthe lowest classification error. However, our proposed, much faster, automated\napproach using deep Convolutional Neural Nets comes very close to human\naccuracy. Contrary to previous findings in the literature, we find that\nmachines following a typical flat classification approach commonly used in\nmachine learning performs better than forcing machines to adopt a hierarchical,\nlocal per parent node approach used by human taxonomic experts. Finally, we\npublicly share our unique dataset to serve as a public benchmark dataset in\nthis field.\n",
"title": "Human experts vs. machines in taxa recognition"
} | null | null | null | null | true | null | 2568 | null | Default | null | null |
null | {
"abstract": " Thermoelectric (TE) materials achieve localised conversion between thermal\nand electric energies, and the conversion efficiency is determined by a figure\nof merit zT. Up to date, two-dimensional electron gas (2DEG) related TE\nmaterials hold the records for zT near room-temperature. A sharp increase in zT\nup to ~2.0 was observed previously for superlattice materials such as PbSeTe,\nBi2Te3/Sb2Te3 and SrNb0.2Ti0.8O3/SrTiO3, when the thicknesses of these TE\nmaterials were spatially confine within sub-nanometre scale. The\ntwo-dimensional confinement of carriers enlarges the density of states near the\nFermi energy3-6 and triggers electron phonon coupling. This overcomes the\nconventional {\\sigma}-S trade-off to more independently improve S, and thereby\nfurther increases thermoelectric power factors (PF=S2{\\sigma}). Nevertheless,\npractical applications of the present 2DEG materials for high power energy\nconversions are impeded by the prerequisite of spatial confinement, as the\namount of TE material is insufficient. Here, we report similar TE properties to\n2DEGs but achieved in SrNb0.2Ti0.8O3 films with thickness within sub-micrometer\nscale by regulating interfacial and lattice polarizations. High power factor\n(up to 103 {\\mu}Wcm-1K-2) and zT value (up to 1.6) were observed for the film\nmaterials near room-temperature and below. Even reckon in the thickness of the\nsubstrate, an integrated power factor of both film and substrate approaching to\nbe 102 {\\mu}Wcm-1K-2 was achieved in a 2 {\\mu}m-thick SrNb0.2Ti0.8O3 film grown\non a 100 {\\mu}m-thick SrTiO3 substrate. The dependence of high TE performances\non size-confinement is reduced by ~103 compared to the conventional\n2DEG-related TE materials. As-grown oxide films are less toxic and not\ndependent on large amounts of heavy elements, potentially paving the way\ntowards applications in localised refrigeration and electric power generations.\n",
"title": "A micrometer-thick oxide film with high thermoelectric performance at temperature ranging from 20-400 K"
} | null | null | null | null | true | null | 2569 | null | Default | null | null |
null | {
"abstract": " Over 50 million scholarly articles have been published: they constitute a\nunique repository of knowledge. In particular, one may infer from them\nrelations between scientific concepts, such as synonyms and hyponyms.\nArtificial neural networks have been recently explored for relation extraction.\nIn this work, we continue this line of work and present a system based on a\nconvolutional neural network to extract relations. Our model ranked first in\nthe SemEval-2017 task 10 (ScienceIE) for relation extraction in scientific\narticles (subtask C).\n",
"title": "MIT at SemEval-2017 Task 10: Relation Extraction with Convolutional Neural Networks"
} | null | null | null | null | true | null | 2570 | null | Default | null | null |
null | {
"abstract": " Bayesian optimization is a sample-efficient approach to global optimization\nthat relies on theoretically motivated value heuristics (acquisition functions)\nto guide its search process. Fully maximizing acquisition functions produces\nthe Bayes' decision rule, but this ideal is difficult to achieve since these\nfunctions are frequently non-trivial to optimize. This statement is especially\ntrue when evaluating queries in parallel, where acquisition functions are\nroutinely non-convex, high-dimensional, and intractable. We first show that\nacquisition functions estimated via Monte Carlo integration are consistently\namenable to gradient-based optimization. Subsequently, we identify a common\nfamily of acquisition functions, including EI and UCB, whose properties not\nonly facilitate but justify use of greedy approaches for their maximization.\n",
"title": "Maximizing acquisition functions for Bayesian optimization"
} | null | null | [
"Statistics"
]
| null | true | null | 2571 | null | Validated | null | null |
null | {
"abstract": " We present a MUSE and KMOS dynamical study 405 star-forming galaxies at\nredshift z=0.28-1.65 (median redshift z=0.84). Our sample are representative of\nstar-forming, main-sequence galaxies, with star-formation rates of\nSFR=0.1-30Mo/yr and stellar masses M=10^8-10^11Mo. For 49+/-4% of our sample,\nthe dynamics suggest rotational support, 24+/-3% are unresolved systems and\n5+/-2% appear to be early-stage major mergers with components on 8-30kpc\nscales. The remaining 22+/-5% appear to be dynamically complex, irregular (or\nface-on systems). For galaxies whose dynamics suggest rotational support, we\nderive inclination corrected rotational velocities and show these systems lie\non a similar scaling between stellar mass and specific angular momentum as\nlocal spirals with j*=J/M*\\propto M^(2/3) but with a redshift evolution that\nscales as j*\\propto M^{2/3}(1+z)^(-1). We identify a correlation between\nspecific angular momentum and disk stability such that galaxies with the\nhighest specific angular momentum, log(j*/M^(2/3))>2.5, are the most stable,\nwith Toomre Q=1.10+/-0.18, compared to Q=0.53+/-0.22 for galaxies with\nlog(j*/M^(2/3))<2.5. At a fixed mass, the HST morphologies of galaxies with the\nhighest specific angular momentum resemble spiral galaxies, whilst those with\nlow specific angular momentum are morphologically complex and dominated by\nseveral bright star-forming regions. This suggests that angular momentum plays\na major role in defining the stability of gas disks: at z~1, massive galaxies\nthat have disks with low specific angular momentum, appear to be globally\nunstable, clumpy and turbulent systems. In contrast, galaxies with high\nspecific angular have evolved in to stable disks with spiral structures.\n",
