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": " Transition metal dichalcogenides (TMDs) exhibit a remarkable exciton physics\nincluding optically accessible (bright) as well as spin- and momentum-forbidden\n(dark) excitonic states. So far the dark exciton landscape has not been\nrevealed leaving in particular the spectral position of momentum-forbidden dark\nstates completely unclear. This has a significant impact on the technological\napplication potential of TMDs, since the nature of the energetically lowest\nstate determines, if the material is a direct-gap semiconductor. Here, we show\nhow dark states can be experimentally revealed by probing the intra-excitonic\n1s-2p transition. Distinguishing the optical response shortly after the\nexcitation (< 100$\\,$fs) and after the exciton thermalization (> 1$\\,$ps)\nallows us to demonstrate the relative position of bright and dark excitons. We\nfind both in theory and experiment a clear blue-shift in the optical response\ndemonstrating for the first time the transition of bright exciton populations\ninto lower lying momentum- and spin-forbidden dark excitonic states in\nmonolayer WSe$_2$.\n",
"title": "Mapping of the dark exciton landscape in transition metal dichalcogenides"
}
| null | null | null | null | true | null |
15201
| null |
Default
| null | null |
null |
{
"abstract": " We review and modify the active set algorithm by Duembgen et al. (2011) for\nnonparametric maximum-likelihood estimation of a log-concave density. This\nparticular estimation problem is embedded into a more general framework\nincluding also the estimation of a log-convex tail inflation function as\nproposed by McCullagh and Polson (2012).\n",
"title": "Active set algorithms for estimating shape-constrained density ratios"
}
| null | null | null | null | true | null |
15202
| null |
Default
| null | null |
null |
{
"abstract": " Let $M$ be a II$_1$ factor with a von Neumann subalgebra $Q\\subset M$ that\nhas infinite index under any projection in $Q'\\cap M$ (e.g., $Q$ abelian; or\n$Q$ an irreducible subfactor with infinite Jones index). We prove that given\nany separable subalgebra $B$ of the ultrapower II$_1$ factor $M^\\omega$, for a\nnon-principal ultrafilter $\\omega$ on $\\Bbb N$, there exists a unitary element\n$u\\in M^\\omega$ such that $uBu^*$ is orthogonal to $Q^\\omega$.\n",
"title": "Asymptotic orthogonalization of subalgebras in II$_1$ factors"
}
| null | null | null | null | true | null |
15203
| null |
Default
| null | null |
null |
{
"abstract": " In statistics and machine learning, approximation of an intractable\nintegration is often achieved by using the unbiased Monte Carlo estimator, but\nthe variances of the estimation are generally high in many applications.\nControl variates approaches are well-known to reduce the variance of the\nestimation. These control variates are typically constructed by employing\npredefined parametric functions or polynomials, determined by using those\nsamples drawn from the relevant distributions. Instead, we propose to construct\nthose control variates by learning neural networks to handle the cases when\ntest functions are complex. In many applications, obtaining a large number of\nsamples for Monte Carlo estimation is expensive, which may result in\noverfitting when training a neural network. We thus further propose to employ\nauxiliary random variables induced by the original ones to extend data samples\nfor training the neural networks. We apply the proposed control variates with\naugmented variables to thermodynamic integration and reinforcement learning.\nExperimental results demonstrate that our method can achieve significant\nvariance reduction compared with other alternatives.\n",
"title": "Neural Control Variates for Variance Reduction"
}
| null | null |
[
"Statistics"
] | null | true | null |
15204
| null |
Validated
| null | null |
null |
{
"abstract": " In this work, which is based on an essential linear analysis carried out by\nChristodoulou, we study the evolution of tidal energy for the motion of two\ngravitating incompressible fluid balls with free boundaries obeying the\nEuler-Poisson equations. The orbital energy is defined as the mechanical energy\nof the two bodies' center of mass. According to the classical analysis of\nKepler and Newton, when the fluids are replaced by point masses, the conic\ncurve describing the trajectories of the masses is a hyperbola when the orbital\nenergy is positive and an ellipse when the orbital energy is negative. The\norbital energy is conserved in the case of point masses. If the point masses\nare initially very far, then the orbital energy is positive, corresponding to\nhyperbolic motion. However, in the motion of fluid bodies the orbital energy is\nno longer conserved because part of the conserved energy is used in deforming\nthe boundaries of the bodies. In this case the total energy\n$\\tilde{\\mathcal{E}}$ can be decomposed into a sum\n$\\tilde{\\mathcal{E}}:=\\widetilde{\\mathcal{E}_{\\mathrm{orbital}}}+\\widetilde{\\mathcal{E}_{\\mathrm{tidal}}}$,\nwith $\\widetilde{\\mathcal{E}_{\\mathrm{tidal}}}$ measuring the energy used in\ndeforming the boundaries, such that if\n$\\widetilde{\\mathcal{E}_{\\mathrm{orbital}}}<-c<0$ for some absolute constant\n$c>0$, then the orbit of the bodies must be bounded. In this work we prove that\nunder appropriate conditions on the initial configuration of the system, the\nfluid boundaries and velocity remain regular up to the point of the first\nclosest approach in the orbit, and that the tidal energy\n$\\widetilde{\\mathcal{E}_{\\mathrm{tidal}}}$ can be made arbitrarily large\nrelative to the total energy $\\tilde{\\mathcal{E}}$. In particular under these\nconditions $\\widetilde{\\mathcal{E}_{\\mathrm{orbital}}}$, which is initially\npositive, becomes negative before the point of the first closest approach.\n",
"title": "On tidal energy in Newtonian two-body motion"
}
| null | null |
[
"Physics",
"Mathematics"
] | null | true | null |
15205
| null |
Validated
| null | null |
null |
{
"abstract": " Classical linear regression is considered for a case when regression\nparameters depend on the external random environment. The last is described as\na continuous time Markov chain with finite state space. Here the expected\nsojourn times in various states are additional regressors. Necessary formulas\nfor an estimation of regression parameters have been derived. The numerical\nexample illustrates the results obtained.\n",
"title": "Markov-Modulated Linear Regression"
}
| null | null | null | null | true | null |
15206
| null |
Default
| null | null |
null |
{
"abstract": " In this paper we consider the Witten Laplacian on 0-forms and give sufficient\nconditions under which the Witten Laplacian admits a compact resolvent. These\nconditions are imposed on the potential itself, involving the control of high\norder derivatives by lower ones, as well as the control of the positive\neigenvalues of the Hessian matrix. This compactness criterion for resolvent is\ninspired by the one for the Fokker-Planck operator. Our method relies on the\nnilpotent group techniques developed by Helffer-Nourrigat [Hypoellipticité\nmaximale pour des opérateurs polynômes de champs de vecteurs, 1985].\n",
"title": "Compactness of the resolvent for the Witten Laplacian"
}
| null | null | null | null | true | null |
15207
| null |
Default
| null | null |
null |
{
"abstract": " For $p > 1$ let a function $\\varphi_p(x) = x^2/2$ if $|x|\\le 1$ and\n$\\varphi_p(x) = 1/p|x|^p -1/p + 1/2$ if $|x| > 1$. For a random variable $\\xi$\nlet $\\tau_{\\varphi_p}(\\xi)$ denote $\\inf\\{c\\ge 0 :\\;\n\\forall_{\\lambda\\in\\mathbb{R}}\\;\n\\ln\\mathbb{E}\\exp(\\lambda\\xi)\\le\\varphi_p(c\\lambda)\\}$; $\\tau_{\\varphi_p}$ is a\nnorm in a space $Sub_{\\varphi_p}(\\Omega) =\\{\\xi:\n\\; \\tau_{\\varphi_p}(\\xi) <\\infty\\}$ of $\\varphi_p$-subgaussian random\nvariables which we call {\\it subgaussian of rank $p$ random variables}. For $p\n= 2$ we have the classic subgaussian random variables. The Azuma inequality\ngives an estimate on the probability of the deviations of a zero-mean\nmartingale $(\\xi_n)_{n\\ge 0}$ with bounded increments from zero. In its classic\nform is assumed that $\\xi_0 = 0$. In this paper it is shown a version of the\nAzuma inequality under assumption that $\\xi_0$ is any subgaussian of rank $p$\nrandom variable.\n",
"title": "On the Azuma inequality in spaces of subgaussian of rank $p$ random variables"
}
| null | null | null | null | true | null |
15208
| null |
Default
| null | null |
null |
{
"abstract": " Here we report small-angle neutron scattering (SANS) measurements and\ntheoretical modeling of U$_3$Al$_2$Ge$_3$. Analysis of the SANS data reveals a\nphase transition to sinusoidally modulated magnetic order, at\n$T_{\\mathrm{N}}=63$~K to be second order, and a first order phase transition to\nferromagnetic order at $T_{\\mathrm{c}}=48$~K. Within the sinusoidally modulated\nmagnetic phase ($T_{\\mathrm{c}} < T < T_{\\mathrm{N}}$), we uncover a dramatic\nchange, by a factor of three, in the ordering wave-vector as a function of\ntemperature. These observations all indicate that U$_3$Al$_2$Ge$_3$ is a close\nrealization of the three-dimensional Axial Next-Nearest-Neighbor Ising model, a\nprototypical framework for describing commensurate to incommensurate phase\ntransitions in frustrated magnets.\n",
"title": "Realization of the Axial Next-Nearest-Neighbor Ising model in U$_3$Al$_2$Ge$_3$"
}
| null | null | null | null | true | null |
15209
| null |
Default
| null | null |
null |
{
"abstract": " We derive asymptotic formulas for the solution of the derivative nonlinear\nSchrödinger equation on the half-line under the assumption that the initial\nand boundary values lie in the Schwartz class. The formulas clearly show the\neffect of the boundary on the solution. The approach is based on a nonlinear\nsteepest descent analysis of an associated Riemann-Hilbert problem.\n",
"title": "Long-time asymptotics for the derivative nonlinear Schrödinger equation on the half-line"
}
| null | null | null | null | true | null |
15210
| null |
Default
| null | null |
null |
{
"abstract": " We develop the general theory for the construction of Extended Topological\nQuantum Field Theories (ETQFTs) associated with the Costantino-Geer-Patureau\nquantum invariants of closed 3-manifolds. In order to do so, we introduce\nrelative modular categories, a class of ribbon categories which are modeled on\nrepresentations of unrolled quantum groups, and which can be thought of as a\nnon-semisimple analogue to modular categories. Our approach exploits a\n2-categorical version of the universal construction introduced by Blanchet,\nHabegger, Masbaum, and Vogel. The 1+1+1-EQFTs thus obtained are realized by\nsymmetric monoidal 2-functors which are defined over non-rigid 2-categories of\nadmissible cobordisms decorated with colored ribbon graphs and cohomology\nclasses, and which take values in 2-categories of complete graded linear\ncategories. In particular, our construction extends the family of graded\n2+1-TQFTs defined for the unrolled version of quantum $\\mathfrak{sl}_2$ by\nBlanchet, Costantino, Geer, and Patureau to a new family of graded ETQFTs. The\nnon-semisimplicity of the theory is witnessed by the presence of non-semisimple\ngraded linear categories associated with critical 1-manifolds.\n",
"title": "Non-Semisimple Extended Topological Quantum Field Theories"
}
| null | null |
[
"Mathematics"
] | null | true | null |
15211
| null |
Validated
| null | null |
null |
{
"abstract": " Social network analysis provides meaningful information about behavior of\nnetwork members that can be used for diverse applications such as\nclassification, link prediction. However, network analysis is computationally\nexpensive because of feature learning for different applications. In recent\nyears, many researches have focused on feature learning methods in social\nnetworks. Network embedding represents the network in a lower dimensional\nrepresentation space with the same properties which presents a compressed\nrepresentation of the network. In this paper, we introduce a novel algorithm\nnamed \"CARE\" for network embedding that can be used for different types of\nnetworks including weighted, directed and complex. Current methods try to\npreserve local neighborhood information of nodes, whereas the proposed method\nutilizes local neighborhood and community information of network nodes to cover\nboth local and global structure of social networks. CARE builds customized\npaths, which are consisted of local and global structure of network nodes, as a\nbasis for network embedding and uses the Skip-gram model to learn\nrepresentation vector of nodes. Subsequently, stochastic gradient descent is\napplied to optimize our objective function and learn the final representation\nof nodes. Our method can be scalable when new nodes are appended to network\nwithout information loss. Parallelize generation of customized random walks is\nalso used for speeding up CARE. We evaluate the performance of CARE on multi\nlabel classification and link prediction tasks. Experimental results on various\nnetworks indicate that the proposed method outperforms others in both Micro and\nMacro-f1 measures for different size of training data.\n",
"title": "Community Aware Random Walk for Network Embedding"
}
| null | null | null | null | true | null |
15212
| null |
Default
| null | null |
null |
{
"abstract": " Understanding the nature of bulges in disc galaxies can provide important\ninsights into the formation and evolution of galaxies. For instance, the\npresence of a classical bulge suggests a relatively violent history, in\ncontrast, the presence of simply an inner disc (also referred to as a\n\"pseudobulge\") indicates the occurrence of secular evolution processes in the\nmain disc. However, we still lack criteria to effectively categorise bulges,\nlimiting our ability to study their impact on the evolution of the host\ngalaxies. Here we present a recipe to separate inner discs from classical\nbulges by combining four different parameters from photometric and kinematic\nanalyses: The bulge Sérsic index $n_\\mathrm{b}$, the concentration index\n$C_{20,50}$, the Kormendy (1977) relation and the inner slope of the radial\nvelocity dispersion profile $\\nabla\\sigma$. With that recipe we provide a\ndetailed bulge classification for a sample of 45 galaxies from the\nintegral-field spectroscopic survey CALIFA. To aid in categorising bulges\nwithin these galaxies, we perform 2D image decomposition to determine bulge\nSérsic index, bulge-to-total light ratio, surface brightness and effective\nradius of the bulge and use growth curve analysis to derive a new concentration\nindex, $C_{20,50}$. We further extract the stellar kinematics from CALIFA data\ncubes and analyse the radial velocity dispersion profile. The results of the\ndifferent approaches are in good agreement and allow a safe classification for\napproximately $95\\%$ of the galaxies. In particular, we show that our new\n\"inner\" concentration index performs considerably better than the traditionally\nused $C_{50,90}$ when yielding the nature of bulges. We also found that a\ncombined use of this index and the Kormendy (1977) relation gives a very robust\nindication of the physical nature of the bulge.\n",
