text
null | inputs
dict | prediction
null | prediction_agent
null | annotation
list | annotation_agent
null | multi_label
bool 1
class | explanation
null | id
stringlengths 1
5
| metadata
null | status
stringclasses 2
values | event_timestamp
null | metrics
null |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
null |
{
"abstract": " We investigate the effect of stress fluctuations on the stochastic dynamics\nof an inclusion embedded in a viscous gel. We show that, in non-equilibrium\nsystems, stress fluctuations give rise to an effective attraction towards the\nboundaries of the confining domain, which is reminiscent of an active Casimir\neffect. We apply this generic result to the dynamics of deformations of the\ncell nucleus and we demonstrate the appearance of a fluctuation maximum at a\ncritical level of activity, in agreement with recent experiments [E. Makhija,\nD. S. Jokhun, and G. V. Shivashankar, Proc. Natl. Acad. Sci. U.S.A. 113, E32\n(2016)].\n",
"title": "Maximal fluctuations of confined actomyosin gels: dynamics of the cell nucleus"
}
| null | null | null | null | true | null |
5101
| null |
Default
| null | null |
null |
{
"abstract": " Quantifying and estimating wildlife population sizes is a foundation of\nwildlife management. However, many carnivore species are cryptic, leading to\ninnate difficulties in estimating their populations. We evaluated the potential\nfor using more rigorous statistical models to estimate the populations of black\nbears (Ursus americanus) in Wisconisin, and their applicability to other\nfurbearers such as bobcats (Lynx rufus). To estimate black bear populations, we\ndeveloped an AAH state-space model in a Bayesian framework based on Norton\n(2015) that can account for variation in harvest and population demographics\nover time. Our state-space model created an accurate estimate of the black bear\npopulation in Wisconsin based on age-at-harvest data and improves on previous\nmodels by using little demographic data, no auxiliary data, and not being\nsensitive to initial population size. The increased accuracy of the AAH\nstate-space models should allow for better management by being able to set\naccurate quotas to ensure a sustainable harvest for the black bear population\nin each zone. We also evaluated the demography and annual harvest data of\nbobcats in Wisconsin to determine trends in harvest, method, and hunter\nparticipation in Wisconsin. Each trait of harvested bobcats that we tested\n(mass, male:female sex ratio, and age) changed over time, and because these\nwere interrelated, and we inferred that harvest selection for larger size\nbiased harvests in favor of a) larger, b) older, and c) male bobcats by hound\nhunters. We also found an increase in the proportion of bobcats that were\nharvested by hound hunting compared to trapping, and that hound hunters were\nmore likely to create taxidermy mounts than trappers. We also found that\ndecreasing available bobcat tags and increasing success have occurred over\ntime, and correlate with substantially increasing both hunter populations and\nhunter interest.\n",
"title": "Age-at-harvest models as monitoring and harvest management tools for Wisconsin carnivores"
}
| null | null | null | null | true | null |
5102
| null |
Default
| null | null |
null |
{
"abstract": " Information planning enables faster learning with fewer training examples. It\nis particularly applicable when training examples are costly to obtain. This\nwork examines the advantages of information planning for text data by focusing\non three supervised models: Naive Bayes, supervised LDA and deep neural\nnetworks. We show that planning based on entropy and mutual information\noutperforms random selection baseline and therefore accelerates learning.\n",
"title": "Information Planning for Text Data"
}
| null | null | null | null | true | null |
5103
| null |
Default
| null | null |
null |
{
"abstract": " Electron tracking based Compton imaging is a key technique to improve the\nsensitivity of Compton cameras by measuring the initial direction of recoiled\nelectrons. To realize this technique in semiconductor Compton cameras, we\npropose a new detector concept, Si-CMOS hybrid detector. It is a Si detector\nbump-bonded to a CMOS readout integrated circuit to obtain electron trajectory\nimages. To acquire the energy and the event timing, signals from N-side are\nalso read out in this concept. By using an ASIC for the N-side readout, the\ntiming resolution of few us is achieved. In this paper, we present the results\nof two prototypes with 20 um pitch pixels. The images of the recoiled electron\ntrajectories are obtained with them successfully. The energy resolutions (FWHM)\nare 4.1 keV (CMOS) and 1.4 keV (N-side) at 59.5 keV. In addition, we confirmed\nthat the initial direction of the electron is determined using the\nreconstruction algorithm based on the graph theory approach. These results show\nthat Si-CMOS hybrid detectors can be used for electron tracking based Compton\nimaging.\n",
"title": "Development of Si-CMOS hybrid detectors towards electron tracking based Compton imaging in semiconductor detectors"
}
| null | null | null | null | true | null |
5104
| null |
Default
| null | null |
null |
{