"title": "Angular momentum evolution of galaxies over the past 10-Gyr: A MUSE and KMOS dynamical survey of 400 star-forming galaxies from z=0.3-1.7"
} | null | null | null | null | true | null | 2572 | null | Default | null | null |
null | {
"abstract": " The aim of fine-grained recognition is to identify sub-ordinate categories in\nimages like different species of birds. Existing works have confirmed that, in\norder to capture the subtle differences across the categories, automatic\nlocalization of objects and parts is critical. Most approaches for object and\npart localization relied on the bottom-up pipeline, where thousands of region\nproposals are generated and then filtered by pre-trained object/part models.\nThis is computationally expensive and not scalable once the number of\nobjects/parts becomes large. In this paper, we propose a nonparametric\ndata-driven method for object and part localization. Given an unlabeled test\nimage, our approach transfers annotations from a few similar images retrieved\nin the training set. In particular, we propose an iterative transfer strategy\nthat gradually refine the predicted bounding boxes. Based on the located\nobjects and parts, deep convolutional features are extracted for recognition.\nWe evaluate our approach on the widely-used CUB200-2011 dataset and a new and\nlarge dataset called Birdsnap. On both datasets, we achieve better results than\nmany state-of-the-art approaches, including a few using oracle (manually\nannotated) bounding boxes in the test images.\n",
"title": "Iterative Object and Part Transfer for Fine-Grained Recognition"
} | null | null | null | null | true | null | 2573 | null | Default | null | null |
null | {
"abstract": " There are many hard conjectures in graph theory, like Tutte's 5-flow\nconjecture, and the 5-cycle double cover conjecture, which would be true in\ngeneral if they would be true for cubic graphs. Since most of them are\ntrivially true for 3-edge-colorable cubic graphs, cubic graphs which are not\n3-edge-colorable, often called {\\em snarks}, play a key role in this context.\nHere, we survey parameters measuring how far apart a non 3-edge-colorable graph\nis from being 3-edge-colorable. We study their interrelation and prove some new\nresults. Besides getting new insight into the structure of snarks, we show that\nsuch measures give partial results with respect to these important conjectures.\nThe paper closes with a list of open problems and conjectures.\n",
"title": "On measures of edge-uncolorability of cubic graphs: A brief survey and some new results"
} | null | null | null | null | true | null | 2574 | null | Default | null | null |
null | {
"abstract": " Modern theories of galaxy formation predict that galaxies impact on their\ngaseous surroundings, playing the fundamental role of regulating the amount of\ngas converted into stars. While star-forming galaxies are believed to provide\nfeedback through galactic winds, Quasi-Stellar Objects (QSOs) are believed\ninstead to provide feedback through the heat generated by accretion onto a\ncentral supermassive black hole. A quantitative difference in the impact of\nfeedback on the gaseous environments of star-forming galaxies and QSOs has not\nbeen established through direct observations. Using the Sherwood cosmological\nsimulations, we demonstrate that measurements of neutral hydrogen in the\nvicinity of star-forming galaxies and QSOs during the era of peak galaxy\nformation show excess LyA absorption extending up to comoving radii of about\n150 kpc for star-forming galaxies and 300 - 700 kpc for QSOs. Simulations\nincluding supernovae-driven winds with the wind velocity scaling like the\nescape velocity of the halo account for the absorption around star-forming\ngalaxies but not QSOs.\n",
"title": "Gas around galaxy haloes - III: hydrogen absorption signatures around galaxies and QSOs in the Sherwood simulation suite"
} | null | null | null | null | true | null | 2575 | null | Default | null | null |
null | {