"title": "A combined photometric and kinematic recipe for evaluating the nature of bulges using the CALIFA sample"
}
| null | null | null | null | true | null |
15213
| null |
Default
| null | null |
null |
{
"abstract": " Growth in both size and complexity of modern data challenges the\napplicability of traditional likelihood-based inference. Composite likelihood\n(CL) methods address the difficulties related to model selection and\ncomputational intractability of the full likelihood by combining a number of\nlow-dimensional likelihood objects into a single objective function used for\ninference. This paper introduces a procedure to combine partial likelihood\nobjects from a large set of feasible candidates and simultaneously carry out\nparameter estimation. The new method constructs estimating equations balancing\nstatistical efficiency and computing cost by minimizing an approximate distance\nfrom the full likelihood score subject to a L1-norm penalty representing the\navailable computing resources. This results in truncated CL equations\ncontaining only the most informative partial likelihood score terms. An\nasymptotic theory within a framework where both sample size and data dimension\ngrow is developed and finite-sample properties are illustrated through\nnumerical examples.\n",
"title": "Fast construction of efficient composite likelihood equations"
}
| null | null | null | null | true | null |
15214
| null |
Default
| null | null |
null |
{
"abstract": " We report a measurement of $KLL$ dielectronic recombination in charge states\nfrom Kr$^{+34}$ through Kr$^{+28}$, in order to investigate the contribution of\nBreit interaction for a wide range of resonant states. Highly charged Kr ions\nwere produced in an electron beam ion trap, while the electron-ion collision\nenergy was scanned over a range of dielectronic recombination resonances. The\nsubsequent $K\\alpha$ x rays were recorded both along and perpendicular to the\nelectron beam axis, which allowed the observation of the influence of Breit\ninteraction on the angular distribution of the x rays. Experimental results are\nin good agreement with distorted-wave calculations. We demonstrate, both\ntheoretically and experimentally, that there is a strong state-selective\ninfluence of the Breit interaction that can be traced back to the angular and\nradial properties of the wavefunctions in the dielectronic capture.\n",
"title": "State-selective influence of the Breit interaction on the angular distribution of emitted photons following dielectronic recombination"
}
| null | null | null | null | true | null |
15215
| null |
Default
| null | null |
null |
{
"abstract": " Growing interest in automatic speaker verification (ASV)systems has lead to\nsignificant quality improvement of spoofing attackson them. Many research works\nconfirm that despite the low equal er-ror rate (EER) ASV systems are still\nvulnerable to spoofing attacks. Inthis work we overview different acoustic\nfeature spaces and classifiersto determine reliable and robust countermeasures\nagainst spoofing at-tacks. We compared several spoofing detection systems,\npresented so far,on the development and evaluation datasets of the Automatic\nSpeakerVerification Spoofing and Countermeasures (ASVspoof) Challenge\n2015.Experimental results presented in this paper demonstrate that the useof\nmagnitude and phase information combination provides a substantialinput into\nthe efficiency of the spoofing detection systems. Also wavelet-based features\nshow impressive results in terms of equal error rate. Inour overview we compare\nspoofing performance for systems based on dif-ferent classifiers. Comparison\nresults demonstrate that the linear SVMclassifier outperforms the conventional\nGMM approach. However, manyresearchers inspired by the great success of deep\nneural networks (DNN)approaches in the automatic speech recognition, applied\nDNN in thespoofing detection task and obtained quite low EER for known and\nun-known type of spoofing attacks.\n",
"title": "Anti-spoofing Methods for Automatic SpeakerVerification System"
}
| null | null | null | null | true | null |
15216
| null |
Default
| null | null |
null |
{
"abstract": " Face deidentification is an active topic amongst privacy and security\nresearchers. Early deidentification methods relying on image blurring or\npixelization were replaced in recent years with techniques based on formal\nanonymity models that provide privacy guaranties and at the same time aim at\nretaining certain characteristics of the data even after deidentification. The\nlatter aspect is particularly important, as it allows to exploit the\ndeidentified data in applications for which identity information is irrelevant.\nIn this work we present a novel face deidentification pipeline, which ensures\nanonymity by synthesizing artificial surrogate faces using generative neural\nnetworks (GNNs). The generated faces are used to deidentify subjects in images\nor video, while preserving non-identity-related aspects of the data and\nconsequently enabling data utilization. Since generative networks are very\nadaptive and can utilize a diverse set of parameters (pertaining to the\nappearance of the generated output in terms of facial expressions, gender,\nrace, etc.), they represent a natural choice for the problem of face\ndeidentification. To demonstrate the feasibility of our approach, we perform\nexperiments using automated recognition tools and human annotators. Our results\nshow that the recognition performance on deidentified images is close to\nchance, suggesting that the deidentification process based on GNNs is highly\neffective.\n",
"title": "Face Deidentification with Generative Deep Neural Networks"
}
| null | null | null | null | true | null |
15217
| null |
Default
| null | null |
null |
{
"abstract": " Comment on \"Dependency distance: a new perspective on syntactic patterns in\nnatural language\" by Haitao Liu et al\n",
"title": "Towards a theory of word order. Comment on \"Dependency distance: a new perspective on syntactic patterns in natural language\" by Haitao Liu et al"
}
| null | null | null | null | true | null |
15218
| null |
Default
| null | null |
null |
{
"abstract": " Precision pulsar timing requires optimization against measurement errors and\nastrophysical variance from the neutron stars themselves and the interstellar\nmedium. We investigate optimization of arrival time precision as a function of\nradio frequency and bandwidth. We find that increases in bandwidth that reduce\nthe contribution from receiver noise are countered by the strong chromatic\ndependence of interstellar effects and intrinsic pulse-profile evolution. The\nresulting optimal frequency range is therefore telescope and pulsar dependent.\nWe demonstrate the results for five pulsars included in current pulsar timing\narrays and determine that they are not optimally observed at current center\nfrequencies. For those objects, we find that better choices of total bandwidth\nas well as center frequency can improve the arrival-time precision. Wideband\nreceivers centered at somewhat higher frequencies with respect to the currently\nadopted receivers can reduce required overall integration times and provide\nsignificant improvements in arrival time uncertainty by a factor of ~sqrt(2) in\nmost cases, assuming a fixed integration time. We also discuss how timing\nprograms can be extended to pulsars with larger dispersion measures through the\nuse of higher-frequency observations.\n",
"title": "Optimal Frequency Ranges for Sub-Microsecond Precision Pulsar Timing"
}
| null | null | null | null | true | null |
15219
| null |
Default
| null | null |
null |
{
"abstract": " This article was withdrawn because (1) it was uploaded without the\nco-authors' knowledge or consent, and (2) there are allegations of plagiarism.\n",
"title": "Submodular Mini-Batch Training in Generative Moment Matching Networks"
}
| null | null | null | null | true | null |
15220
| null |
Default
| null | null |
null |
{
"abstract": " Traffic speed is a key indicator for the efficiency of an urban\ntransportation system. Accurate modeling of the spatiotemporally varying\ntraffic speed thus plays a crucial role in urban planning and development. This\npaper addresses the problem of efficient fine-grained traffic speed prediction\nusing big traffic data obtained from static sensors. Gaussian processes (GPs)\nhave been previously used to model various traffic phenomena, including flow\nand speed. However, GPs do not scale with big traffic data due to their cubic\ntime complexity. In this work, we address their efficiency issues by proposing\nlocal GPs to learn from and make predictions for correlated subsets of data.\nThe main idea is to quickly group speed variables in both spatial and temporal\ndimensions into a finite number of clusters, so that future and unobserved\ntraffic speed queries can be heuristically mapped to one of such clusters. A\nlocal GP corresponding to that cluster can then be trained on the fly to make\npredictions in real-time. We call this method localization. We use non-negative\nmatrix factorization for localization and propose simple heuristics for cluster\nmapping. We additionally leverage on the expressiveness of GP kernel functions\nto model road network topology and incorporate side information. Extensive\nexperiments using real-world traffic data collected in the two U.S. cities of\nPittsburgh and Washington, D.C., show that our proposed local GPs significantly\nimprove both runtime performances and prediction accuracies compared to the\nbaseline global and local GPs.\n",
"title": "Local Gaussian Processes for Efficient Fine-Grained Traffic Speed Prediction"
}
| null | null |
[
"Computer Science"
] | null | true | null |
15221
| null |
Validated
| null | null |
null |
{
"abstract": " We construct a family of vertex algebras associated with a family of\nsymplectic singularity/resolution, called hypertoric varieties. While the\nhypertoric varieties are constructed by a certain Hamiltonian reduction\nassociated with a torus action, our vertex algebras are constructed by\n(semi-infinite) BRST reduction. The construction works algebro-geometrically\nand we construct sheaves of $\\hbar$-adic vertex algebras over hypertoric\nvarieties which localize the vertex algebras. We show when the vertex algebras\nare vertex operator algebras by giving explicit conformal vectors. We also show\nthat the Zhu algebras of the vertex algebras, associative algebras associated\nwith non-negatively graded vertex algebras, gives a certain family of filtered\nquantizations of the coordinate rings of the hypertoric varieties.\n",
"title": "Vertex algebras associated with hypertoric varieties"
}
| null | null | null | null | true | null |
15222
| null |
Default
| null | null |
null |
{
"abstract": " Approximate model counting for bit-vector SMT formulas (generalizing \\#SAT)\nhas many applications such as probabilistic inference and quantitative\ninformation-flow security, but it is computationally difficult. Adding random\nparity constraints (XOR streamlining) and then checking satisfiability is an\neffective approximation technique, but it requires a prior hypothesis about the\nmodel count to produce useful results. We propose an approach inspired by\nstatistical estimation to continually refine a probabilistic estimate of the\nmodel count for a formula, so that each XOR-streamlined query yields as much\ninformation as possible. We implement this approach, with an approximate\nprobability model, as a wrapper around an off-the-shelf SMT solver or SAT\nsolver. Experimental results show that the implementation is faster than the\nmost similar previous approaches which used simpler refinement strategies. The\ntechnique also lets us model count formulas over floating-point constraints,\nwhich we demonstrate with an application to a vulnerability in differential\nprivacy mechanisms.\n",
"title": "Bit-Vector Model Counting using Statistical Estimation"
}
| null | null | null | null | true | null |
15223
| null |
Default
| null | null |
null |
{
"abstract": " We develop a family of reformulations of an arbitrary consistent linear\nsystem into a stochastic problem. The reformulations are governed by two\nuser-defined parameters: a positive definite matrix defining a norm, and an\narbitrary discrete or continuous distribution over random matrices. Our\nreformulation has several equivalent interpretations, allowing for researchers\nfrom various communities to leverage their domain specific insights. In\nparticular, our reformulation can be equivalently seen as a stochastic\noptimization problem, stochastic linear system, stochastic fixed point problem\nand a probabilistic intersection problem. We prove sufficient, and necessary\nand sufficient conditions for the reformulation to be exact.\nFurther, we propose and analyze three stochastic algorithms for solving the\nreformulated problem---basic, parallel and accelerated methods---with global\nlinear convergence rates. The rates can be interpreted as condition numbers of\na matrix which depends on the system matrix and on the reformulation\nparameters. This gives rise to a new phenomenon which we call stochastic\npreconditioning, and which refers to the problem of finding parameters (matrix\nand distribution) leading to a sufficiently small condition number. Our basic\nmethod can be equivalently interpreted as stochastic gradient descent,\nstochastic Newton method, stochastic proximal point method, stochastic fixed\npoint method, and stochastic projection method, with fixed stepsize (relaxation\nparameter), applied to the reformulations.\n",
"title": "Stochastic Reformulations of Linear Systems: Algorithms and Convergence Theory"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
15224
| null |
Validated
| null | null |
null |
{
"abstract": " We examine dense self-gravitating stellar systems dominated by a central\npotential, such as nuclear star clusters hosting a central supermassive black\nhole. Different dynamical properties of these systems evolve on vastly\ndifferent timescales. In particular, the orbital-plane orientations are\ntypically driven into internal thermodynamic equilibrium by vector resonant\nrelaxation before the orbital eccentricities or semimajor axes relax. We show\nthat the statistical mechanics of such systems exhibit a striking resemblance\nto liquid crystals, with analogous ordered-nematic and disordered-isotropic\nphases. The ordered phase consists of bodies orbiting in a disk in both\ndirections, with the disk thickness depending on temperature, while the\ndisordered phase corresponds to a nearly isotropic distribution of the orbit\nnormals. We show that below a critical value of the total angular momentum, the\nsystem undergoes a first-order phase transition between the ordered and\ndisordered phases. At the critical point the phase transition becomes\nsecond-order while for higher angular momenta there is a smooth crossover. We\nalso find metastable equilibria containing two identical disks with mutual\ninclinations between $90^{\\circ}$ and $180^\\circ$.\n",
"title": "Isotropic-Nematic Phase Transitions in Gravitational Systems"
}
| null | null | null | null | true | null |
15225
| null |
Default
| null | null |
null |
{