"abstract": " In this paper we establish the best constant of an anisotropic\nGagliardo-Nirenberg-type inequality related to the\nBenjamin-Ono-Zakharov-Kuznetsov equation. As an application of our results, we\nprove the uniform bound of solutions for such a equation in the energy space.\n",
"title": "Sharp constant of an anisotropic Gagliardo-Nirenberg-type inequality and applications"
}
| null | null | null | null | true | null |
5105
| null |
Default
| null | null |
null |
{
"abstract": " A draft addendum to ICH E9 has been released for public consultation in\nAugust 2017. The addendum focuses on two topics particularly relevant for\nrandomized confirmatory clinical trials: estimands and sensitivity analyses.\nThe need to amend ICH E9 grew out of the realization of a lack of alignment\nbetween the objectives of a clinical trial stated in the protocol and the\naccompanying quantification of the \"treatment effect\" reported in a regulatory\nsubmission. We embed time-to-event endpoints in the estimand framework, and\ndiscuss how the four estimand attributes described in the addendum apply to\ntime-to-event endpoints. We point out that if the proportional hazards\nassumption is not met, the estimand targeted by the most prevalent methods used\nto analyze time-to-event endpoints, logrank test and Cox regression, depends on\nthe censoring distribution. We discuss for a large randomized clinical trial\nhow the analyses for the primary and secondary endpoints as well as the\nsensitivity analyses actually performed in the trial can be seen in the context\nof the addendum. To the best of our knowledge, this is the first attempt to do\nso for a trial with a time-to-event endpoint. Questions that remain open with\nthe addendum for time-to-event endpoints and beyond are formulated, and\nrecommendations for planning of future trials are given. We hope that this will\nprovide a contribution to developing a common framework based on the final\nversion of the addendum that can be applied to design, protocols, statistical\nanalysis plans, and clinical study reports in the future.\n",
"title": "Treatment Effect Quantification for Time-to-event Endpoints -- Estimands, Analysis Strategies, and beyond"
}
| null | null |
[
"Statistics"
] | null | true | null |
5106
| null |
Validated
| null | null |
null |
{
"abstract": " Given a characteristic, we define a character of the Siegel modular group of\nlevel 2, the computations of their values are also obtained. By using our\ntheorems, some key theorems of Igusa [1] can be recovered.\n",
"title": "A character of Siegel modular group of level 2 from theta constants"
}
| null | null |
[
"Mathematics"
] | null | true | null |
5107
| null |
Validated
| null | null |
null |
{
"abstract": " We investigate a new class of topological antiferromagnetic (AF) Chern\ninsulators driven by electronic interactions in two-dimensional systems without\ninversion symmetry. Despite the absence of a net magnetization, AF Chern\ninsulators (AFCI) possess a nonzero Chern number $C$ and exhibit the quantum\nanomalous Hall effect (QAHE). Their existence is guaranteed by the bifurcation\nof the boundary line of Weyl points between a quantum spin Hall insulator and a\ntopologically trivial phase with the emergence of AF long-range order. As a\nconcrete example, we study the phase structure of the honeycomb lattice\nKane-Mele model as a function of the inversion-breaking ionic potential and the\nHubbard interaction. We find an easy $z$-axis $C=1$ AFCI phase and a spin-flop\ntransition to a topologically trivial $xy$-plane collinear antiferromagnet. We\npropose experimental realizations of the AFCI and QAHE in correlated electron\nmaterials and cold atom systems.\n",
"title": "Antiferromagnetic Chern insulators in non-centrosymmetric systems"
}
| null | null | null | null | true | null |
5108
| null |
Default
| null | null |
null |
{
"abstract": " The relation between a cosmological halo concentration and its mass (cMr) is\na powerful tool to constrain cosmological models of halo formation and\nevolution. On the scale of galaxy clusters the cMr has so far been determined\nmostly with X-ray and gravitational lensing data. The use of independent\ntechniques is helpful in assessing possible systematics. Here we provide one of\nthe few determinations of the cMr by the dynamical analysis of the\nprojected-phase-space distribution of cluster members. Based on the WINGS and\nOmegaWINGS data sets, we used the Jeans analysis with the MAMPOSSt technique to\ndetermine masses and concentrations for 49 nearby clusters, each of which has\n~60 spectroscopic members or more within the virial region, after removal of\nsubstructures. Our cMr is in statistical agreement with theoretical predictions\nbased on LambdaCDM cosmological simulations. Our cMr is different from most\nprevious observational determinations because of its flatter slope and lower\nnormalization. It is however in agreement with two recent cMr obtained using\nthe lensing technique on the CLASH and LoCuSS cluster data sets. In the future\nwe will extend our analysis to galaxy systems of lower mass and at higher\nredshifts.\n",
"title": "The concentration-mass relation of clusters of galaxies from the OmegaWINGS survey"
}
| null | null | null | null | true | null |
5109
| null |
Default
| null | null |
null |
{