"abstract": " Deep learning involves a difficult non-convex optimization problem with a\nlarge number of weights between any two adjacent layers of a deep structure. To\nhandle large data sets or complicated networks, distributed training is needed,\nbut the calculation of function, gradient, and Hessian is expensive. In\nparticular, the communication and the synchronization cost may become a\nbottleneck. In this paper, we focus on situations where the model is\ndistributedly stored, and propose a novel distributed Newton method for\ntraining deep neural networks. By variable and feature-wise data partitions,\nand some careful designs, we are able to explicitly use the Jacobian matrix for\nmatrix-vector products in the Newton method. Some techniques are incorporated\nto reduce the running time as well as the memory consumption. First, to reduce\nthe communication cost, we propose a diagonalization method such that an\napproximate Newton direction can be obtained without communication between\nmachines. Second, we consider subsampled Gauss-Newton matrices for reducing the\nrunning time as well as the communication cost. Third, to reduce the\nsynchronization cost, we terminate the process of finding an approximate Newton\ndirection even though some nodes have not finished their tasks. Details of some\nimplementation issues in distributed environments are thoroughly investigated.\nExperiments demonstrate that the proposed method is effective for the\ndistributed training of deep neural networks. In compared with stochastic\ngradient methods, it is more robust and may give better test accuracy.\n",
"title": "Distributed Newton Methods for Deep Neural Networks"
} | null | null | null | null | true | null | 2576 | null | Default | null | null |
null | {
"abstract": " The attention for personalized mental health care is thriving. Research data\nspecific to the individual, such as time series sensor data or data from\nintensive longitudinal studies, is relevant from a research perspective, as\nanalyses on these data can reveal the heterogeneity among the participants and\nprovide more precise and individualized results than with group-based methods.\nHowever, using this data for self-management and to help the individual to\nimprove his or her mental health has proven to be challenging.\nThe present work describes a novel approach to automatically generate\npersonalized advice for the improvement of the well-being of individuals by\nusing time series data from intensive longitudinal studies: Automated Impulse\nResponse Analysis (AIRA). AIRA analyzes vector autoregression models of\nwell-being by generating impulse response functions. These impulse response\nfunctions are used in simulations to determine which variables in the model\nhave the largest influence on the other variables and thus on the well-being of\nthe participant. The effects found can be used to support self-management.\nWe demonstrate the practical usefulness of AIRA by performing analysis on\nlongitudinal self-reported data about psychological variables. To evaluate its\neffectiveness and efficacy, we ran its algorithms on two data sets ($N=4$ and\n$N=5$), and discuss the results. Furthermore, we compare AIRA's output to the\nresults of a previously published study and show that the results are\ncomparable. By automating Impulse Response Function Analysis, AIRA fulfills the\nneed for accurate individualized models of health outcomes at a low resource\ncost with the potential for upscaling.\n",
"title": "Personalized advice for enhancing well-being using automated impulse response analysis --- AIRA"
} | null | null | null | null | true | null | 2577 | null | Default | null | null |
null | {
"abstract": " Robust estimation is much more challenging in high dimensions than it is in\none dimension: Most techniques either lead to intractable optimization problems\nor estimators that can tolerate only a tiny fraction of errors. Recent work in\ntheoretical computer science has shown that, in appropriate distributional\nmodels, it is possible to robustly estimate the mean and covariance with\npolynomial time algorithms that can tolerate a constant fraction of\ncorruptions, independent of the dimension. However, the sample and time\ncomplexity of these algorithms is prohibitively large for high-dimensional\napplications. In this work, we address both of these issues by establishing\nsample complexity bounds that are optimal, up to logarithmic factors, as well\nas giving various refinements that allow the algorithms to tolerate a much\nlarger fraction of corruptions. Finally, we show on both synthetic and real\ndata that our algorithms have state-of-the-art performance and suddenly make\nhigh-dimensional robust estimation a realistic possibility.\n",
"title": "Being Robust (in High Dimensions) Can Be Practical"
} | null | null | [
"Computer Science",
"Statistics"
]
| null | true | null | 2578 | null | Validated | null | null |
null | {
"abstract": " The graphene/MoS2 heterojunction formed by joining the two components\nlaterally in a single plane promises to exhibit a low-resistance contact\naccording to the Schottky-Mott rule. Here we provide an atomic-scale\ndescription of the structural, electronic, and magnetic properties of this type\nof junction. We first identify the energetically favorable structures in which\nthe preference of forming C-S or C-Mo bonds at the boundary depends on the\nchemical conditions. We find that significant charge transfer between graphene\nand MoS2 is localized at the boundary. We show that the abundant 1D boundary\nstates substantially pin the Fermi level in the lateral contact between\ngraphene and MoS2, in close analogy to the effect of 2D interfacial states in\nthe contacts between 3D materials. Furthermore, we propose specific ways in\nwhich these effects can be exploited to achieve spin-polarized currents.\n",
"title": "Properties of In-Plane Graphene/MoS2 Heterojunctions"
} | null | null | [
"Physics"
]
| null | true | null | 2579 | null | Validated | null | null |
null | {