"abstract": " We propose a new type of Hopf semimetals indexed by a pair of numbers\n$(p,q)$, where the Hopf number is given by $pq$. The Fermi surface is given by\nthe preimage of the Hopf map, which is nontrivially linked for a nonzero Hopf\nnumber. The Fermi surface forms a torus link, whose examples are the Hopf link\nindexed by $(1,1)$, the Solomon's knot $(2,1)$, the double Hopf-link $(2,2)$\nand the double trefoil-knot $(3,2)$. We may choose $p$ or $q$ as a half\ninteger, where torus-knot Fermi surfaces such as the trefoil knot $(3/2,1)$ are\nrealized. It is even possible to make the Hopf number an arbitrary rational\nnumber, where a semimetal whose Fermi surface forms open strings is generated.\n",
"title": "Topological Semimetals carrying Arbitrary Hopf Numbers: Hopf-Link, Solomon's-Knot, Trefoil-Knot and Other Semimetals"
}
| null | null | null | null | true | null |
15226
| null |
Default
| null | null |
null |
{
"abstract": " We present an extensive study of the key problem of online learning where\nalgorithms are allowed to abstain from making predictions. In the adversarial\nsetting, we show how existing online algorithms and guarantees can be adapted\nto this problem. In the stochastic setting, we first point out a bias problem\nthat limits the straightforward extension of algorithms such as UCB-N to\ntime-varying feedback graphs, as needed in this context. Next, we give a new\nalgorithm, UCB-GT, that exploits historical data and is adapted to time-varying\nfeedback graphs. We show that this algorithm benefits from more favorable\nregret guarantees than a possible, but limited, extension of UCB-N. We further\nreport the results of a series of experiments demonstrating that UCB-GT largely\noutperforms that extension of UCB-N, as well as more standard baselines.\n",
"title": "Online Learning with Abstention"
}
| null | null | null | null | true | null |
15227
| null |
Default
| null | null |
null |
{
"abstract": " We report experimental studies of the influence of symmetric dual-loop\noptical feedback on the RF linewidth and timing jitter of self-mode-locked\ntwo-section quantum dash lasers emitting at 1550 nm. Various feedback schemes\nwere investigated and optimum levels determined for narrowest RF linewidth and\nlow timing jitter, for single-loop and symmetric dual-loop feedback. Two\nsymmetric dual-loop configurations, with balanced and unbalanced feedback\nratios, were studied. We demonstrate that unbalanced symmetric dual loop\nfeedback, with the inner cavity resonant and fine delay tuning of the outer\nloop, gives narrowest RF linewidth and reduced timing jitter over a wide range\nof delay, unlike single and balanced symmetric dual-loop configurations. This\nconfiguration with feedback lengths 80 and 140 m narrows the RF linewidth by\n4-67x and 10-100x, respectively, across the widest delay range, compared to\nfree-running. For symmetric dual-loop feedback, the influence of different\npower split ratios through the feedback loops was determined. Our results show\nthat symmetric dual-loop feedback is markedly more effective than single-loop\nfeedback in reducing RF linewidth and timing jitter, and is much less sensitive\nto delay phase, making this technique ideal for applications where robustness\nand alignment tolerance are essential.\n",
"title": "Stabilization of self-mode-locked quantum dash lasers by symmetric dual-loop optical feedback"
}
| null | null | null | null | true | null |
15228
| null |
Default
| null | null |
null |
{
"abstract": " This paper provides a set of sensitivity analysis and activity identification\nresults for a class of convex functions with a strong geometric structure, that\nwe coined \"mirror-stratifiable\". These functions are such that there is a\nbijection between a primal and a dual stratification of the space into\npartitioning sets, called strata. This pairing is crucial to track the strata\nthat are identifiable by solutions of parametrized optimization problems or by\niterates of optimization algorithms. This class of functions encompasses all\nregularizers routinely used in signal and image processing, machine learning,\nand statistics. We show that this \"mirror-stratifiable\" structure enjoys a nice\nsensitivity theory, allowing us to study stability of solutions of optimization\nproblems to small perturbations, as well as activity identification of\nfirst-order proximal splitting-type algorithms. Existing results in the\nliterature typically assume that, under a non-degeneracy condition, the active\nset associated to a minimizer is stable to small perturbations and is\nidentified in finite time by optimization schemes. In contrast, our results do\nnot require any non-degeneracy assumption: in consequence, the optimal active\nset is not necessarily stable anymore, but we are able to track precisely the\nset of identifiable strata.We show that these results have crucial implications\nwhen solving challenging ill-posed inverse problems via regularization, a\ntypical scenario where the non-degeneracy condition is not fulfilled. Our\ntheoretical results, illustrated by numerical simulations, allow to\ncharacterize the instability behaviour of the regularized solutions, by\nlocating the set of all low-dimensional strata that can be potentially\nidentified by these solutions.\n",
"title": "Sensitivity Analysis for Mirror-Stratifiable Convex Functions"
}
| null | null | null | null | true | null |
15229
| null |
Default
| null | null |
null |
{
"abstract": " Online writers and journalism media are increasingly combining visualization\n(and other multimedia content) with narrative text to create narrative\nvisualizations. Often, however, the two elements are presented independently of\none another. We propose an approach to automatically integrate text and\nvisualization elements. We begin with a writer's narrative that presumably can\nbe supported with visual data evidence. We leverage natural language\nprocessing, quantitative narrative analysis, and information visualization to\n(1) automatically extract narrative components (who, what, when, where) from\ndata-rich stories, and (2) integrate the supporting data evidence with the text\nto develop a narrative visualization. We also employ bidirectional interaction\nfrom text to visualization and visualization to text to support reader\nexploration in both directions. We demonstrate the approach with a case study\nin the data-rich field of sports journalism.\n",
"title": "Coupling Story to Visualization: Using Textual Analysis as a Bridge Between Data and Interpretation"
}
| null | null | null | null | true | null |
15230
| null |
Default
| null | null |
null |
{
"abstract": " Accurate prediction of suitable discourse connectives (however, furthermore,\netc.) is a key component of any system aimed at building coherent and fluent\ndiscourses from shorter sentences and passages. As an example, a dialog system\nmight assemble a long and informative answer by sampling passages extracted\nfrom different documents retrieved from the Web. We formulate the task of\ndiscourse connective prediction and release a dataset of 2.9M sentence pairs\nseparated by discourse connectives for this task. Then, we evaluate the\nhardness of the task for human raters, apply a recently proposed decomposable\nattention (DA) model to this task and observe that the automatic predictor has\na higher F1 than human raters (32 vs. 30). Nevertheless, under specific\nconditions the raters still outperform the DA model, suggesting that there is\nheadroom for future improvements.\n",
"title": "Automatic Prediction of Discourse Connectives"
}
| null | null | null | null | true | null |
15231
| null |
Default
| null | null |
null |
{
"abstract": " Although reinforcement learning methods can achieve impressive results in\nsimulation, the real world presents two major challenges: generating samples is\nexceedingly expensive, and unexpected perturbations or unseen situations cause\nproficient but specialized policies to fail at test time. Given that it is\nimpractical to train separate policies to accommodate all situations the agent\nmay see in the real world, this work proposes to learn how to quickly and\neffectively adapt online to new tasks. To enable sample-efficient learning, we\nconsider learning online adaptation in the context of model-based reinforcement\nlearning. Our approach uses meta-learning to train a dynamics model prior such\nthat, when combined with recent data, this prior can be rapidly adapted to the\nlocal context. Our experiments demonstrate online adaptation for continuous\ncontrol tasks on both simulated and real-world agents. We first show simulated\nagents adapting their behavior online to novel terrains, crippled body parts,\nand highly-dynamic environments. We also illustrate the importance of\nincorporating online adaptation into autonomous agents that operate in the real\nworld by applying our method to a real dynamic legged millirobot. We\ndemonstrate the agent's learned ability to quickly adapt online to a missing\nleg, adjust to novel terrains and slopes, account for miscalibration or errors\nin pose estimation, and compensate for pulling payloads.\n",
"title": "Learning to Adapt in Dynamic, Real-World Environments Through Meta-Reinforcement Learning"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
15232
| null |
Validated
| null | null |
null |
{
"abstract": " In this paper, we propose a novel splitting receiver, which involves joint\nprocessing of coherently and non-coherently received signals. Using a passive\nRF power splitter, the received signal at each receiver antenna is split into\ntwo streams which are then processed by a conventional coherent detection (CD)\ncircuit and a power-detection (PD) circuit, respectively. The streams of the\nsignals from all the receiver antennas are then jointly used for information\ndetection. We show that the splitting receiver creates a three-dimensional\nreceived signal space, due to the joint coherent and non-coherent processing.\nWe analyze the achievable rate of a splitting receiver, which shows that the\nsplitting receiver provides a rate gain of $3/2$ compared to either the\nconventional (CD-based) coherent receiver or the PD-based non-coherent receiver\nin the high SNR regime. We also analyze the symbol error rate (SER) for\npractical modulation schemes, which shows that the splitting receiver achieves\nasymptotic SER reduction by a factor of at least $\\sqrt{M}-1$ for $M$-QAM\ncompared to either the conventional (CD-based) coherent receiver or the\nPD-based non-coherent receiver.\n",
"title": "A Novel Receiver Design with Joint Coherent and Non-Coherent Processing"
}
| null | null | null | null | true | null |
15233
| null |
Default
| null | null |
null |
{
"abstract": " The paper establishes the equality condition in the I-MMSE proof of the\nentropy power inequality (EPI). This is done by establishing an exact\nexpression for the deficit between the two sides of the EPI. Interestingly, a\nnecessary condition for the equality is established by making a connection to\nthe famous Cauchy functional equation.\n",
"title": "Comment on the Equality Condition for the I-MMSE Proof of Entropy Power Inequality"
}
| null | null | null | null | true | null |
15234
| null |
Default
| null | null |
null |
{
"abstract": " The gap between our ability to collect interesting data and our ability to\nanalyze these data is growing at an unprecedented rate. Recent algorithmic\nattempts to fill this gap have employed unsupervised tools to discover\nstructure in data. Some of the most successful approaches have used\nprobabilistic models to uncover latent thematic structure in discrete data.\nDespite the success of these models on textual data, they have not generalized\nas well to image data, in part because of the spatial and temporal structure\nthat may exist in an image stream.\nWe introduce a novel unsupervised machine learning framework that\nincorporates the ability of convolutional autoencoders to discover features\nfrom images that directly encode spatial information, within a Bayesian\nnonparametric topic model that discovers meaningful latent patterns within\ndiscrete data. By using this hybrid framework, we overcome the fundamental\ndependency of traditional topic models on rigidly hand-coded data\nrepresentations, while simultaneously encoding spatial dependency in our topics\nwithout adding model complexity. We apply this model to the motivating\napplication of high-level scene understanding and mission summarization for\nexploratory marine robots. Our experiments on a seafloor dataset collected by a\nmarine robot show that the proposed hybrid framework outperforms current\nstate-of-the-art approaches on the task of unsupervised seafloor terrain\ncharacterization.\n",
"title": "Feature discovery and visualization of robot mission data using convolutional autoencoders and Bayesian nonparametric topic models"
}
| null | null | null | null | true | null |
15235
| null |
Default
| null | null |
null |
{
"abstract": " This paper is about models for a vector of probabilities whose elements must\nhave a multiplicative structure and sum to 1 at the same time; in certain\napplications, as basket analysis, these models may be seen as a constrained\nversion of quasi-independence. After reviewing the basic properties of these\nmodels, their geometric features as a curved exponential family are\ninvestigated. A new algorithm for computing maximum likelihood estimates is\npresented and new insights are provided on the underlying geometry. The\nasymptotic distribution of three statistics for hypothesis testing are derived\nand a small simulation study is presented to investigate the accuracy of\nasymptotic approximations.\n",
"title": "Multiplicative models for frequency data, estimation and testing"
}
| null | null | null | null | true | null |
15236
| null |
Default
| null | null |
null |
{
"abstract": " The family of Information Dispersal Algorithms is applied to distributed\nsystems for secure and reliable storage and transmission. In comparison with\nperfect secret sharing it achieves a significantly smaller memory overhead and\nbetter performance, but provides only incremental confidentiality. Therefore,\neven if it is not possible to explicitly reconstruct data from less than the\nrequired amount of fragments, it is still possible to deduce some information\nabout the nature of data by looking at preserved data patterns inside a\nfragment. The idea behind this paper is to provide a lightweight data\nfragmentation scheme, that would combine the space efficiency and simplicity\nthat could be find in Information Dispersal Algorithms with a computational\nlevel of data confidentiality.\n",
"title": "An Efficient Keyless Fragmentation Algorithm for Data Protection"
}
| null | null | null | null | true | null |
15237
| null |
Default
| null | null |
null |
{
"abstract": " A group theoretical formulation of Schramm--Loewner-evolution-type growth\nprocesses corresponding to Wess--Zumino--Witten theories is developed that\nmakes it possible to construct stochastic differential equations associated\nwith more general null vectors than the ones considered in the most fundamental\nexample in [Alekseev et al., Lett. Math. Phys. 97, 243-261 (2011)]. Also given\nare examples of Schramm--Loewner-evolution-type growth processes associated\nwith null vectors of conformal weight $4$ in the basic representations of\n$\\widehat{\\mathfrak{sl}}_{2}$ and $\\widehat{\\mathfrak{sl}}_{3}$.\n",