"abstract": " Effects of the structural distortion associated with the $\\rm OsO_6$\noctahedral rotation and tilting on the electronic band structure and magnetic\nanisotropy energy for the $5d^3$ compound NaOsO$_3$ are investigated using the\ndensity functional theory (DFT) and within a three-orbital model. Comparison of\nthe essential features of the DFT band structures with the three-orbital model\nfor both the undistorted and distorted structures provides insight into the\norbital and directional asymmetry in the electron hopping terms resulting from\nthe structural distortion. The orbital mixing terms obtained in the transformed\nhopping Hamiltonian resulting from the octahedral rotations are shown to\naccount for the fine features in the DFT band structure. Staggered\nmagnetization and the magnetic character of states near the Fermi energy\nindicate weak coupling behavior.\n",
"title": "Effects of the structural distortion on the electronic band structure of {\\boldmath $\\rm Na Os O_3$} studied within density functional theory and a three-orbital model"
}
| null | null | null | null | true | null |
5110
| null |
Default
| null | null |
null |
{
"abstract": " If $E$ is an elliptic curve with a point of order two, then work of Klagsbrun\nand Lemke Oliver shows that the distribution of\n$\\dim_{\\mathbb{F}_2}\\mathrm{Sel}_\\phi(E^d/\\mathbb{Q}) - \\dim_{\\mathbb{F}_2}\n\\mathrm{Sel}_{\\hat\\phi}(E^{\\prime d}/\\mathbb{Q})$ within the quadratic twist\nfamily tends to the discrete normal distribution $\\mathcal{N}(0,\\frac{1}{2}\n\\log \\log X)$ as $X \\rightarrow \\infty$.\nWe consider the distribution of $\\mathrm{dim}_{\\mathbb{F}_2}\n\\mathrm{Sel}_\\phi(E^d/\\mathbb{Q})$ within such a quadratic twist family when\n$\\dim_{\\mathbb{F}_2} \\mathrm{Sel}_\\phi(E^d/\\mathbb{Q}) - \\dim_{\\mathbb{F}_2}\n\\mathrm{Sel}_{\\hat\\phi}(E^{\\prime d}/\\mathbb{Q})$ has a fixed value $u$.\nSpecifically, we show that for every $r$, the limiting probability that\n$\\dim_{\\mathbb{F}_2} \\mathrm{Sel}_\\phi(E^d/\\mathbb{Q}) = r$ is given by an\nexplicit constant $\\alpha_{r,u}$. The constants $\\alpha_{r,u}$ are closely\nrelated to the $u$-probabilities introduced in Cohen and Lenstra's work on the\ndistribution of class groups, and thus provide a connection between the\ndistribution of Selmer groups of elliptic curves and random abelian groups.\nOur analysis of this problem has two steps. The first step uses algebraic and\ncombinatorial methods to directly relate the ranks of the Selmer groups in\nquestion to the dimensions of the kernels of random $\\mathbb{F}_2$-matrices.\nThis proves that the density of twists with a given $\\phi$-Selmer rank $r$ is\ngiven by $\\alpha_{r,u}$ for an unusual notion of density. The second step of\nthe analysis utilizes techniques from analytic number theory to show that this\nresult implies the correct asymptotics in terms of the natural notion of\ndensity.\n",
"title": "On the Joint Distribution Of $\\mathrm{Sel}_ϕ(E/\\mathbb{Q})$ and $\\mathrm{Sel}_{\\hatϕ}(E^\\prime/\\mathbb{Q})$ in Quadratic Twist Families"
}
| null | null | null | null | true | null |
5111
| null |
Default
| null | null |
null |
{
"abstract": " Machine learning is essentially the sciences of playing with data. An\nadaptive data selection strategy, enabling to dynamically choose different data\nat various training stages, can reach a more effective model in a more\nefficient way. In this paper, we propose a deep reinforcement learning\nframework, which we call \\emph{\\textbf{N}eural \\textbf{D}ata \\textbf{F}ilter}\n(\\textbf{NDF}), to explore automatic and adaptive data selection in the\ntraining process. In particular, NDF takes advantage of a deep neural network\nto adaptively select and filter important data instances from a sequential\nstream of training data, such that the future accumulative reward (e.g., the\nconvergence speed) is maximized. In contrast to previous studies in data\nselection that is mainly based on heuristic strategies, NDF is quite generic\nand thus can be widely suitable for many machine learning tasks. Taking neural\nnetwork training with stochastic gradient descent (SGD) as an example,\ncomprehensive experiments with respect to various neural network modeling\n(e.g., multi-layer perceptron networks, convolutional neural networks and\nrecurrent neural networks) and several applications (e.g., image classification\nand text understanding) demonstrate that NDF powered SGD can achieve comparable\naccuracy with standard SGD process by using less data and fewer iterations.\n",
"title": "Learning What Data to Learn"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
5112
| null |
Validated
| null | null |
null |
{
"abstract": " Deformation estimation of elastic object assuming an internal organ is\nimportant for the computer navigation of surgery. The aim of this study is to\nestimate the deformation of an entire three-dimensional elastic object using\ndisplacement information of very few observation points. A learning approach\nwith a neural network was introduced to estimate the entire deformation of an\nobject. We applied our method to two elastic objects; a rectangular\nparallelepiped model, and a human liver model reconstructed from computed\ntomography data. The average estimation error for the human liver model was\n0.041 mm when the object was deformed up to 66.4 mm, from only around 3 %\nobservations. These results indicate that the deformation of an entire elastic\nobject can be estimated with an acceptable level of error from limited\nobservations by applying a trained neural network to a new deformation.\n",