"abstract": " Let $p$ be a prime. A $p$-group $G$ is defined to be semi-extraspecial if for\nevery maximal subgroup $N$ in $Z(G)$ the quotient $G/N$ is a an extraspecial\ngroup. In addition, we say that $G$ is ultraspecial if $G$ is semi-extraspecial\nand $|G:G'| = |G'|^2$. In this paper, we prove that every $p$-group of\nnilpotence class $2$ is isomorphic to a subgroup of some ultraspecial group.\nGiven a prime $p$ and a positive integer $n$, we provide a framework to\nconstruct of all the ultraspecial groups order $p^{3n}$ that contain an abelian\nsubgroup of order $p^{2n}$. In the literature, it has been proved that every\nultraspecial group $G$ order $p^{3n}$ with at least two abelian subgroups of\norder $p^{2n}$ can be associated to a semifield. We provide a generalization of\nsemifield, and then we show that every semi-extraspecial group $G$ that is the\nproduct of two abelian subgroups can be associated with this generalization of\nsemifield.\n",
"title": "Semi-extraspecial groups with an abelian subgroup of maximal possible order"
} | null | null | null | null | true | null | 2580 | null | Default | null | null |
null | {
"abstract": " The lack of diversity in a genetic algorithm's population may lead to a bad\nperformance of the genetic operators since there is not an equilibrium between\nexploration and exploitation. In those cases, genetic algorithms present a fast\nand unsuitable convergence.\nIn this paper we develop a novel hybrid genetic algorithm which attempts to\nobtain a balance between exploration and exploitation. It confronts the\ndiversity problem using the named greedy diversification operator. Furthermore,\nthe proposed algorithm applies a competition between parent and children so as\nto exploit the high quality visited solutions. These operators are complemented\nby a simple selection mechanism designed to preserve and take advantage of the\npopulation diversity.\nAdditionally, we extend our proposal to the field of memetic algorithms,\nobtaining an improved model with outstanding results in practice.\nThe experimental study shows the validity of the approach as well as how\nimportant is taking into account the exploration and exploitation concepts when\ndesigning an evolutionary algorithm.\n",
"title": "Genetic and Memetic Algorithm with Diversity Equilibrium based on Greedy Diversification"
} | null | null | null | null | true | null | 2581 | null | Default | null | null |
null | {
"abstract": " In this work, we analyze the problem of adoption of mobile money in Pakistan\nby using the call detail records of a major telecom company as our input. Our\nresults highlight the fact that different sections of the society have\ndifferent patterns of adoption of digital financial services but user mobility\nrelated features are the most important one when it comes to adopting and using\nmobile money services.\n",
"title": "Determinants of Mobile Money Adoption in Pakistan"
} | null | null | null | null | true | null | 2582 | null | Default | null | null |
null | {
"abstract": " We prove Cherlin's conjecture, concerning binary primitive permutation\ngroups, for those groups with socle isomorphic to $\\mathrm{PSL}_2(q)$,\n${^2\\mathrm{B}_2}(q)$, ${^2\\mathrm{G}_2}(q)$ or $\\mathrm{PSU}_3(q)$. Our method\nuses the notion of a \"strongly non-binary action\".\n",
"title": "Cherlin's conjecture for almost simple groups of Lie rank 1"
} | null | null | [
"Mathematics"
]
| null | true | null | 2583 | null | Validated | null | null |
null | {
"abstract": " This article concerns the expressive power of depth in neural nets with ReLU\nactivations and bounded width. We are particularly interested in the following\nquestions: what is the minimal width $w_{\\text{min}}(d)$ so that ReLU nets of\nwidth $w_{\\text{min}}(d)$ (and arbitrary depth) can approximate any continuous\nfunction on the unit cube $[0,1]^d$ aribitrarily well? For ReLU nets near this\nminimal width, what can one say about the depth necessary to approximate a\ngiven function? Our approach to this paper is based on the observation that,\ndue to the convexity of the ReLU activation, ReLU nets are particularly\nwell-suited for representing convex functions. In particular, we prove that\nReLU nets with width $d+1$ can approximate any continuous convex function of\n$d$ variables arbitrarily well. These results then give quantitative depth\nestimates for the rate of approximation of any continuous scalar function on\nthe $d$-dimensional cube $[0,1]^d$ by ReLU nets with width $d+3.$\n",
"title": "Universal Function Approximation by Deep Neural Nets with Bounded Width and ReLU Activations"
} | null | null | null | null | true | null | 2584 | null | Default | null | null |
null | {
"abstract": " Jiří Matoušek (1963-2015) had many breakthrough contributions in\nmathematics and algorithm design. His milestone results are not only profound\nbut also elegant. By going beyond the original objects --- such as Euclidean\nspaces or linear programs --- Jirka found the essence of the challenging\nmathematical/algorithmic problems as well as beautiful solutions that were\nnatural to him, but were surprising discoveries to the field.\nIn this short exploration article, I will first share with readers my initial\nencounter with Jirka and discuss one of his fundamental geometric results from\nthe early 1990s. In the age of social and information networks, I will then\nturn the discussion from geometric structures to network structures, attempting\nto take a humble step towards the holy grail of network science, that is to\nunderstand the network essence that underlies the observed\nsparse-and-multifaceted network data. I will discuss a simple result which\nsummarizes some basic algebraic properties of personalized PageRank matrices.\nUnlike the traditional transitive closure of binary relations, the personalized\nPageRank matrices take \"accumulated Markovian closure\" of network data. Some of\nthese algebraic properties are known in various contexts. But I hope featuring\nthem together in a broader context will help to illustrate the desirable\nproperties of this Markovian completion of networks, and motivate systematic\ndevelopments of a network theory for understanding vast and ubiquitous\nmultifaceted network data.\n",