"title": "Schramm--Loewner-evolution-type growth processes corresponding to Wess--Zumino--Witten theories"
}
| null | null | null | null | true | null |
15238
| null |
Default
| null | null |
null |
{
"abstract": " Although the explicit commutativitiy conditions for second-order linear\ntime-varying systems have been appeared in some literature, these are all for\ninitially relaxed systems. This paper presents explicit necessary and\nsufficient commutativity conditions for commutativity of second-order linear\ntime-varying systems with non-zero initial conditions. It has appeared\ninteresting that the second requirement for the commutativity of non-relaxed\nsystems plays an important role on the commutativity conditions when non-zero\ninitial conditions exist. Another highlight is that the commutativity of\nswitched systems is considered and spoiling of commutativity at the switching\ninstants is illustrated for the first time. The simulation results support the\ntheory developed in the paper.\n",
"title": "Explicit Commutativity Conditions for Second-order Linear Time-Varying Systems with Non-Zero Initial Conditions"
}
| null | null | null | null | true | null |
15239
| null |
Default
| null | null |
null |
{
"abstract": " Radiative alpha-capture, ($\\alpha,\\gamma$), reactions play a critical role in\nnucleosynthesis and nuclear energy generation in a variety of astrophysical\nenvironments. The St. George recoil separator at the University of Notre Dame's\nNuclear Science Laboratory was developed to measure ($\\alpha,\\gamma$) reactions\nin inverse kinematics via recoil detection in order to obtain nuclear reaction\ncross sections at the low energies of astrophysical interest, while avoiding\nthe $\\gamma$-background that plagues traditional measurement techniques. Due to\nthe $\\gamma$-ray produced by the nuclear reaction at the target location,\nrecoil nuclei are produced with a variety of energies and angles, all of which\nmust be accepted by St. George in order to accurately determine the reaction\ncross section. We demonstrate the energy acceptance of the St. George recoil\nseparator using primary beams of helium, hydrogen, neon, and oxygen, spanning\nthe magnetic and electric rigidity phase space populated by recoils of\nanticipated ($\\alpha,\\gamma$) reaction measurements. We found the performance\nof St. George meets the design specifications, demonstrating its suitability\nfor ($\\alpha,\\gamma$) reaction measurements of astrophysical interest.\n",
"title": "Energy Acceptance of the St. George Recoil Separator"
}
| null | null | null | null | true | null |
15240
| null |
Default
| null | null |
null |
{
"abstract": " We recall first Gallai-simplicial complex $\\Delta_{\\Gamma}(G)$ associated to\nGallai graph $\\Gamma(G)$ of a planar graph $G$. The Euler characteristic is a\nvery useful topological and homotopic invariant to classify surfaces. In\nTheorems 3.2 and 3.4, we compute Euler characteristics of Gallai-simplicial\ncomplexes associated to triangular ladder and prism graphs, respectively.\nLet $G$ be a finite simple graph on $n$ vertices of the form $n=3l+2$ or\n$3l+3$. In Theorem 4.4, we prove that $G$ will be $f$-Gallai graph for the\nfollowing types of constructions of $G$.\nType 1. When $n=3l+2$. $G=\\mathbb{S}_{4l}$ is a graph consisting of two\ncopies of star graphs $S_{2l}$ and $S'_{2l}$ with $l\\geq 2$ having $l$ common\nvertices.\nType 2. When $n=3l+3$. $G=\\mathbb{S}_{4l+1}$ is a graph consisting of two\nstar graphs $S_{2l}$ and $S_{2l+1}$ with $l\\geq 2$ having $l$ common vertices.\n",
"title": "Topological and Algebraic Characterizations of Gallai-Simplicial Complexes"
}
| null | null |
[
"Mathematics"
] | null | true | null |
15241
| null |
Validated
| null | null |
null |
{
"abstract": " Ensuring that classifiers are non-discriminatory or fair with respect to a\nsensitive feature (e.g., race or gender) is a topical problem. Progress in this\ntask requires fixing a definition of fairness, and there have been several\nproposals in this regard over the past few years. Several of these, however,\nassume either binary sensitive features (thus precluding categorical or\nreal-valued sensitive groups), or result in non-convex objectives (thus\nadversely affecting the optimisation landscape). In this paper, we propose a\nnew definition of fairness that generalises some existing proposals, while\nallowing for generic sensitive features and resulting in a convex objective.\nThe key idea is to enforce that the expected losses (or risks) across each\nsubgroup induced by the sensitive feature are commensurate. We show how this\nrelates to the rich literature on risk measures from mathematical finance. As a\nspecial case, this leads to a new convex fairness-aware objective based on\nminimising the conditional value at risk (CVaR).\n",
"title": "Fairness risk measures"
}
| null | null | null | null | true | null |
15242
| null |
Default
| null | null |
null |
{
"abstract": " Currently, eXtensible Access Control Markup Language (XACML) has becoming the\nstandard for implementing access control policies and consequently more\nattention is dedicated to testing the correctness of XACML policies. In\nparticular, coverage measures can be adopted for assessing test strategy\neffectiveness in exercising the policy elements. This study introduces a set of\nXACML coverage criteria and describes the access control infrastructure, based\non a monitor engine, enabling the coverage criterion selection and the on-line\ntracing of the testing activity. Examples of infrastructure usage and of\nassessment of different test strategies are provided.\n",
"title": "On-line tracing of XACML-based policy coverage criteria"
}
| null | null | null | null | true | null |
15243
| null |
Default
| null | null |
null |
{
"abstract": " Based on the results published recently [J. Phys. A: Math. Theor. 50, 065201\n(2017)], the universal finite-size contributions to the free energy of the\nsquare lattice Ising model on the $L\\times M$ rectangle, with open boundary\nconditions in both directions, are calculated exactly in the finite-size\nscaling limit $L,M\\to\\infty$, $T\\to T_\\mathrm{c}$, with fixed temperature\nscaling variable $x\\propto(T/T_\\mathrm{c}-1)M$ and fixed aspect ratio\n$\\rho\\propto L/M$. We derive exponentially fast converging series for the\nrelated Casimir potential and Casimir force scaling functions. At the critical\npoint $T=T_\\mathrm{c}$ we confirm predictions from conformal field theory by\nCardy & Peschel [Nucl. Phys. B 300, 377 (1988)] and by Kleban & Vassileva [J.\nPhys. A: Math. Gen. 24, 3407 (1991)]. The presence of corners and the related\ncorner free energy has dramatic impact on the Casimir scaling functions and\nleads to a logarithmic divergence of the Casimir potential scaling function at\ncriticality.\n",
"title": "The square lattice Ising model on the rectangle II: Finite-size scaling limit"
}
| null | null |
[
"Physics",
"Mathematics"
] | null | true | null |
15244
| null |
Validated
| null | null |
null |
{
"abstract": " Purpose: MRI cell tracking can be used to monitor immune cells involved in\nthe immunotherapy response, providing insight into the mechanism of action,\ntemporal progression of tumour growth and individual potency of therapies. To\nevaluate whether MRI could be used to track immune cell populations in response\nto immunotherapy, CD8+ cytotoxic T cells (CTLs), CD4+CD25+FoxP3+ regulatory T\ncells (Tregs) and myeloid derived suppressor cells (MDSCs) were labelled with\nsuperparamagnetic iron oxide (SPIO) particles.\nMethods: SPIO-labelled cells were injected into mice (one cell type/mouse)\nimplanted with an HPV-based cervical cancer model. Half of these mice were also\nvaccinated with DepoVaxTM, a lipid-based vaccine platform that was developed to\nenhance the potency of peptide-based vaccines.\nResults: MRI visualization of CTLs, Tregs and MDSCs was apparent 24 hours\npost-injection, with hypointensities due to iron labelled cells clearing\napproximately 72 hours post-injection. Vaccination resulted in increased\nrecruitment of CTLs and decreased recruitment of MDSCs and Tregs to the tumour.\nWe also found that MDSC and Treg recruitment was positively correlated with\nfinal tumour volume.\nConclusion: This type of analysis can be used to non-invasively study changes\nin immune cell recruitment in individual mice over time, potentially allowing\nimproved application and combination of immunotherapies.\n",
"title": "Using MRI Cell Tracking to Monitor Immune Cell Recruitment in Response to a Peptide-Based Cancer Vaccine"
}
| null | null |
[
"Physics"
] | null | true | null |
15245
| null |
Validated
| null | null |
null |
{
"abstract": " We present a novel algorithm for learning the spectral density of large scale\nnetworks using stochastic trace estimation and the method of maximum entropy.\nThe complexity of the algorithm is linear in the number of non-zero elements of\nthe matrix, offering a computational advantage over other algorithms. We apply\nour algorithm to the problem of community detection in large networks. We show\nstate-of-the-art performance on both synthetic and real datasets.\n",
"title": "Entropic Spectral Learning in Large Scale Networks"
}
| null | null | null | null | true | null |
15246
| null |
Default
| null | null |
null |
{
"abstract": " We prove an equivalence between the infinitesimal Torelli theorem for top\nforms on a hypersurface contained inside a Grassmannian $\\mathbb G$ and the\ntheory of adjoint volume forms presented in L. Rizzi, F. Zucconi, \"Generalized\nadjoint forms on algebraic varieties\", Ann. Mat. Pura e Applicata, in press.\nMore precisely, via this theory and a suitable generalization of Macaulay's\ntheorem we show that the differential of the period map vanishes on an\ninfinitesimal deformation if and only if certain explicitly given twisted\nvolume forms go in the generalized Jacobi ideal of $X$ via the cup product\nhomomorphism.\n",
"title": "On Green's proof of infinitesimal Torelli theorem for hypersurfaces"
}
| null | null | null | null | true | null |
15247
| null |
Default
| null | null |
null |
{
"abstract": " We analyze three-dimensional hydrodynamical simulations of the interaction of\njets and the bubbles they inflate with the intra-cluster medium (ICM), and show\nthat the heating of the ICM by mixing hot bubble gas with the ICM operates over\ntens of millions of years, and hence can smooth the sporadic activity of the\njets. The inflation process of hot bubbles by propagating jets forms many\nvortices, and these vortices mix the hot bubble gas with the ICM. The mixing,\nhence the heating of the ICM, starts immediately after the jets are launched,\nbut continues for tens of millions of years. We suggest that the smoothing of\nthe active galactic nucleus (AGN) sporadic activity by the long-lived vortices\naccounts for the recent finding of a gentle energy coupling between AGN heating\nand the ICM.\n",
"title": "Gentle heating by mixing in cooling flow clusters"
}
| null | null |
[
"Physics"
] | null | true | null |
15248
| null |
Validated
| null | null |
null |
{
"abstract": " This paper tackles the reduction of redundant repeating generation that is\noften observed in RNN-based encoder-decoder models. Our basic idea is to\njointly estimate the upper-bound frequency of each target vocabulary in the\nencoder and control the output words based on the estimation in the decoder.\nOur method shows significant improvement over a strong RNN-based\nencoder-decoder baseline and achieved its best results on an abstractive\nsummarization benchmark.\n",
"title": "Cutting-off Redundant Repeating Generations for Neural Abstractive Summarization"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
15249
| null |
Validated
| null | null |
null |
{
"abstract": " We consider the Lasso for a noiseless experiment where one has observations\n$X \\beta^0$ and uses the penalized version of basis pursuit. We compute for\nsome special designs the compatibility constant, a quantity closely related to\nthe restricted eigenvalue. We moreover show the dependence of the (penalized)\nprediction error on this compatibility constant. This exercise illustrates that\ncompatibility is necessarily entering into the bounds for the (penalized)\nprediction error and that the bounds in the literature therefore are - up to\nconstants - tight. We also give conditions that show that in the noisy case the\ndominating term for the prediction error is given by the prediction error of\nthe noiseless case.\n",
"title": "Some exercises with the Lasso and its compatibility constant"
}
| null | null | null | null | true | null |
15250
| null |
Default
| null | null |
null |
{
"abstract": " Transformer lifetime assessments plays a vital role in reliable operation of\npower systems. In this paper, leveraging sensory data, an approach in\nestimating transformer lifetime is presented. The winding hottest-spot\ntemperature, which is the pivotal driver that impacts transformer aging, is\nmeasured hourly via a temperature sensor, then transformer loss of life is\ncalculated based on the IEEE Std. C57.91-2011. A Cumulative Moving Average\n(CMA) model is subsequently applied to the data stream of the transformer loss\nof life to provide hourly estimates until convergence. Numerical examples\ndemonstrate the effectiveness of the proposed approach for the transformer\nlifetime estimation, and explores its efficiency and practical merits.\n",
"title": "Leveraging Sensory Data in Estimating Transformer Lifetime"
}
| null | null | null | null | true | null |
15251
| null |
Default
| null | null |
null |
{
"abstract": " Future projection of climate is typically obtained by combining outputs from\nmultiple Earth System Models (ESMs) for several climate variables such as\ntemperature and precipitation. While IPCC has traditionally used a simple model\noutput average, recent work has illustrated potential advantages of using a\nmultitask learning (MTL) framework for projections of individual climate\nvariables. In this paper we introduce a framework for hierarchical multitask\nlearning (HMTL) with two levels of tasks such that each super-task, i.e., task\nat the top level, is itself a multitask learning problem over sub-tasks. For\nclimate projections, each super-task focuses on projections of specific climate\nvariables spatially using an MTL formulation. For the proposed HMTL approach, a\ngroup lasso regularization is added to couple parameters across the\nsuper-tasks, which in the climate context helps exploit relationships among the\nbehavior of different climate variables at a given spatial location. We show\nthat some recent works on MTL based on learning task dependency structures can\nbe viewed as special cases of HMTL. Experiments on synthetic and real climate\ndata show that HMTL produces better results than decoupled MTL methods applied\nseparately on the super-tasks and HMTL significantly outperforms baselines for\nclimate projection.\n",
"title": "Spatial Projection of Multiple Climate Variables using Hierarchical Multitask Learning"
}
| null | null | null | null | true | null |
15252
| null |
Default
| null | null |
null |
{