"title": "Deformation estimation of an elastic object by partial observation using a neural network"
}
| null | null | null | null | true | null |
5113
| null |
Default
| null | null |
null |
{
"abstract": " We have performed an empirical comparison of two distinct notions of discrete\nRicci curvature for graphs or networks, namely, the Forman-Ricci curvature and\nOllivier-Ricci curvature. Importantly, these two discretizations of the Ricci\ncurvature were developed based on different properties of the classical smooth\nnotion, and thus, the two notions shed light on different aspects of network\nstructure and behavior. Nevertheless, our extensive computational analysis in a\nwide range of both model and real-world networks shows that the two\ndiscretizations of Ricci curvature are highly correlated in many networks.\nMoreover, we show that if one considers the augmented Forman-Ricci curvature\nwhich also accounts for the two-dimensional simplicial complexes arising in\ngraphs, the observed correlation between the two discretizations is even\nhigher, especially, in real networks. Besides the potential theoretical\nimplications of these observations, the close relationship between the two\ndiscretizations has practical implications whereby Forman-Ricci curvature can\nbe employed in place of Ollivier-Ricci curvature for faster computation in\nlarger real-world networks whenever coarse analysis suffices.\n",
"title": "Comparative analysis of two discretizations of Ricci curvature for complex networks"
}
| null | null | null | null | true | null |
5114
| null |
Default
| null | null |
null |
{
"abstract": " We present a deep generative model for learning to predict classes not seen\nat training time. Unlike most existing methods for this problem, that represent\neach class as a point (via a semantic embedding), we represent each seen/unseen\nclass using a class-specific latent-space distribution, conditioned on class\nattributes. We use these latent-space distributions as a prior for a supervised\nvariational autoencoder (VAE), which also facilitates learning highly\ndiscriminative feature representations for the inputs. The entire framework is\nlearned end-to-end using only the seen-class training data. The model infers\ncorresponding attributes of a test image by maximizing the VAE lower bound; the\ninferred attributes may be linked to labels not seen when training. We further\nextend our model to a (1) semi-supervised/transductive setting by leveraging\nunlabeled unseen-class data via an unsupervised learning module, and (2)\nfew-shot learning where we also have a small number of labeled inputs from the\nunseen classes. We compare our model with several state-of-the-art methods\nthrough a comprehensive set of experiments on a variety of benchmark data sets.\n",
"title": "Zero-Shot Learning via Class-Conditioned Deep Generative Models"
}
| null | null | null | null | true | null |
5115
| null |
Default
| null | null |
null |
{
"abstract": " We investigate the dynamics of a nonlinear system modeling tumor growth with\ndrug application. The tumor is viewed as a mixture consisting of proliferating,\nquiescent and dead cells as well as a nutrient in the presence of a drug. The\nsystem is given by a multi-phase flow model: the densities of the different\ncells are governed by a set of transport equations, the density of the nutrient\nand the density of the drug are governed by rather general diffusion equations,\nwhile the velocity of the tumor is given by Darcy's equation. The domain\noccupied by the tumor in this setting is a growing continuum $\\Omega$ with\nboundary $\\partial \\Omega$ both of which evolve in time. Global-in-time weak\nsolutions are obtained using an approach based on the vanishing viscosity of\nthe Brinkman's regularization. Both the solutions and the domain are rather\ngeneral, no symmetry assumption is required and the result holds for large\ninitial data.\n",
"title": "On the vanishing viscosity approximation of a nonlinear model for tumor growth"
}
| null | null |
[
"Mathematics"
] | null | true | null |
5116
| null |
Validated
| null | null |
null |
{
"abstract": " I present a family of algorithms to reduce noise in astrophysical im- ages\nand image sequences, preserving more information from the original data than is\nretained by conventional techniques. The family uses locally adaptive filters\n(\"noise gates\") in the Fourier domain, to separate coherent image structure\nfrom background noise based on the statistics of local neighborhoods in the\nimage. Processing of solar data limited by simple shot noise or by additive\nnoise reveals image structure not easily visible in the originals, preserves\nphotometry of observable features, and reduces shot noise by a factor of 10 or\nmore with little to no apparent loss of resolution, revealing faint features\nthat were either not directly discernible or not sufficiently strongly detected\nfor quantitative analysis. The method works best on image sequences containing\nrelated subjects, for example movies of solar evolution, but is also applicable\nto single images provided that there are enough pixels. The adaptive filter\nuses the statistical properties of noise and of local neighborhoods in the\ndata, to discriminate between coherent features and incoherent noise without\nreference to the specific shape or evolution of the those features. The\ntechnique can potentially be modified in a straightforward way to exploit\nadditional a priori knowledge about the functional form of the noise.\n",