"title": "Network Essence: PageRank Completion and Centrality-Conforming Markov Chains"
} | null | null | [
"Computer Science",
"Statistics"
]
| null | true | null | 2585 | null | Validated | null | null |
null | {
"abstract": " Modern neural networks are often augmented with an attention mechanism, which\ntells the network where to focus within the input. We propose in this paper a\nnew framework for sparse and structured attention, building upon a smoothed max\noperator. We show that the gradient of this operator defines a mapping from\nreal values to probabilities, suitable as an attention mechanism. Our framework\nincludes softmax and a slight generalization of the recently-proposed sparsemax\nas special cases. However, we also show how our framework can incorporate\nmodern structured penalties, resulting in more interpretable attention\nmechanisms, that focus on entire segments or groups of an input. We derive\nefficient algorithms to compute the forward and backward passes of our\nattention mechanisms, enabling their use in a neural network trained with\nbackpropagation. To showcase their potential as a drop-in replacement for\nexisting ones, we evaluate our attention mechanisms on three large-scale tasks:\ntextual entailment, machine translation, and sentence summarization. Our\nattention mechanisms improve interpretability without sacrificing performance;\nnotably, on textual entailment and summarization, we outperform the standard\nattention mechanisms based on softmax and sparsemax.\n",
"title": "A Regularized Framework for Sparse and Structured Neural Attention"
} | null | null | null | null | true | null | 2586 | null | Default | null | null |
null | {
"abstract": " Intrinsically motivated spontaneous exploration is a key enabler of\nautonomous lifelong learning in human children. It allows them to discover and\nacquire large repertoires of skills through self-generation, self-selection,\nself-ordering and self-experimentation of learning goals. We present the\nunsupervised multi-goal reinforcement learning formal framework as well as an\nalgorithmic approach called intrinsically motivated goal exploration processes\n(IMGEP) to enable similar properties of autonomous learning in machines. The\nIMGEP algorithmic architecture relies on several principles: 1) self-generation\nof goals as parameterized reinforcement learning problems; 2) selection of\ngoals based on intrinsic rewards; 3) exploration with parameterized\ntime-bounded policies and fast incremental goal-parameterized policy search; 4)\nsystematic reuse of information acquired when targeting a goal for improving\nother goals. We present a particularly efficient form of IMGEP that uses a\nmodular representation of goal spaces as well as intrinsic rewards based on\nlearning progress. We show how IMGEPs automatically generate a learning\ncurriculum within an experimental setup where a real humanoid robot can explore\nmultiple spaces of goals with several hundred continuous dimensions. While no\nparticular target goal is provided to the system beforehand, this curriculum\nallows the discovery of skills of increasing complexity, that act as stepping\nstone for learning more complex skills (like nested tool use). We show that\nlearning several spaces of diverse problems can be more efficient for learning\ncomplex skills than only trying to directly learn these complex skills. We\nillustrate the computational efficiency of IMGEPs as these robotic experiments\nuse a simple memory-based low-level policy representations and search\nalgorithm, enabling the whole system to learn online and incrementally on a\nRaspberry Pi 3.\n",
"title": "Intrinsically Motivated Goal Exploration Processes with Automatic Curriculum Learning"
} | null | null | null | null | true | null | 2587 | null | Default | null | null |
null | {
"abstract": " Previous research has traditionally analyzed emoji sentiment from the point\nof view of the reader of the content not the author. Here, we analyze emoji\nsentiment from the point of view of the author and present a emoji sentiment\nbenchmark that was built from an employee happiness dataset where emoji happen\nto be annotated with daily happiness of the author of the comment. The data\nspans over 3 years, and 4k employees of 56 companies based in Barcelona. We\ncompare sentiment of writers to readers. Results indicate that, there is an 82%\nagreement in how emoji sentiment is perceived by readers and writers. Finally,\nwe report that when authors use emoji they report higher levels of happiness.\nEmoji use was not found to be correlated with differences in author moodiness.\n",
"title": "Sentiment Perception of Readers and Writers in Emoji use"
} | null | null | null | null | true | null | 2588 | null | Default | null | null |
null | {