"abstract": " Random impedance networks are widely used as a model to describe plasmon\nresonances in disordered metal-dielectric nanocomposites. In order to study\nthin films, two-dimensional networks are often used despite the fact that such\nnetworks correspond to a two-dimensional electrodynamics [J.P. Clerc et al, J.\nPhys. A 29, 4781 (1996)]. In the present work, we propose a model of\ntwo-dimensional systems with three-dimensional Coulomb interaction and show\nthat this model is equivalent to a planar network with long-range capacitive\nconnections between sites. In a case of a metal film, we get a known dispersion\n$\\omega \\propto \\sqrt{k}$ of plane-wave two-dimensional plasmons. In the\nframework of the proposed model, we study the evolution of resonances with\ndecreasing of metal filling factor. In the subcritical region with metal\nfilling $p$ lower than the percolation threshold $p_c$, we observe a gap with\nLifshitz tails in the spectral density of states (DOS). In the supercritical\nregion $p>p_c$, the DOS demonstrates a crossover between plane-wave\ntwo-dimensional plasmons and resonances associated with small clusters.\n",
"title": "Two-dimensional plasmons in the random impedance network model of disordered thin-film nanocomposites"
}
| null | null | null | null | true | null |
15253
| null |
Default
| null | null |
null |
{
"abstract": " Purpose: To compare two methods that use x-ray spectral information to image\nexternally administered contrast agents: K-edge subtraction and basis-function\ndecomposition (the A-space method), Methods: The K-edge method uses narrow band\nx-ray spectra with energies infinitesimally below and above the contrast\nmaterial K-edge energy. The A-space method uses a broad spectrum x-ray tube\nsource and measures the transmitted spectrum with photon counting detectors\nwith pulse height analysis. The methods are compared by their signal to noise\nratio (SNR) divided by the patient dose for an imaging task to decide whether\ncontrast material is present in a soft tissue background. The performance with\niodine or gadolinium containing contrast material is evaluated as a function of\nobject thickness and the x-ray tube voltage of the A-space method. Results: For\na tube voltages above 60 kV and soft tissue thicknesses from 5 to 25 g/cm^2,\nthe A-space method has a larger SNR per dose than the K-edge subtraction method\nfor either iodine or gadolinium containing contrast agent. Conclusion: Even\nwith the unrealistic spectra assumed for the K-edge method, the A-space method\nhas a substantially larger SNR per patient dose.\n",
"title": "K-edge subtraction vs. A-space processing for x-ray imaging of contrast agents: SNR"
}
| null | null | null | null | true | null |
15254
| null |
Default
| null | null |
null |
{
"abstract": " In many environments only a tiny subset of all states yield high reward. In\nthese cases, few of the interactions with the environment provide a relevant\nlearning signal. Hence, we may want to preferentially train on those\nhigh-reward states and the probable trajectories leading to them. To this end,\nwe advocate for the use of a backtracking model that predicts the preceding\nstates that terminate at a given high-reward state. We can train a model which,\nstarting from a high value state (or one that is estimated to have high value),\npredicts and sample for which the (state, action)-tuples may have led to that\nhigh value state. These traces of (state, action) pairs, which we refer to as\nRecall Traces, sampled from this backtracking model starting from a high value\nstate, are informative as they terminate in good states, and hence we can use\nthese traces to improve a policy. We provide a variational interpretation for\nthis idea and a practical algorithm in which the backtracking model samples\nfrom an approximate posterior distribution over trajectories which lead to\nlarge rewards. Our method improves the sample efficiency of both on- and\noff-policy RL algorithms across several environments and tasks.\n",
"title": "Recall Traces: Backtracking Models for Efficient Reinforcement Learning"
}
| null | null |
[
"Statistics"
] | null | true | null |
15255
| null |
Validated
| null | null |
null |
{
"abstract": " We study the dynamics of overdamped Brownian particles diffusing in\nconservative force fields and undergoing stochastic resetting to a given\nlocation with a generic space-dependent rate of resetting. We present a\nsystematic approach involving path integrals and elements of renewal theory\nthat allows to derive analytical expressions for a variety of statistics of the\ndynamics such as (i) the propagator prior to first reset; (ii) the distribution\nof the first-reset time, and (iii) the spatial distribution of the particle at\nlong times. We apply our approach to several representative and hitherto\nunexplored examples of resetting dynamics. A particularly interesting example\nfor which we find analytical expressions for the statistics of resetting is\nthat of a Brownian particle trapped in a harmonic potential with a rate of\nresetting that depends on the instantaneous energy of the particle. We find\nthat using energy-dependent resetting processes is more effective in achieving\nspatial confinement of Brownian particles on a faster timescale than by\nperforming quenches of parameters of the harmonic potential.\n",
"title": "Path-integral formalism for stochastic resetting: Exactly solved examples and shortcuts to confinement"
}
| null | null | null | null | true | null |
15256
| null |
Default
| null | null |
null |
{
"abstract": " The Ensemble Kalman methodology in an inverse problems setting can be viewed\nas an iterative scheme, which is a weakly tamed discretization scheme for a\ncertain stochastic differential equation (SDE). Assuming a suitable\napproximation result, dynamical properties of the SDE can be rigorously pulled\nback via the discrete scheme to the original Ensemble Kalman inversion.\nThe results of this paper make a step towards closing the gap of the missing\napproximation result by proving a strong convergence result in a simplified\nmodel of a scalar stochastic differential equation. We focus here on a toy\nmodel with similar properties than the one arising in the context of Ensemble\nKalman filter. The proposed model can be interpreted as a single particle\nfilter for a linear map and thus forms the basis for further analysis. The\ndifficulty in the analysis arises from the formally derived limiting SDE with\nnon-globally Lipschitz continuous nonlinearities both in the drift and in the\ndiffusion. Here the standard Euler-Maruyama scheme might fail to provide a\nstrongly convergent numerical scheme and taming is necessary. In contrast to\nthe strong taming usually used, the method presented here provides a weaker\nform of taming.\nWe present a strong convergence analysis by first proving convergence on a\ndomain of high probability by using a cut-off or localisation, which then\nleads, combined with bounds on moments for both the SDE and the numerical\nscheme, by a bootstrapping argument to strong convergence.\n",
"title": "A strongly convergent numerical scheme from Ensemble Kalman inversion"
}
| null | null | null | null | true | null |
15257
| null |
Default
| null | null |
null |
{
"abstract": " Positron Emission Tomography (PET) is a functional imaging modality widely\nused in neuroscience studies. To obtain meaningful quantitative results from\nPET images, attenuation correction is necessary during image reconstruction.\nFor PET/MR hybrid systems, PET attenuation is challenging as Magnetic Resonance\n(MR) images do not reflect attenuation coefficients directly. To address this\nissue, we present deep neural network methods to derive the continuous\nattenuation coefficients for brain PET imaging from MR images. With only Dixon\nMR images as the network input, the existing U-net structure was adopted and\nanalysis using forty patient data sets shows it is superior than other Dixon\nbased methods. When both Dixon and zero echo time (ZTE) images are available,\nwe have proposed a modified U-net structure, named GroupU-net, to efficiently\nmake use of both Dixon and ZTE information through group convolution modules\nwhen the network goes deeper. Quantitative analysis based on fourteen real\npatient data sets demonstrates that both network approaches can perform better\nthan the standard methods, and the proposed network structure can further\nreduce the PET quantification error compared to the U-net structure.\n",
"title": "Attenuation correction for brain PET imaging using deep neural network based on dixon and ZTE MR images"
}
| null | null | null | null | true | null |
15258
| null |
Default
| null | null |
null |
{
"abstract": " Electrochemistry is the underlying mechanism in a variety of energy\nconversion and storage systems, and it is well known that the composition,\nstructure, and properties of electrochemical materials near active interfaces\noften deviates substantially and inhomogeneously from the bulk properties. A\nuniversal challenge facing the development of electrochemical systems is our\nlack of understanding of physical and chemical rates at local length scales,\nand the recently developed electrochemical strain microscopy (ESM) provides a\npromising method to probe crucial local information regarding the underlying\nelectrochemical mechanisms. Here we develop a computational model that couples\nmechanics and electrochemistry relevant for ESM experiments, with the goal to\nenable quantitative analysis of electrochemical processes underneath a charged\nscanning probe. We show that the model captures the essence of a number of\ndifferent ESM experiments, making it possible to de-convolute local ionic\nconcentration and diffusivity via combined ESM mapping, spectroscopy, and\nrelaxation studies. Through the combination of ESM experiments and\ncomputations, it is thus possible to obtain deep insight into the local\nelectrochemistry at the nanoscale.\n",
"title": "Resolving Local Electrochemistry at the Nanoscale via Electrochemical Strain Microscopy: Modeling and Experiments"
}
| null | null |
[
"Physics"
] | null | true | null |
15259
| null |
Validated
| null | null |
null |
{
"abstract": " We build new algebraic structures, which we call genuine equivariant operads,\nwhich can be thought of as a hybrid between equivariant operads and coefficient\nsystems. We then prove an Elmendorf-Piacenza type theorem stating that\nequivariant operads, with their graph model structure, are equivalent to\ngenuine equivariant operads, with their projective model structure.\nAs an application, we build explicit models for the $N_{\\infty}$-operads of\nBlumberg and Hill.\n",
"title": "Genuine equivariant operads"
}
| null | null | null | null | true | null |
15260
| null |
Default
| null | null |
null |
{
"abstract": " Given the n vertices of a convex polygon in cyclic order, can the triangle of\nmaximum area inscribed in P be determined by an algorithm with O(n) time\ncomplexity? A purported linear-time algorithm by Dobkin and Snyder from 1979\nhas recently been shown to be incorrect by Keikha, Löffler, Urhausen, and van\nder Hoog. These authors give an alternative algorithm with O(n log n) time\ncomplexity. Here we give an algorithm with linear time complexity.\n",
"title": "A linear-time algorithm for the maximum-area inscribed triangle in a convex polygon"
}
| null | null | null | null | true | null |
15261
| null |
Default
| null | null |
null |
{
"abstract": " For the analysis of molecular processes, the estimation of time-scales, i.e.,\ntransition rates, is very important. Estimating the transition rates between\nmolecular conformations is -- from a mathematical point of view -- an invariant\nsubspace projection problem. A certain infinitesimal generator acting on\nfunction space is projected to a low-dimensional rate matrix. This projection\ncan be performed in two steps. First, the infinitesimal generator is\ndiscretized, then the invariant subspace is approxi-mated and used for the\nsubspace projection. In our approach, the discretization will be based on a\nVoronoi tessellation of the conformational space. We will show that the\ndiscretized infinitesimal generator can simply be approximated by the geometric\naverage of the Boltzmann weights of the Voronoi cells. Thus, there is a direct\ncorrela-tion between the potential energy surface of molecular structures and\nthe transition rates of conformational changes. We present results for a\n2d-diffusion process and Alanine dipeptide.\n",
"title": "Estimation of the infinitesimal generator by square-root approximation"
}
| null | null | null | null | true | null |
15262
| null |
Default
| null | null |
null |
{
"abstract": " We present microlensing events in the 2015 Korea Microlensing Telescope\nNetwork (KMTNet) data and our procedure for identifying these events. In\nparticular, candidates were detected with a novel \"completed event\"\nmicrolensing event-finder algorithm. The algorithm works by making linear fits\nto a (t0,teff,u0) grid of point-lens microlensing models. This approach is\nrendered computationally efficient by restricting u0 to just two values (0 and\n1), which we show is quite adequate. The implementation presented here is\nspecifically tailored to the commission-year character of the 2015 data, but\nthe algorithm is quite general and has already been applied to a completely\ndifferent (non-KMTNet) data set. We outline expected improvements for 2016 and\nfuture KMTNet data. The light curves of the 660 \"clear microlensing\" and 182\n\"possible microlensing\" events that were found in 2015 are presented along with\nour policy for their public release.\n",
"title": "Korea Microlensing Telescope Network Microlensing Events from 2015: Event-Finding Algorithm, Vetting, and Photometry"
}
| null | null | null | null | true | null |
15263
| null |
Default
| null | null |
null |
{
"abstract": " As researchers use computational methods to study complex social behaviors at\nscale, the validity of this computational social science depends on the\nintegrity of the data. On July 2, 2015, Jason Baumgartner published a dataset\nadvertised to include ``every publicly available Reddit comment'' which was\nquickly shared on Bittorrent and the Internet Archive. This data quickly became\nthe basis of many academic papers on topics including machine learning, social\nbehavior, politics, breaking news, and hate speech. We have discovered\nsubstantial gaps and limitations in this dataset which may contribute to bias\nin the findings of that research. In this paper, we document the dataset,\nsubstantial missing observations in the dataset, and the risks to research\nvalidity from those gaps. In summary, we identify strong risks to research that\nconsiders user histories or network analysis, moderate risks to research that\ncompares counts of participation, and lesser risk to machine learning research\nthat avoids making representative claims about behavior and participation on\nReddit.\n",
"title": "Caveat Emptor, Computational Social Science: Large-Scale Missing Data in a Widely-Published Reddit Corpus"
}
| null | null | null | null | true | null |
15264
| null |
Default
| null | null |
null |
{