"title": "Noise-gating to clean astrophysical image data"
}
| null | null |
[
"Physics"
] | null | true | null |
5117
| null |
Validated
| null | null |
null |
{
"abstract": " Dropout, a stochastic regularisation technique for training of neural\nnetworks, has recently been reinterpreted as a specific type of approximate\ninference algorithm for Bayesian neural networks. The main contribution of the\nreinterpretation is in providing a theoretical framework useful for analysing\nand extending the algorithm. We show that the proposed framework suffers from\nseveral issues; from undefined or pathological behaviour of the true posterior\nrelated to use of improper priors, to an ill-defined variational objective due\nto singularity of the approximating distribution relative to the true\nposterior. Our analysis of the improper log uniform prior used in variational\nGaussian dropout suggests the pathologies are generally irredeemable, and that\nthe algorithm still works only because the variational formulation annuls some\nof the pathologies. To address the singularity issue, we proffer Quasi-KL (QKL)\ndivergence, a new approximate inference objective for approximation of\nhigh-dimensional distributions. We show that motivations for variational\nBernoulli dropout based on discretisation and noise have QKL as a limit.\nProperties of QKL are studied both theoretically and on a simple practical\nexample which shows that the QKL-optimal approximation of a full rank Gaussian\nwith a degenerate one naturally leads to the Principal Component Analysis\nsolution.\n",
"title": "Variational Bayesian dropout: pitfalls and fixes"
}
| null | null | null | null | true | null |
5118
| null |
Default
| null | null |
null |
{
"abstract": " Wind energy forecasting helps to manage power production, and hence, reduces\nenergy cost. Deep Neural Networks (DNN) mimics hierarchical learning in the\nhuman brain and thus possesses hierarchical, distributed, and multi-task\nlearning capabilities. Based on aforementioned characteristics, we report Deep\nBelief Network (DBN) based forecast engine for wind power prediction because of\nits good generalization and unsupervised pre-training attributes. The proposed\nDBN-WP forecast engine, which exhibits stochastic feature generation\ncapabilities and is composed of multiple Restricted Boltzmann Machines,\ngenerates suitable features for wind power prediction using atmospheric\nproperties as input. DBN-WP, due to its unsupervised pre-training of RBM layers\nand generalization capabilities, is able to learn the fluctuations in the\nmeteorological properties and thus is able to perform effective mapping of the\nwind power. In the deep network, a regression layer is appended at the end to\npredict sort-term wind power. It is experimentally shown that the deep learning\nand unsupervised pre-training capabilities of DBN based model has comparable\nand in some cases better results than hybrid and complex learning techniques\nproposed for wind power prediction. The proposed prediction system based on\nDBN, achieves mean values of RMSE, MAE and SDE as 0.124, 0.083 and 0.122,\nrespectively. Statistical analysis of several independent executions of the\nproposed DBN-WP wind power prediction system demonstrates the stability of the\nsystem. The proposed DBN-WP architecture is easy to implement and offers\ngeneralization as regards the change in location of the wind farm is concerned.\n",
"title": "Deep Belief Networks Based Feature Generation and Regression for Predicting Wind Power"
}
| null | null | null | null | true | null |
5119
| null |
Default
| null | null |
null |
{
"abstract": " First-order iterative optimization methods play a fundamental role in large\nscale optimization and machine learning. This paper presents control\ninterpretations for such optimization methods. First, we give loop-shaping\ninterpretations for several existing optimization methods and show that they\nare composed of basic control elements such as PID and lag compensators. Next,\nwe apply the small gain theorem to draw a connection between the convergence\nrate analysis of optimization methods and the input-output gain computations of\ncertain complementary sensitivity functions. These connections suggest that\nstandard classical control synthesis tools may be brought to bear on the design\nof optimization algorithms.\n",
"title": "Control Interpretations for First-Order Optimization Methods"
}
| null | null | null | null | true | null |
5120
| null |
Default
| null | null |
null |
{