"abstract": " The concept of a C-class of differential equations goes back to E. Cartan\nwith the upshot that generic equations in a C-class can be solved without\nintegration. While Cartan's definition was in terms of differential invariants\nbeing first integrals, all results exhibiting C-classes that we are aware of\nare based on the fact that a canonical Cartan geometry associated to the\nequations in the class descends to the space of solutions. For sufficiently low\norders, these geometries belong to the class of parabolic geometries and the\nresults follow from the general characterization of geometries descending to a\ntwistor space.\nIn this article we answer the question of whether a canonical Cartan geometry\ndescends to the space of solutions in the case of scalar ODEs of order at least\nfour and of systems of ODEs of order at least three. As in the lower order\ncases, this is characterized by the vanishing of the generalized Wilczynski\ninvariants, which are defined via the linearization at a solution. The\ncanonical Cartan geometries (which are not parabolic geometries) are a slight\nvariation of those available in the literature based on a recent general\nconstruction. All the verifications needed to apply this construction for the\nclasses of ODEs we study are carried out in the article, which thus also\nprovides a complete alternative proof for the existence of canonical Cartan\nconnections associated to higher order (systems of) ODEs.\n",
"title": "On C-class equations"
} | null | null | null | null | true | null | 2589 | null | Default | null | null |
null | {
"abstract": " Interval estimation of quantiles has been treated by many in the literature.\nHowever, to the best of our knowledge there has been no consideration for\ninterval estimation when the data are available in grouped format. Motivated by\nthis, we introduce several methods to obtain confidence intervals for quantiles\nwhen only grouped data is available. Our preferred method for interval\nestimation is to approximate the underlying density using the Generalized\nLambda Distribution (GLD) to both estimate the quantiles and variance of the\nquantile estimators. We compare the GLD method with some other methods that we\nalso introduce which are based on a frequency approximation approach and a\nlinear interpolation approximation of the density. Our methods are strongly\nsupported by simulations showing that excellent coverage can be achieved for a\nwide number of distributions. These distributions include highly-skewed\ndistributions such as the log-normal, Dagum and Singh-Maddala distributions. We\nalso apply our methods to real data and show that inference can be carried out\non published outcomes that have been summarized only by a histogram. Our\nmethods are therefore useful for a broad range of applications. We have also\ncreated a web application that can be used to conveniently calculate the\nestimators.\n",
"title": "Confidence Intervals for Quantiles from Histograms and Other Grouped Data"
} | null | null | null | null | true | null | 2590 | null | Default | null | null |
null | {
"abstract": " In recent years, a large body of research has demonstrated that judgments and\nbehaviors can propagate from person to person. Phenomena as diverse as\npolitical mobilization, health practices, altruism, and emotional states\nexhibit similar dynamics of social contagion. The precise mechanisms of\njudgment propagation are not well understood, however, because it is difficult\nto control for confounding factors such as homophily or dynamic network\nstructures. We introduce a novel experimental design that renders possible the\nstringent study of judgment propagation. In this design, experimental chains of\nindividuals can revise their initial judgment in a visual perception task after\nobserving a predecessor's judgment. The positioning of a very good performer at\nthe top of a chain created a performance gap, which triggered waves of judgment\npropagation down the chain. We evaluated the dynamics of judgment propagation\nexperimentally. Despite strong social influence within pairs of individuals,\nthe reach of judgment propagation across a chain rarely exceeded a social\ndistance of three to four degrees of separation. Furthermore, computer\nsimulations showed that the speed of judgment propagation decayed exponentially\nwith the social distance from the source. We show that information distortion\nand the overweighting of other people's errors are two individual-level\nmechanisms hindering judgment propagation at the scale of the chain. Our\nresults contribute to the understanding of social contagion processes, and our\nexperimental method offers numerous new opportunities to study judgment\npropagation in the laboratory.\n",
"title": "Reach and speed of judgment propagation in the laboratory"
} | null | null | null | null | true | null | 2591 | null | Default | null | null |
null | {
"abstract": " Texture characterization is a key problem in image understanding and pattern\nrecognition. In this paper, we present a flexible shape-based texture\nrepresentation using shape co-occurrence patterns. More precisely, texture\nimages are first represented by tree of shapes, each of which is associated\nwith several geometrical and radiometric attributes. Then four typical kinds of\nshape co-occurrence patterns based on the hierarchical relationship of the\nshapes in the tree are learned as codewords. Three different coding methods are\ninvestigated to learn the codewords, with which, any given texture image can be\nencoded into a descriptive vector. In contrast with existing works, the\nproposed method not only inherits the strong ability to depict geometrical\naspects of textures and the high robustness to variations of imaging conditions\nfrom the shape-based method, but also provides a flexible way to consider shape\nrelationships and to compute high-order statistics on the tree. To our\nknowledge, this is the first time to use co-occurrence patterns of explicit\nshapes as a tool for texture analysis. Experiments on various texture datasets\nand scene datasets demonstrate the efficiency of the proposed method.\n",