"abstract": " Skeleton-based human action recognition has attracted a lot of research\nattention during the past few years. Recent works attempted to utilize\nrecurrent neural networks to model the temporal dependencies between the 3D\npositional configurations of human body joints for better analysis of human\nactivities in the skeletal data. The proposed work extends this idea to spatial\ndomain as well as temporal domain to better analyze the hidden sources of\naction-related information within the human skeleton sequences in both of these\ndomains simultaneously. Based on the pictorial structure of Kinect's skeletal\ndata, an effective tree-structure based traversal framework is also proposed.\nIn order to deal with the noise in the skeletal data, a new gating mechanism\nwithin LSTM module is introduced, with which the network can learn the\nreliability of the sequential data and accordingly adjust the effect of the\ninput data on the updating procedure of the long-term context representation\nstored in the unit's memory cell. Moreover, we introduce a novel multi-modal\nfeature fusion strategy within the LSTM unit in this paper. The comprehensive\nexperimental results on seven challenging benchmark datasets for human action\nrecognition demonstrate the effectiveness of the proposed method.\n",
"title": "Skeleton-Based Action Recognition Using Spatio-Temporal LSTM Network with Trust Gates"
}
| null | null | null | null | true | null |
15265
| null |
Default
| null | null |
null |
{
"abstract": " In this work, we propose a novel method for quantifying distances between\nToeplitz structured covariance matrices. By exploiting the spectral\nrepresentation of Toeplitz matrices, the proposed distance measure is defined\nbased on an optimal mass transport problem in the spectral domain. This may\nthen be interpreted in the covariance domain, suggesting a natural way of\ninterpolating and extrapolating Toeplitz matrices, such that the positive\nsemi-definiteness and the Toeplitz structure of these matrices are preserved.\nThe proposed distance measure is also shown to be contractive with respect to\nboth additive and multiplicative noise, and thereby allows for a quantification\nof the decreased distance between signals when these are corrupted by noise.\nFinally, we illustrate how this approach can be used for several applications\nin signal processing. In particular, we consider interpolation and\nextrapolation of Toeplitz matrices, as well as clustering problems and tracking\nof slowly varying stochastic processes.\n",
"title": "Interpolation and Extrapolation of Toeplitz Matrices via Optimal Mass Transport"
}
| null | null | null | null | true | null |
15266
| null |
Default
| null | null |
null |
{
"abstract": " A cognitive radar adapts the transmit waveform in response to changes in the\nradar and target environment. In this work, we analyze the recently proposed\nsub-Nyquist cognitive radar wherein the total transmit power in a multi-band\ncognitive waveform remains the same as its full-band conventional counterpart.\nFor such a system, we derive lower bounds on the mean-squared-error (MSE) of a\nsingle-target time delay estimate. We formulate a procedure to select the\noptimal bands, and recommend distribution of the total power in different bands\nto enhance the accuracy of delay estimation. In particular, using Cramér-Rao\nbounds, we show that equi-width subbands in cognitive radar always have better\ndelay estimation than the conventional radar. Further analysis using Ziv-Zakai\nbound reveals that cognitive radar performs well in low signal-to-noise (SNR)\nregions.\n",
"title": "Performance of time delay estimation in a cognitive radar"
}
| null | null | null | null | true | null |
15267
| null |
Default
| null | null |
null |
{
"abstract": " Imitation learning algorithms learn viable policies by imitating an expert's\nbehavior when reward signals are not available. Generative Adversarial\nImitation Learning (GAIL) is a state-of-the-art algorithm for learning policies\nwhen the expert's behavior is available as a fixed set of trajectories. We\nevaluate in terms of the expert's cost function and observe that the\ndistribution of trajectory-costs is often more heavy-tailed for GAIL-agents\nthan the expert at a number of benchmark continuous-control tasks. Thus,\nhigh-cost trajectories, corresponding to tail-end events of catastrophic\nfailure, are more likely to be encountered by the GAIL-agents than the expert.\nThis makes the reliability of GAIL-agents questionable when it comes to\ndeployment in risk-sensitive applications like robotic surgery and autonomous\ndriving. In this work, we aim to minimize the occurrence of tail-end events by\nminimizing tail risk within the GAIL framework. We quantify tail risk by the\nConditional-Value-at-Risk (CVaR) of trajectories and develop the Risk-Averse\nImitation Learning (RAIL) algorithm. We observe that the policies learned with\nRAIL show lower tail-end risk than those of vanilla GAIL. Thus the proposed\nRAIL algorithm appears as a potent alternative to GAIL for improved reliability\nin risk-sensitive applications.\n",
"title": "RAIL: Risk-Averse Imitation Learning"
}
| null | null | null | null | true | null |
15268
| null |
Default
| null | null |
null |
{
"abstract": " Shape memory alloys often show a complex hierarchical morphology in the\nmartensitic state. To understand the formation of this twin-within-twins\nmicrostructure, we examine epitaxial Ni-Mn-Ga films as a model system. In-situ\nscanning electron microscopy experiments show beautiful complex twinning\npatterns with a number of different mesoscopic twin boundaries and macroscopic\ntwin boundaries between already twinned regions. We explain the appearance and\ngeometry of these patterns by constructing an internally twinned martensitic\nnucleus, which can take the shape of a diamond or a parallelogram, within the\nbasic phenomenological theory of martensite. These nucleus contains already the\nseeds of different possible mesoscopic twin boundaries. Nucleation and growth\nof these nuclei determines the creation of the hierarchical space-filling\nmartensitic microstructure. This is in contrast to previous approaches to\nexplain a hierarchical martensitic microstructure. This new picture of creation\nand anisotropic, well-oriented growth of twinned martensitic nuclei explains\nthe morphology and exact geometrical features of our experimentally observed\ntwins-within-twins microstructure on the meso- and macroscopic scale.\n",
"title": "Nucleation and growth of hierarchical martensite in epitaxial shape memory films"
}
| null | null | null | null | true | null |
15269
| null |
Default
| null | null |
null |
{
"abstract": " Recent advances in representation learning on graphs, mainly leveraging graph\nconvolutional networks, have brought a substantial improvement on many\ngraph-based benchmark tasks. While novel approaches to learning node embeddings\nare highly suitable for node classification and link prediction, their\napplication to graph classification (predicting a single label for the entire\ngraph) remains mostly rudimentary, typically using a single global pooling step\nto aggregate node features or a hand-designed, fixed heuristic for hierarchical\ncoarsening of the graph structure. An important step towards ameliorating this\nis differentiable graph coarsening---the ability to reduce the size of the\ngraph in an adaptive, data-dependent manner within a graph neural network\npipeline, analogous to image downsampling within CNNs. However, the previous\nprominent approach to pooling has quadratic memory requirements during training\nand is therefore not scalable to large graphs. Here we combine several recent\nadvances in graph neural network design to demonstrate that competitive\nhierarchical graph classification results are possible without sacrificing\nsparsity. Our results are verified on several established graph classification\nbenchmarks, and highlight an important direction for future research in\ngraph-based neural networks.\n",
"title": "Towards Sparse Hierarchical Graph Classifiers"
}
| null | null | null | null | true | null |
15270
| null |
Default
| null | null |
null |
{
"abstract": " Deep convolutional neural networks (CNNs) based approaches are the\nstate-of-the-art in various computer vision tasks, including face recognition.\nConsiderable research effort is currently being directed towards further\nimproving deep CNNs by focusing on more powerful model architectures and better\nlearning techniques. However, studies systematically exploring the strengths\nand weaknesses of existing deep models for face recognition are still\nrelatively scarce in the literature. In this paper, we try to fill this gap and\nstudy the effects of different covariates on the verification performance of\nfour recent deep CNN models using the Labeled Faces in the Wild (LFW) dataset.\nSpecifically, we investigate the influence of covariates related to: image\nquality -- blur, JPEG compression, occlusion, noise, image brightness,\ncontrast, missing pixels; and model characteristics -- CNN architecture, color\ninformation, descriptor computation; and analyze their impact on the face\nverification performance of AlexNet, VGG-Face, GoogLeNet, and SqueezeNet. Based\non comprehensive and rigorous experimentation, we identify the strengths and\nweaknesses of the deep learning models, and present key areas for potential\nfuture research. Our results indicate that high levels of noise, blur, missing\npixels, and brightness have a detrimental effect on the verification\nperformance of all models, whereas the impact of contrast changes and\ncompression artifacts is limited. It has been found that the descriptor\ncomputation strategy and color information does not have a significant\ninfluence on performance.\n",
"title": "Strengths and Weaknesses of Deep Learning Models for Face Recognition Against Image Degradations"
}
| null | null | null | null | true | null |
15271
| null |
Default
| null | null |
null |
{
"abstract": " We introduce a Bayesian approach for modeling Voigt profiles in absorption\nspectroscopy and its implementation in the python package, BayesVP, publicly\navailable at this https URL. The code fits the\nabsorption line profiles within specified wavelength ranges and generates\nposterior distributions for the column density, Doppler parameter, and\nredshifts of the corresponding absorbers. The code uses publicly available\nefficient parallel sampling packages to sample posterior and thus can be run on\nparallel platforms. BayesVP supports simultaneous fitting for multiple\nabsorption components in high-dimensional parameter space. We provide other\nuseful utilities in the package, such as explicit specification of priors of\nmodel parameters, continuum model, Bayesian model comparison criteria, and\nposterior sampling convergence check.\n",
"title": "BayesVP: a Bayesian Voigt profile fitting package"
}
| null | null |
[
"Physics"
] | null | true | null |
15272
| null |
Validated
| null | null |
null |
{
"abstract": " We consider the minimization of submodular functions subject to ordering\nconstraints. We show that this optimization problem can be cast as a convex\noptimization problem on a space of uni-dimensional measures, with ordering\nconstraints corresponding to first-order stochastic dominance. We propose new\ndiscretization schemes that lead to simple and efficient algorithms based on\nzero-th, first, or higher order oracles; these algorithms also lead to\nimprovements without isotonic constraints. Finally, our experiments show that\nnon-convex loss functions can be much more robust to outliers for isotonic\nregression, while still leading to an efficient optimization problem.\n",
"title": "Efficient Algorithms for Non-convex Isotonic Regression through Submodular Optimization"
}
| null | null | null | null | true | null |
15273
| null |
Default
| null | null |
null |
{
"abstract": " The evolution from superconducting LiTi2O4-delta to insulating Li4Ti5O12 thin\nfilms has been studied by precisely adjusting the oxygen pressure during the\nsample fabrication process. In the superconducting LiTi2O4-delta films, with\nthe increase of oxygen pressure, the oxygen vacancies are filled, and the\nc-axis lattice constant decreases gradually. With the increase of the oxygen\npressure to a certain critical value, the c-axis lattice constant becomes\nstable, which implies that the Li4Ti5O12 phase comes into being. The process of\noxygen filling is manifested by the angular bright-field images of the scanning\ntransmission electron microscopy techniques. The temperature of\nmagnetoresistance changed from positive and negative shows a non-monotonous\nbehavior with the increase of oxygen pressure. The theoretical explanation of\nthe oxygen effects on the structure and superconductivity of LiTi2O4-delta has\nalso been discussed in this work.\n",
"title": "The effects of oxygen in spinel oxide Li1+xTi2-xO4-delta thin films"
}
| null | null | null | null | true | null |
15274
| null |
Default
| null | null |
null |
{
"abstract": " Classes of locally compact groups having qualitative uncertainty principle\nfor Gabor transform have been investigated. These include Moore groups,\nHeisenberg Group $\\mathbb{H}_n, \\mathbb{H}_{n} \\times D,$ where $D$ is discrete\ngroup and other low dimensional nilpotent Lie groups.\n",
"title": "Qualitative uncertainty principle for Gabor transform on certain locally compact groups"
}
| null | null | null | null | true | null |
15275
| null |
Default
| null | null |
null |
{
"abstract": " Additively separable hedonic games and fractional hedonic games have received\nconsiderable attention. They are coalition forming games of selfish agents\nbased on their mutual preferences. Most of the work in the literature\ncharacterizes the existence and structure of stable outcomes (i.e., partitions\nin coalitions), assuming that preferences are given. However, there is little\ndiscussion on this assumption. In fact, agents receive different utilities if\nthey belong to different partitions, and thus it is natural for them to declare\ntheir preferences strategically in order to maximize their benefit. In this\npaper we consider strategyproof mechanisms for additively separable hedonic\ngames and fractional hedonic games, that is, partitioning methods without\npayments such that utility maximizing agents have no incentive to lie about\ntheir true preferences. We focus on social welfare maximization and provide\nseveral lower and upper bounds on the performance achievable by strategyproof\nmechanisms for general and specific additive functions. In most of the cases we\nprovide tight or asymptotically tight results. All our mechanisms are simple\nand can be computed in polynomial time. Moreover, all the lower bounds are\nunconditional, that is, they do not rely on any computational or complexity\nassumptions.\n",
"title": "Strategyproof Mechanisms for Additively Separable Hedonic Games and Fractional Hedonic Games"
}
| null | null | null | null | true | null |
15276
| null |
Default
| null | null |
null |
{
"abstract": " We present a novel method to solve image analogy problems : it allows to\nlearn the relation between paired images present in training data, and then\ngeneralize and generate images that correspond to the relation, but were never\nseen in the training set. Therefore, we call the method Conditional Analogy\nGenerative Adversarial Network (CAGAN), as it is based on adversarial training\nand employs deep convolutional neural networks. An especially interesting\napplication of that technique is automatic swapping of clothing on fashion\nmodel photos. Our work has the following contributions. First, the definition\nof the end-to-end trainable CAGAN architecture, which implicitly learns\nsegmentation masks without expensive supervised labeling data. Second,\nexperimental results show plausible segmentation masks and often convincing\nswapped images, given the target article. Finally, we discuss the next steps\nfor that technique: neural network architecture improvements and more advanced\napplications.\n",