"abstract": " Machine understanding of complex images is a key goal of artificial\nintelligence. One challenge underlying this task is that visual scenes contain\nmultiple inter-related objects, and that global context plays an important role\nin interpreting the scene. A natural modeling framework for capturing such\neffects is structured prediction, which optimizes over complex labels, while\nmodeling within-label interactions. However, it is unclear what principles\nshould guide the design of a structured prediction model that utilizes the\npower of deep learning components. Here we propose a design principle for such\narchitectures that follows from a natural requirement of permutation\ninvariance. We prove a necessary and sufficient characterization for\narchitectures that follow this invariance, and discuss its implication on model\ndesign. Finally, we show that the resulting model achieves new state of the art\nresults on the Visual Genome scene graph labeling benchmark, outperforming all\nrecent approaches.\n",
"title": "Mapping Images to Scene Graphs with Permutation-Invariant Structured Prediction"
}
| null | null | null | null | true | null |
5121
| null |
Default
| null | null |
null |
{
"abstract": " Currently, approximately 30% of epileptic patients treated with antiepileptic\ndrugs (AEDs) remain resistant to treatment (known as refractory patients). This\nproject seeks to understand the underlying similarities in refractory patients\nvs. other epileptic patients, identify features contributing to drug resistance\nacross underlying phenotypes for refractory patients, and develop predictive\nmodels for drug resistance in epileptic patients. In this study, epileptic\npatient data was examined to attempt to observe discernable similarities or\ndifferences in refractory patients (case) and other non-refractory patients\n(control) to map underlying mechanisms in causality. For the first part of the\nstudy, unsupervised algorithms such as Kmeans, Spectral Clustering, and\nGaussian Mixture Models were used to examine patient features projected into a\nlower dimensional space. Results from this study showed a high degree of\nnon-linearity in the underlying feature space. For the second part of this\nstudy, classification algorithms such as Logistic Regression, Gradient Boosted\nDecision Trees, and SVMs, were tested on the reduced-dimensionality features,\nwith accuracy results of 0.83(+/-0.3) testing using 7 fold cross validation.\nObservations of test results indicate using a radial basis function kernel PCA\nto reduce features ingested by a Gradient Boosted Decision Tree Ensemble lead\nto gains in improved accuracy in mapping a binary decision to highly non-linear\nfeatures collected from epileptic patients.\n",
"title": "Identifying Similarities in Epileptic Patients for Drug Resistance Prediction"
}
| null | null | null | null | true | null |
5122
| null |
Default
| null | null |
null |
{
"abstract": " Disentangled representations, where the higher level data generative factors\nare reflected in disjoint latent dimensions, offer several benefits such as\nease of deriving invariant representations, transferability to other tasks,\ninterpretability, etc. We consider the problem of unsupervised learning of\ndisentangled representations from large pool of unlabeled observations, and\npropose a variational inference based approach to infer disentangled latent\nfactors. We introduce a regularizer on the expectation of the approximate\nposterior over observed data that encourages the disentanglement. We also\npropose a new disentanglement metric which is better aligned with the\nqualitative disentanglement observed in the decoder's output. We empirically\nobserve significant improvement over existing methods in terms of both\ndisentanglement and data likelihood (reconstruction quality).\n",
"title": "Variational Inference of Disentangled Latent Concepts from Unlabeled Observations"
}
| null | null | null | null | true | null |
5123
| null |
Default
| null | null |
null |
{
"abstract": " The color of hot-dip galvanized steel sheet was adjusted in a reproducible\nway using a liquid Zn-Ti metallic bath, air atmosphere, and controlling the\nbath temperature as the only experimental parameter. Coloring was found only\nfor sample s cooled in air and dipped into Ti-containing liquid Zn. For samples\ndipped into a 0.15 wt pct Ti-containing Zn bath, the color remained metallic\n(gray) below a 792 K (519 C) bath temperature; it was yellow at 814 K, violet\nat 847 K, and blue at 873 K. With the increasing bath temperature, the\nthickness of the adhered Zn-Ti layer gradually decreased from 52 to 32\nmicrometers, while the thickness of the outer TiO2 layer gradually increased\nfrom 24 to 69 nm. Due to small Al contamination of the Zn bath, a thin (around\n2 nm) alumina-rich layer is found between the outer TiO2 layer and the inner\nmacroscopic Zn layer. It is proven that the color change was governed by the\nformation of thin outer TiO2 layer; different colors appear depending on the\nthickness of this layer, mostly due to the destructive interference of visible\nlight on this transparent nano-layer. A complex model was built to explain the\nresults using known relationships of chemical thermodynamics, adhesion, heat\nflow, kinetics of chemical reactions, diffusion, and optics.\n",
"title": "Designing the color of hot-dip galvanized steel sheet through destructive light interference using a Zn-Ti liquid metallic bath"
}
| null | null | null | null | true | null |
5124
| null |
Default
| null | null |
null |
{