"title": "Texture Characterization by Using Shape Co-occurrence Patterns"
} | null | null | null | null | true | null | 2592 | null | Default | null | null |
null | {
"abstract": " Images and spectra of the open cluster NGC 3105 have been obtained with GMOS\non Gemini South. The (i', g'-i') color-magnitude diagram (CMD) constructed from\nthese data extends from the brightest cluster members to g'~23. This is 4 - 5\nmag fainter than previous CMDs at visible wavelengths and samples cluster\nmembers with sub-solar masses. Assuming a half-solar metallicity, comparisons\nwith isochrones yield a distance of 6.6+/-0.3 kpc. An age of at least 32 Myr is\nfound based on the photometric properties of the brightest stars, coupled with\nthe apparent absence of pre-main sequence stars in the lower regions of the\nCMD. The luminosity function of stars between 50 and 70 arcsec from the cluster\ncenter is consistent with a Chabrier lognormal mass function. However, at radii\nsmaller than 50 arcsec there is a higher specific frequency of the most massive\nmain sequence stars than at larger radii. Photometry obtained from archival\nSPITZER images reveals that some of the brightest stars near NGC 3105 have\nexcess infrared emission, presumably from warm dust envelopes. Halpha emission\nis detected in a few early-type stars in and around the cluster, building upon\nprevious spectroscopic observations that found Be stars near NGC 3105. The\nequivalent width of the NaD lines in the spectra of early type stars is\nconsistent with the reddening found from comparisons with isochrones. Stars\nwith i'~18.5 that fall near the cluster main sequence have a spectral-type A5V,\nand a distance modulus that is consistent with that obtained by comparing\nisochrones with the CMD is found assuming solar neighborhood intrinsic\nbrightnesses for these stars.\n",
"title": "NGC 3105: A Young Cluster in the Outer Galaxy"
} | null | null | null | null | true | null | 2593 | null | Default | null | null |
null | {
"abstract": " We present an exact ground state solution of a quantum dimer model introduced\nin Ref.[1], which features ordinary bosonic spin-singlet dimers as well as\nfermionic dimers that can be viewed as bound states of spinons and holons in a\nhole-doped resonating valence bond liquid. Interestingly, this model captures\nseveral essential properties of the metallic pseudogap phase in high-$T_c$\ncuprate superconductors. We identify a line in parameter space where the exact\nground state wave functions can be constructed at an arbitrary density of\nfermionic dimers. At this exactly solvable line the ground state has a huge\ndegeneracy, which can be interpreted as a flat band of fermionic excitations.\nPerturbing around the exactly solvable line, this degeneracy is lifted and the\nground state is a fractionalized Fermi liquid with a small pocket Fermi surface\nin the low doping limit.\n",
"title": "Exact solution of a two-species quantum dimer model for pseudogap metals"
} | null | null | null | null | true | null | 2594 | null | Default | null | null |
null | {
"abstract": " We study the instability of standing wave solutions for nonlinear\nSchrödinger equations with a one-dimensional harmonic potential in\ndimension $N\\ge 2$. We prove that if the nonlinearity is $L^2$-critical or\nsupercritical in dimension $N-1$, then any ground states are strongly unstable\nby blowup.\n",
"title": "Strong instability of standing waves for nonlinear Schrödinger equations with a partial confinement"
} | null | null | [
"Mathematics"
]
| null | true | null | 2595 | null | Validated | null | null |
null | {
"abstract": " This paper proposes the matrix-weighted consensus algorithm, which is a\ngeneralization of the consensus algorithm in the literature. Given a networked\ndynamical system where the interconnections between agents are weighted by\nnonnegative definite matrices instead of nonnegative scalars, consensus and\nclustering phenomena naturally exist. We examine algebraic and algebraic graph\nconditions for achieving a consensus, and provide an algorithm for finding all\nclusters of a given system. Finally, we illustrate two applications of the\nproposed consensus algorithm in clustered consensus and in bearing-based\nformation control.\n",
"title": "Theory and Applications of Matrix-Weighted Consensus"
} | null | null | null | null | true | null | 2596 | null | Default | null | null |
null | {
"abstract": " We study a strategic version of the multi-armed bandit problem, where each\narm is an individual strategic agent and we, the principal, pull one arm each\nround. When pulled, the arm receives some private reward $v_a$ and can choose\nan amount $x_a$ to pass on to the principal (keeping $v_a-x_a$ for itself). All\nnon-pulled arms get reward $0$. Each strategic arm tries to maximize its own\nutility over the course of $T$ rounds. Our goal is to design an algorithm for\nthe principal incentivizing these arms to pass on as much of their private\nrewards as possible.\nWhen private rewards are stochastically drawn each round ($v_a^t \\leftarrow\nD_a$), we show that:\n- Algorithms that perform well in the classic adversarial multi-armed bandit\nsetting necessarily perform poorly: For all algorithms that guarantee low\nregret in an adversarial setting, there exist distributions $D_1,\\ldots,D_k$\nand an approximate Nash equilibrium for the arms where the principal receives\nreward $o(T)$.\n- Still, there exists an algorithm for the principal that induces a game\namong the arms where each arm has a dominant strategy. When each arm plays its\ndominant strategy, the principal sees expected reward $\\mu'T - o(T)$, where\n$\\mu'$ is the second-largest of the means $\\mathbb{E}[D_{a}]$. This algorithm\nmaintains its guarantee if the arms are non-strategic ($x_a = v_a$), and also\nif there is a mix of strategic and non-strategic arms.\n",