"title": "The Conditional Analogy GAN: Swapping Fashion Articles on People Images"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
15277
| null |
Validated
| null | null |
null |
{
"abstract": " The inverse problem of antiplane elasticity on determination of the profiles\nof $n$ uniformly stressed inclusions is studied. The inclusions are in ideal\ncontact with the surrounding matrix, the stress field inside the inclusions is\nuniform, and at infinity the body is subjected to antiplane uniform shear. The\nexterior of the inclusions, an $n$-connected domain, is treated as the image by\na conformal map of an $n$-connected slit domain with the slits lying in the\nsame line. The inverse problem is solved by quadratures by reducing it to two\nRiemann-Hilbert problems on a Riemann surface of genus $n-1$. Samples of two\nand three symmetric and non-symmetric uniformly stressed inclusions are\nreported.\n",
"title": "Inverse antiplane problem on $n$ uniformly stressed inclusions"
}
| null | null |
[
"Mathematics"
] | null | true | null |
15278
| null |
Validated
| null | null |
null |
{
"abstract": " We present a construction of a 2-Hilbert space of sections of a bundle gerbe,\na suitable candidate for a prequantum 2-Hilbert space in higher geometric\nquantisation. We introduce a direct sum on the morphism categories in the\n2-category of bundle gerbes and show that these categories are cartesian\nmonoidal and abelian. Endomorphisms of the trivial bundle gerbe, or higher\nfunctions, carry the structure of a rig-category, which acts on generic\nmorphism categories of bundle gerbes. We continue by presenting a\ncategorification of the hermitean metric on a hermitean line bundle. This is\nachieved by introducing a functorial dual that extends the dual of vector\nbundles to morphisms of bundle gerbes, and constructing a two-variable\nadjunction for the aforementioned rig-module category structure on morphism\ncategories. Its right internal hom is the module action, composed by taking the\ndual of higher functions, while the left internal hom is interpreted as a\nbundle gerbe metric. Sections of bundle gerbes are defined as morphisms from\nthe trivial bundle gerbe to a given bundle gerbe. The resulting categories of\nsections carry a rig-module structure over the category of finite-dimensional\nHilbert spaces. A suitable definition of 2-Hilbert spaces is given, modifying\nprevious definitions by the use of two-variable adjunctions. We prove that the\ncategory of sections of a bundle gerbe fits into this framework, thus obtaining\na 2-Hilbert space of sections. In particular, this can be constructed for\nprequantum bundle gerbes in problems of higher geometric quantisation. We\ndefine a dimensional reduction functor and show that the categorical structures\nintroduced on bundle gerbes naturally reduce to their counterparts on hermitean\nline bundles with connections. In several places in this thesis, we provide\nexamples, making 2-Hilbert spaces of sections and dimensional reduction very\nexplicit.\n",
"title": "Categorical Structures on Bundle Gerbes and Higher Geometric Prequantisation"
}
| null | null | null | null | true | null |
15279
| null |
Default
| null | null |
null |
{
"abstract": " This article describes the final solution of team monkeytyping, who finished\nin second place in the YouTube-8M video understanding challenge. The dataset\nused in this challenge is a large-scale benchmark for multi-label video\nclassification. We extend the work in [1] and propose several improvements for\nframe sequence modeling. We propose a network structure called Chaining that\ncan better capture the interactions between labels. Also, we report our\napproaches in dealing with multi-scale information and attention pooling. In\naddition, We find that using the output of model ensemble as a side target in\ntraining can boost single model performance. We report our experiments in\nbagging, boosting, cascade, and stacking, and propose a stacking algorithm\ncalled attention weighted stacking. Our final submission is an ensemble that\nconsists of 74 sub models, all of which are listed in the appendix.\n",
"title": "The Monkeytyping Solution to the YouTube-8M Video Understanding Challenge"
}
| null | null |
[
"Computer Science"
] | null | true | null |
15280
| null |
Validated
| null | null |
null |
{
"abstract": " Brain Electroencephalography (EEG) classification is widely applied to\nanalyze cerebral diseases in recent years. Unfortunately, invalid/noisy EEGs\ndegrade the diagnosis performance and most previously developed methods ignore\nthe necessity of EEG selection for classification. To this end, this paper\nproposes a novel maximum weight clique-based EEG selection approach, named\nmwcEEGs, to map EEG selection to searching maximum similarity-weighted cliques\nfrom an improved Fréchet distance-weighted undirected EEG graph\nsimultaneously considering edge weights and vertex weights. Our mwcEEGs\nimproves the classification performance by selecting intra-clique pairwise\nsimilar and inter-clique discriminative EEGs with similarity threshold\n$\\delta$. Experimental results demonstrate the algorithm effectiveness compared\nwith the state-of-the-art time series selection algorithms on real-world EEG\ndatasets.\n",
"title": "Brain EEG Time Series Selection: A Novel Graph-Based Approach for Classification"
}
| null | null |
[
"Statistics",
"Quantitative Biology"
] | null | true | null |
15281
| null |
Validated
| null | null |
null |
{
"abstract": " We present a case-study demonstrating the usefulness of Bayesian hierarchical\nmixture modelling for investigating cognitive processes. In sentence\ncomprehension, it is widely assumed that the distance between linguistic\nco-dependents affects the latency of dependency resolution: the longer the\ndistance, the longer the retrieval time (the distance-based account). An\nalternative theory, direct-access, assumes that retrieval times are a mixture\nof two distributions: one distribution represents successful retrievals (these\nare independent of dependency distance) and the other represents an initial\nfailure to retrieve the correct dependent, followed by a reanalysis that leads\nto successful retrieval. We implement both models as Bayesian hierarchical\nmodels and show that the direct-access model explains Chinese relative clause\nreading time data better than the distance account.\n",
"title": "Modelling dependency completion in sentence comprehension as a Bayesian hierarchical mixture process: A case study involving Chinese relative clauses"
}
| null | null | null | null | true | null |
15282
| null |
Default
| null | null |
null |
{
"abstract": " Ensemble Kalman filter (EnKF) is an important data assimilation method for\nhigh dimensional geophysical systems. Efficient implementation of EnKF in\npractice often involves the localization technique, which updates each\ncomponent using only information within a local radius. This paper rigorously\nanalyzes the local EnKF (LEnKF) for linear systems, and shows that the filter\nerror can be dominated by the ensemble covariance, as long as 1) the sample\nsize exceeds the logarithmic of state dimension and a constant that depends\nonly on the local radius; 2) the forecast covariance matrix admits a stable\nlocalized structure. In particular, this indicates that with small system and\nobservation noises, the filter error will be accurate in long time even if the\ninitialization is not. The analysis also reveals an intrinsic inconsistency\ncaused by the localization technique, and a stable localized structure is\nnecessary to control this inconsistency. While this structure is usually taken\nfor granted for the operation of LEnKF, it can also be rigorously proved for\nlinear systems with sparse local observations and weak local interactions.\nThese theoretical results are also validated by numerical implementation of\nLEnKF on a simple stochastic turbulence in two dynamical regimes.\n",
"title": "Performance analysis of local ensemble Kalman filter"
}
| null | null | null | null | true | null |
15283
| null |
Default
| null | null |
null |
{
"abstract": " This note is devoted to the study of the homology class of a compact Poisson\ntransversal in a Poisson manifold. For specific classes of Poisson structures,\nsuch as unimodular Poisson structures and Poisson manifolds with closed leaves,\nwe prove that all their compact Poisson transversals represent non-trivial\nhomology classes, generalizing the symplectic case. We discuss several examples\nin which this property does not hold, as well as a weaker version of this\nproperty, which holds for log-symplectic structures. Finally, we extend our\nresults to Dirac geometry.\n",
"title": "The homology class of a Poisson transversal"
}
| null | null | null | null | true | null |
15284
| null |
Default
| null | null |
null |
{
"abstract": " This paper provides an alternative approach to the theory of dynamic\nprogramming, designed to accommodate the kinds of recursive preference\nspecifications that have become popular in economic and financial analysis,\nwhile still supporting traditional additively separable rewards. The approach\nexploits the theory of monotone convex operators, which turns out to be well\nsuited to dynamic maximization. The intuition is that convexity is preserved\nunder maximization, so convexity properties found in preferences extend\nnaturally to the Bellman operator.\n",
"title": "Discrete Time Dynamic Programming with Recursive Preferences: Optimality and Applications"
}
| null | null | null | null | true | null |
15285
| null |
Default
| null | null |
null |
{
"abstract": " The construction of anisotropic triangulations is desirable for various\napplications, such as the numerical solving of partial differential equations\nand the representation of surfaces in graphics. To solve this notoriously\ndifficult problem in a practical way, we introduce the discrete Riemannian\nVoronoi diagram, a discrete structure that approximates the Riemannian Voronoi\ndiagram. This structure has been implemented and was shown to lead to good\ntriangulations in $\\mathbb{R}^2$ and on surfaces embedded in $\\mathbb{R}^3$ as\ndetailed in our experimental companion paper.\nIn this paper, we study theoretical aspects of our structure. Given a finite\nset of points $\\cal P$ in a domain $\\Omega$ equipped with a Riemannian metric,\nwe compare the discrete Riemannian Voronoi diagram of $\\cal P$ to its\nRiemannian Voronoi diagram. Both diagrams have dual structures called the\ndiscrete Riemannian Delaunay and the Riemannian Delaunay complex. We provide\nconditions that guarantee that these dual structures are identical. It then\nfollows from previous results that the discrete Riemannian Delaunay complex can\nbe embedded in $\\Omega$ under sufficient conditions, leading to an anisotropic\ntriangulation with curved simplices. Furthermore, we show that, under similar\nconditions, the simplices of this triangulation can be straightened.\n",
"title": "Anisotropic triangulations via discrete Riemannian Voronoi diagrams"
}
| null | null | null | null | true | null |
15286
| null |
Default
| null | null |
null |
{
"abstract": " Principal component analysis (PCA) and singular value decomposition (SVD) are\nwidely used in statistics, machine learning, and applied mathematics. It has\nbeen well studied in the case of homoskedastic noise, where the noise levels of\nthe contamination are homogeneous.\nIn this paper, we consider PCA and SVD in the presence of heteroskedastic\nnoise, which arises naturally in a range of applications. We introduce a\ngeneral framework for heteroskedastic PCA and propose an algorithm called\nHeteroPCA, which involves iteratively imputing the diagonal entries to remove\nthe bias due to heteroskedasticity. This procedure is computationally efficient\nand provably optimal under the generalized spiked covariance model. A key\ntechnical step is a deterministic robust perturbation analysis on the singular\nsubspace, which can be of independent interest. The effectiveness of the\nproposed algorithm is demonstrated in a suite of applications, including\nheteroskedastic low-rank matrix denoising, Poisson PCA, and SVD based on\nheteroskedastic and incomplete data.\n",
"title": "Heteroskedastic PCA: Algorithm, Optimality, and Applications"
}
| null | null | null | null | true | null |
15287
| null |
Default
| null | null |
null |
{
"abstract": " Twenty-seven years ago, one of the biggest societal changes in human history\nbegan slowly when the technical foundations for the World Wide Web were defined\nby Tim Berners-Lee. Ever since, the Web has grown exponentially, reaching far\nbeyond its original technical foundations and deeply affecting the world today\n- and even more so the society of the future. We have seen that the Web can\ninfluence the realization of human rights and even the pursuit of happiness.\nThe Web provides an infrastructure to help us to learn, to work, to communicate\nwith loved ones, and to provide entertainment. However, it also creates an\nenvironment affected by the digital divide between those who have and those who\ndo not have access. Additionally, the Web provides challenges we must\nunderstand if we are to find a viable balance between data ownership and\nprivacy protection, between over-whelming surveillance and the prevention of\nterrorism. For the Web to succeed, we need to understand its societal\nchallenges including increased crime, the impact of social platforms and\nsocio-economic discrimination, and we must work towards fairness, social\ninclusion, and open governance.\nTen Yars ago, the field of Web Science was created to explore the science\nunderlying the Web from a socio-technical perspective including its\nmathematical properties, engineering principles, and social impacts. Ten years\nlater, we are learning much as the interdisciplinary endeavor to understand the\nWeb's global information space continues to grow.\nIn this article we want to elicit the major lessons we have learned through\nWeb Science and make some cautious predictions of what to expect next.\n",
"title": "A Manifesto for Web Science @ 10"
}
| null | null | null | null | true | null |
15288
| null |
Default
| null | null |
null |
{