"abstract": " We fix a counting function of multiplicities of algebraic points in a\nprojective hypersurface over a number field, and take the sum over all\nalgebraic points of bounded height and fixed degree. An upper bound for the sum\nwith respect to this counting function will be given in terms of the degree of\nthe hypersurface, the dimension of the singular locus, the upper bounds of\nheight, and the degree of the field of definition.\n",
"title": "Counting Multiplicities in a Hypersurface over a Number Field"
}
| null | null | null | null | true | null |
5125
| null |
Default
| null | null |
null |
{
"abstract": " We consider the task of fine-grained sentiment analysis from the perspective\nof multiple instance learning (MIL). Our neural model is trained on document\nsentiment labels, and learns to predict the sentiment of text segments, i.e.\nsentences or elementary discourse units (EDUs), without segment-level\nsupervision. We introduce an attention-based polarity scoring method for\nidentifying positive and negative text snippets and a new dataset which we call\nSPOT (as shorthand for Segment-level POlariTy annotations) for evaluating\nMIL-style sentiment models like ours. Experimental results demonstrate superior\nperformance against multiple baselines, whereas a judgement elicitation study\nshows that EDU-level opinion extraction produces more informative summaries\nthan sentence-based alternatives.\n",
"title": "Multiple Instance Learning Networks for Fine-Grained Sentiment Analysis"
}
| null | null | null | null | true | null |
5126
| null |
Default
| null | null |
null |
{
"abstract": " Computing-in-Memory (CiM) architectures aim to reduce costly data transfers\nby performing arithmetic and logic operations in memory and hence relieve the\npressure due to the memory wall. However, determining whether a given workload\ncan really benefit from CiM, which memory hierarchy and what device technology\nshould be adopted by a CiM architecture requires in-depth study that is not\nonly time consuming but also demands significant expertise in architectures and\ncompilers. This paper presents an energy evaluation framework, Eva-CiM, for\nsystems based on CiM architectures. Eva-CiM encompasses a multi-level (from\ndevice to architecture) comprehensive tool chain by leveraging existing\nmodeling and simulation tools such as GEM5, McPAT [2] and DESTINY [3]. To\nsupport high-confidence prediction, rapid design space exploration and ease of\nuse, Eva-CiM introduces several novel modeling/analysis approaches including\nmodels for capturing memory access and dependency-aware ISA traces, and for\nquantifying interactions between the host CPU and CiM modules. Eva-CiM can\nreadily produce energy estimates of the entire system for a given program, a\nprocessor architecture, and the CiM array and technology specifications.\nEva-CiM is validated by comparing with DESTINY [3] and [4], and enables\nfindings including practical contributions from CiM-supported accesses,\nCiM-sensitive benchmarking as well as the pros and cons of increased memory\nsize for CiM. Eva-CiM also enables exploration over different configurations\nand device technologies, showing 1.3-6.0X energy improvement for SRAM and\n2.0-7.9X for FeFET-RAM, respectively.\n",
"title": "Eva-CiM: A System-Level Energy Evaluation Framework for Computing-in-Memory Architectures"
}
| null | null | null | null | true | null |
5127
| null |
Default
| null | null |
null |
{
"abstract": " Previous work has shown that the one-dimensional (1D) inviscid compressible\nflow (Euler) equations admit a wide variety of scale-invariant solutions\n(including the famous Noh, Sedov, and Guderley shock solutions) when the\nincluded equation of state (EOS) closure model assumes a certain\nscale-invariant form. However, this scale-invariant EOS class does not include\neven simple models used for shock compression of crystalline solids, including\nmany broadly applicable representations of Mie-Grüneisen EOS. Intuitively,\nthis incompatibility naturally arises from the presence of multiple dimensional\nscales in the Mie-Grüneisen EOS, which are otherwise absent from\nscale-invariant models that feature only dimensionless parameters (such as the\nadiabatic index in the ideal gas EOS). The current work extends previous\nefforts intended to rectify this inconsistency, by using a scale-invariant EOS\nmodel to approximate a Mie- Grüneisen EOS form. To this end, the adiabatic\nbulk modulus for the Mie-Grüneisen EOS is constructed, and its key features\nare used to motivate the selection of a scale-invariant approximation form. The\nremaining surrogate model parameters are selected through enforcement of the\nRankine-Hugoniot jump conditions for an infinitely strong shock in a\nMie-Grüneisen material. Finally, the approximate EOS is used in conjunction\nwith the 1D inviscid Euler equations to calculate a semi-analytical,\nGuderley-like imploding shock solution in a metal sphere, and to determine if\nand when the solution may be valid for the underlying Mie-Grüneisen EOS.\n",
"title": "Converging Shock Flows for a Mie-Grüneisen Equation of State"
}
| null | null |
[
"Physics"
] | null | true | null |
5128
| null |
Validated
| null | null |
null |
{