"title": "Multi-armed Bandit Problems with Strategic Arms"
} | null | null | null | null | true | null | 2597 | null | Default | null | null |
null | {
"abstract": " With the recent focus in the accessibility field, researchers from academia\nand industry have been very active in developing innovative techniques and\ntools for assistive technology. Especially with handheld devices getting ever\npowerful and being able to recognize the user's voice, screen magnification for\nindividuals with low-vision, and eye tracking devices used in studies with\nindividuals with physical and intellectual disabilities, the science field is\nquickly adapting and creating conclusions as well as products to help. In this\npaper, we will focus on new technology and tools to help make reading\neasier--including reformatting document presentation (for people with physical\nvision impairments) and text simplification to make information itself easier\nto interpret (for people with intellectual disabilities). A real-world case\nstudy is reported based on our experience to make documents more accessible.\n",
"title": "Emerging Topics in Assistive Reading Technology: From Presentation to Content Accessibility"
} | null | null | null | null | true | null | 2598 | null | Default | null | null |
null | {
"abstract": " Machine learning and quantum computing are two technologies each with the\npotential for altering how computation is performed to address previously\nuntenable problems. Kernel methods for machine learning are ubiquitous for\npattern recognition, with support vector machines (SVMs) being the most\nwell-known method for classification problems. However, there are limitations\nto the successful solution to such problems when the feature space becomes\nlarge, and the kernel functions become computationally expensive to estimate. A\ncore element to computational speed-ups afforded by quantum algorithms is the\nexploitation of an exponentially large quantum state space through controllable\nentanglement and interference. Here, we propose and experimentally implement\ntwo novel methods on a superconducting processor. Both methods represent the\nfeature space of a classification problem by a quantum state, taking advantage\nof the large dimensionality of quantum Hilbert space to obtain an enhanced\nsolution. One method, the quantum variational classifier builds on [1,2] and\noperates through using a variational quantum circuit to classify a training set\nin direct analogy to conventional SVMs. In the second, a quantum kernel\nestimator, we estimate the kernel function and optimize the classifier\ndirectly. The two methods present a new class of tools for exploring the\napplications of noisy intermediate scale quantum computers [3] to machine\nlearning.\n",
"title": "Supervised learning with quantum enhanced feature spaces"
} | null | null | null | null | true | null | 2599 | null | Default | null | null |
null | {
"abstract": " Using password based authentication technique, a system maintains the login\ncredentials (username, password) of the users in a password file. Once the\npassword file is compromised, an adversary obtains both the login credentials.\nWith the advancement of technology, even if a password is maintained in hashed\nformat, then also the adversary can invert the hashed password to get the\noriginal one. To mitigate this threat, most of the systems nowadays store some\nsystem generated fake passwords (also known as honeywords) along with the\noriginal password of a user. This type of setup confuses an adversary while\nselecting the original password. If the adversary chooses any of these\nhoneywords and submits that as a login credential, then system detects the\nattack. A large number of significant work have been done on designing\nmethodologies (identified as $\\text{M}^{\\text{DS}}_{\\text{OA}}$) that can\nprotect password against observation or, shoulder surfing attack. Under this\nattack scenario, an adversary observes (or records) the login information\nentered by a user and later uses those credentials to impersonate the genuine\nuser. In this paper, we have shown that because of their design principle, a\nlarge subset of $\\text{M}^{\\text{DS}}_{\\text{OA}}$ (identified as\n$\\text{M}^{\\text{FODS}}_{\\text{SOA}}$) cannot afford to store honeywords in\npassword file. Thus these methods, belonging to\n$\\text{M}^{\\text{FODS}}_{\\text{SOA}}$, are unable to provide any kind of\nsecurity once password file gets compromised. Through our contribution in this\npaper, by still using the concept of honeywords, we have proposed few generic\nprinciples to mask the original password of\n$\\text{M}^{\\text{FODS}}_{\\text{SOA}}$ category methods. We also consider few\nwell-established methods like S3PAS, CHC, PAS and COP belonging to\n$\\text{M}^{\\text{FODS}}_{\\text{SOA}}$, to show that proposed idea is\nimplementable in practice.\n",
"title": "On The Limitation of Some Fully Observable Multiple Session Resilient Shoulder Surfing Defense Mechanisms"
} | null | null | null | null | true | null | 2600 | null | Default | null | null |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.