"abstract": " We present the second release of value-added catalogues of the LAMOST\nSpectroscopic Survey of the Galactic Anticentre (LSS-GAC DR2). The catalogues\npresent values of radial velocity $V_{\\rm r}$, atmospheric parameters ---\neffective temperature $T_{\\rm eff}$, surface gravity log$g$, metallicity\n[Fe/H], $\\alpha$-element to iron (metal) abundance ratio [$\\alpha$/Fe]\n([$\\alpha$/M]), elemental abundances [C/H] and [N/H], and absolute magnitudes\n${\\rm M}_V$ and ${\\rm M}_{K_{\\rm s}}$ deduced from 1.8 million spectra of 1.4\nmillion unique stars targeted by the LSS-GAC since September 2011 until June\n2014. The catalogues also give values of interstellar reddening, distance and\norbital parameters determined with a variety of techniques, as well as proper\nmotions and multi-band photometry from the far-UV to the mid-IR collected from\nthe literature and various surveys. Accuracies of radial velocities reach\n5kms$^{-1}$ for late-type stars, and those of distance estimates range between\n10 -- 30 per cent, depending on the spectral signal-to-noise ratios. Precisions\nof [Fe/H], [C/H] and [N/H] estimates reach 0.1dex, and those of [$\\alpha$/Fe]\nand [$\\alpha$/M] reach 0.05dex. The large number of stars, the contiguous sky\ncoverage, the simple yet non-trivial target selection function and the robust\nestimates of stellar radial velocities and atmospheric parameters, distances\nand elemental abundances, make the catalogues a valuable data set to study the\nstructure and evolution of the Galaxy, especially the solar-neighbourhood and\nthe outer disk.\n",
"title": "LAMOST Spectroscopic Survey of the Galactic Anticentre (LSS-GAC): the second release of value-added catalogues"
}
| null | null | null | null | true | null |
15289
| null |
Default
| null | null |
null |
{
"abstract": " Machine learning can impact people with legal or ethical consequences when it\nis used to automate decisions in areas such as insurance, lending, hiring, and\npredictive policing. In many of these scenarios, previous decisions have been\nmade that are unfairly biased against certain subpopulations, for example those\nof a particular race, gender, or sexual orientation. Since this past data may\nbe biased, machine learning predictors must account for this to avoid\nperpetuating or creating discriminatory practices. In this paper, we develop a\nframework for modeling fairness using tools from causal inference. Our\ndefinition of counterfactual fairness captures the intuition that a decision is\nfair towards an individual if it is the same in (a) the actual world and (b) a\ncounterfactual world where the individual belonged to a different demographic\ngroup. We demonstrate our framework on a real-world problem of fair prediction\nof success in law school.\n",
"title": "Counterfactual Fairness"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
15290
| null |
Validated
| null | null |
null |
{
"abstract": " Recently, resources and tasks were proposed to go beyond state tracking in\ndialogue systems. An example is the frame tracking task, which requires\nrecording multiple frames, one for each user goal set during the dialogue. This\nallows a user, for instance, to compare items corresponding to different goals.\nThis paper proposes a model which takes as input the list of frames created so\nfar during the dialogue, the current user utterance as well as the dialogue\nacts, slot types, and slot values associated with this utterance. The model\nthen outputs the frame being referenced by each triple of dialogue act, slot\ntype, and slot value. We show that on the recently published Frames dataset,\nthis model significantly outperforms a previously proposed rule-based baseline.\nIn addition, we propose an extensive analysis of the frame tracking task by\ndividing it into sub-tasks and assessing their difficulty with respect to our\nmodel.\n",
"title": "A Frame Tracking Model for Memory-Enhanced Dialogue Systems"
}
| null | null | null | null | true | null |
15291
| null |
Default
| null | null |
null |
{
"abstract": " A generic model for the shape optimization problems we consider in this paper\nis the optimization of the Dirichlet eigenvalues of the Laplace operator with a\nvolume constraint. We deal with an obstacle placement problem which can be\nformulated as the following eigenvalue optimization problem: Fix two positive\nreal numbers $r_1$ and $A$. We consider a disk $B\\subset \\mathbb{R}^2$ having\nradius $r_1$. We want to place an obstacle $P$ of area $A$ within $B$ so as to\nmaximize or minimize the fundamental Dirichlet eigenvalue $\\lambda_1$ for the\nLaplacian on $B\\setminus P$. That is, we want to study the behavior of the\nfunction $\\rho \\mapsto \\lambda_1(B\\setminus\\rho(P))$, where $\\rho$ runs over\nthe set of all rigid motions of the plane fixing the center of mass for $P$\nsuch that $\\rho(P)\\subset B$. In this paper, we consider a non-concentric\nobstacle placement problem. The extremal configurations correspond to the cases\nwhere an axis of symmetry of $P$ coincide with an axis of symmetry of $B$. We\nalso characterize the maximizing and the minimizing configurations in our main\nresult, viz., Theorem 4.1. Equation (6), Propositions 5.1 and 5.2 imply Theorem\n4.1. We give many different generalizations of our result. At the end, we\nprovide some numerical evidence to validate our main theorem for the case where\nthe obstacle $P$ has $\\mathbb{D}_4$ symmetry. For the $n$ odd case, we identify\nsome of the extremal configuration for $\\lambda_1$. We prove that equation (6)\nand Proposition 5.1 hold true for $n$ odd too. We highlight some of the\ndifficulties faced in proving Proposition 5.2 for this case. We provide\nnumerical evidence for $n=5$ and conjecture that Theorem 4.1 holds true for $n$\nodd too.\n",
"title": "How to place an obstacle having a dihedral symmetry centered at a given point inside a disk so as to optimize the fundamental Dirichlet eigenvalue"
}
| null | null | null | null | true | null |
15292
| null |
Default
| null | null |
null |
{
"abstract": " The manifold which admits a genus-$2$ reducible Heegaard splitting is one of\nthe $3$-sphere, $\\mathbb{S}^2 \\times \\mathbb{S}^1$, lens spaces and their\nconnected sums. For each of those manifolds except most lens spaces, the\nmapping class group of the genus-$2$ splitting was shown to be finitely\npresented. In this work, we study the remaining generic lens spaces, and show\nthat the mapping class group of the genus-$2$ Heegaard splitting is finitely\npresented for any lens space by giving its explicit presentation. As an\napplication, we show that the fundamental groups of the spaces of the genus-$2$\nHeegaard splittings of lens spaces are all finitely presented.\n",
"title": "The mapping class groups of reducible Heegaard splittings of genus two"
}
| null | null | null | null | true | null |
15293
| null |
Default
| null | null |
null |
{
"abstract": " The use of programming languages can wax and wane across the decades. We\nexamine the split-apply- combine pattern that is common in statistical\ncomputing, and consider how its invocation or implementation in languages like\nMATLAB and APL differ from R/dplyr. The differences in spelling illustrate how\nthe concept of linguistic relativity applies to programming languages in ways\nthat are analogous to human languages. Finally, we discuss how Julia, by being\na high performance yet general purpose dynamic language, allows its users to\nexpress different abstractions to suit individual preferences.\n",
"title": "Linguistic Relativity and Programming Languages"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
15294
| null |
Validated
| null | null |
null |
{
"abstract": " We establish the link between Mathematical Morphology and the map of\nAsplund's distances between a probe and a grey scale function, using the\nLogarithmic Image Processing scalar multiplication. We demonstrate that the map\nis the logarithm of the ratio between a dilation and an erosion of the function\nby a structuring function: the probe. The dilations and erosions are mappings\nfrom the lattice of the images into the lattice of the positive functions.\nUsing a flat structuring element, the expression of the map of Asplund's\ndistances can be simplified with a dilation and an erosion of the image; these\nmappings stays in the lattice of the images. We illustrate our approach by an\nexample of pattern matching with a non-flat structuring function.\n",
"title": "Double-sided probing by map of Asplund's distances using Logarithmic Image Processing in the framework of Mathematical Morphology"
}
| null | null |
[
"Computer Science",
"Mathematics"
] | null | true | null |
15295
| null |
Validated
| null | null |
null |
{
"abstract": " This book chapter introduces regression approaches and regression adjustment\nfor Approximate Bayesian Computation (ABC). Regression adjustment adjusts\nparameter values after rejection sampling in order to account for the imperfect\nmatch between simulations and observations. Imperfect match between simulations\nand observations can be more pronounced when there are many summary statistics,\na phenomenon coined as the curse of dimensionality. Because of this imperfect\nmatch, credibility intervals obtained with regression approaches can be\ninflated compared to true credibility intervals. The chapter presents the main\nconcepts underlying regression adjustment. A theorem that compares theoretical\nproperties of posterior distributions obtained with and without regression\nadjustment is presented. Last, a practical application of regression adjustment\nin population genetics shows that regression adjustment shrinks posterior\ndistributions compared to rejection approaches, which is a solution to avoid\ninflated credibility intervals.\n",
"title": "Regression approaches for Approximate Bayesian Computation"
}
| null | null | null | null | true | null |
15296
| null |
Default
| null | null |
null |
{
"abstract": " Data and knowledge representation are fundamental concepts in machine\nlearning. The quality of the representation impacts the performance of the\nlearning model directly. Feature learning transforms or enhances raw data to\nstructures that are effectively exploited by those models. In recent years,\nseveral works have been using complex networks for data representation and\nanalysis. However, no feature learning method has been proposed for such\ncategory of techniques. Here, we present an unsupervised feature learning\nmechanism that works on datasets with binary features. First, the dataset is\nmapped into a feature--sample network. Then, a multi-objective optimization\nprocess selects a set of new vertices to produce an enhanced version of the\nnetwork. The new features depend on a nonlinear function of a combination of\npreexisting features. Effectively, the process projects the input data into a\nhigher-dimensional space. To solve the optimization problem, we design two\nmetaheuristics based on the lexicographic genetic algorithm and the improved\nstrength Pareto evolutionary algorithm (SPEA2). We show that the enhanced\nnetwork contains more information and can be exploited to improve the\nperformance of machine learning methods. The advantages and disadvantages of\neach optimization strategy are discussed.\n",
"title": "Feature learning in feature-sample networks using multi-objective optimization"
}
| null | null | null | null | true | null |
15297
| null |
Default
| null | null |
null |
{
"abstract": " The artificial axon is a recently introduced synthetic assembly of supported\nlipid bilayers and voltage gated ion channels, displaying the basic\nelectrophysiology of nerve cells. Here we demonstrate the use of two artificial\naxons as control elements to achieve a simple task. Namely, we steer a remote\ncontrol car towards a light source, using the sensory input dependent firing\nrate of the axons as the control signal for turning left or right. We present\nthe result in the form of the analysis of a movie of the car approaching the\nlight source. In general terms, with this work we pursue a constructivist\napproach to exploring the nexus between machine language at the nerve cell\nlevel and behavior.\n",
"title": "Analog control with two Artificial Axons"
}
| null | null | null | null | true | null |
15298
| null |
Default
| null | null |
null |
{
"abstract": " In this paper a multi-objective mathematical model has been used to optimize\ngrinding parameters include workpiece speed, depth of cut and wheel speed which\nhighly affect the final surface quality. The mathematical model of the\noptimization problem consists of three conflict objective functions subject to\nwheel wear and production rate constraints. Exact methods can solve the NLP\nmodel in few seconds, therefore using Meta-heuristic algorithms which provide\nnear optimal solutions in not suitable. Considering this, five Multi-Objective\nDecision Making methods have been used to solve the multi-objective\nmathematical model using GAMS software to achieve the optimal parameters of the\ngrinding process. The Multi-Objective Decision Making methods provide different\neffective solutions where the decision maker can choose each solution in\ndifferent situations. Different criteria have been considered to evaluate the\nperformance of the five Multi-Objective Decision Making methods. Also,\nTechnique for Order of Preference by Similarity to Ideal Solution method has\nbeen used to obtain the priority of each method and determine which\nMulti-Objective Decision Making method performs better considering all criteria\nsimultaneously. The results indicated that Weighted Sum Method and Goal\nprogramming method are the best Multi-Objective Decision Making methods. The\nWeighted Sum Method and Goal programming provided solutions which are\ncompetitive to each other. In addition, these methods obtained solutions which\nhave minimum grinding time, cost and surface roughness among other\nMulti-Objective Decision Making methods.\n",
"title": "Designing a cost-time-quality-efficient grinding process using MODM methods"
}
| null | null |
[
"Computer Science"
] | null | true | null |
15299
| null |
Validated
| null | null |
null |
{
"abstract": " In this article we study the treewidth of the \\emph{display graph}, an\nauxiliary graph structure obtained from the fusion of phylogenetic (i.e.,\nevolutionary) trees at their leaves. Earlier work has shown that the treewidth\nof the display graph is bounded if the trees are in some formal sense\ntopologically similar. Here we further expand upon this relationship. We\nanalyse a number of reduction rules which are commonly used in the\nphylogenetics literature to obtain fixed parameter tractable algorithms. In\nsome cases (the \\emph{subtree} reduction) the reduction rules behave similarly\nwith respect to treewidth, while others (the \\emph{cluster} reduction) behave\nvery differently, and the behaviour of the \\emph{chain reduction} is\nparticularly intriguing because of its link with graph separators and forbidden\nminors. We also show that the gap between treewidth and Tree Bisection and\nReconnect (TBR) distance can be infinitely large, and that unlike, for example,\nplanar graphs the treewidth of the display graph can be as much as linear in\nits number of vertices. On a slightly different note we show that if a display\ngraph is formed from the fusion of a phylogenetic network and a tree, rather\nthan from two trees, the treewidth of the display graph is bounded whenever the\ntree can be topologically embedded (\"displayed\") within the network. This opens\nthe door to the formulation of the display problem in Monadic Second Order\nLogic (MSOL). A number of other auxiliary results are given. We conclude with a\ndiscussion and list a number of open problems.\n",
"title": "Treewidth distance on phylogenetic trees"
}
| null | null | null | null | true | null |
15300
| null |
Default
| null | null |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.