"abstract": " Consider jointly Gaussian random variables whose conditional independence\nstructure is specified by a graphical model. If we observe realizations of the\nvariables, we can compute the covariance matrix, and it is well known that the\nsupport of the inverse covariance matrix corresponds to the edges of the\ngraphical model. Instead, suppose we only have noisy observations. If the noise\nat each node is independent, we can compute the sum of the covariance matrix\nand an unknown diagonal. The inverse of this sum is (in general) dense. We ask:\ncan the original independence structure be recovered? We address this question\nfor tree structured graphical models. We prove that this problem is\nunidentifiable, but show that this unidentifiability is limited to a small\nclass of candidate trees. We further present additional constraints under which\nthe problem is identifiable. Finally, we provide an O(n^3) algorithm to find\nthis equivalence class of trees.\n",
"title": "Robust estimation of tree structured Gaussian Graphical Model"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
5129
| null |
Validated
| null | null |
null |
{
"abstract": " Follower count is a factor that quantifies the popularity of celebrities. It\nis a reflection of their power, prestige and overall social reach. In this\npaper we investigate whether the social connectivity or the language choice is\nmore correlated to the future follower count of a celebrity. We collect data\nabout tweets, retweets and mentions of 471 Indian celebrities with verified\nTwitter accounts. We build two novel networks to approximate social\nconnectivity of the celebrities. We study various structural properties of\nthese two networks and observe their correlations with future follower counts.\nIn parallel, we analyze the linguistic structure of the tweets (LIWC features,\nsyntax and sentiment features and style and readability features) and observe\nthe correlations of each of these with the future follower count of a\ncelebrity. As a final step we use there features to classify a celebrity in a\nspecific bucket of future follower count (HIGH, MID or LOW). We observe that\nthe network features alone achieve an accuracy of 0.52 while the linguistic\nfeatures alone achieve an accuracy of 0.69 grossly outperforming the network\nfeatures. The network and linguistic features in conjunction produce an\naccuracy of 0.76. We also discuss some final insights that we obtain from\nfurther data analysis celebrities with larger follower counts post tweets that\nhave (i) more words from friend and family LIWC categories, (ii) more positive\nsentiment laden words, (iii) have better language constructs and are (iv) more\nreadable.\n",
"title": "What Propels Celebrity Follower Counts? Language Use or Social Connectivity"
}
| null | null | null | null | true | null |
5130
| null |
Default
| null | null |
null |
{
"abstract": " We show that monochromatic Finsler metrics, i.e., Finsler metrics such that\neach two tangent spaces are isomorphic as normed spaces, are generalized\nBerwald metrics, i.e., there exists an affine connection, possibly with\ntorsion, that preserves the Finsler function\n",
"title": "Monochromatic metrics are generalized Berwald"
}
| null | null | null | null | true | null |
5131
| null |
Default
| null | null |
null |
{
"abstract": " Transcriptional repressor CTCF is an important regulator of chromatin 3D\nstructure, facilitating the formation of topologically associating domains\n(TADs). However, its direct effects on gene regulation is less well understood.\nHere, we utilize previously published ChIP-seq and RNA-seq data to investigate\nthe effects of CTCF on alternative splicing of genes with CTCF sites. We\ncompared the amount of RNA-seq signals in exons upstream and downstream of\nbinding sites following auxin-induced degradation of CTCF in mouse embryonic\nstem cells. We found that changes in gene expression following CTCF depletion\nwere significant, with a general increase in the presence of upstream exons. We\ninfer that a possible mechanism by which CTCF binding contributes to\nalternative splicing is by causing pauses in the transcription mechanism during\nwhich splicing elements are able to concurrently act on upstream exons already\ntranscribed into RNA.\n",
"title": "CTCF Degradation Causes Increased Usage of Upstream Exons in Mouse Embryonic Stem Cells"
}
| null | null | null | null | true | null |
5132
| null |
Default
| null | null |
null |
{
"abstract": " Many of the multi-planet systems discovered to date have been notable for\ntheir compactness, with neighbouring planets closer together than any in the\nSolar System. Interestingly, planet-hosting stars have a wide range of ages,\nsuggesting that such compact systems can survive for extended periods of time.\nWe have used numerical simulations to investigate how quickly systems go\nunstable in relation to the spacing between planets, focusing on hypothetical\nsystems of Earth-mass planets on evenly-spaced orbits (in mutual Hill radii).\nIn general, the further apart the planets are initially, the longer it takes\nfor a pair of planets to undergo a close encounter. We recover the results of\nprevious studies, showing a linear trend in the initial planet spacing between\n3 and 8 mutual Hill radii and the logarithm of the stability time.\nInvestigating thousands of simulations with spacings up to 13 mutual Hill radii\nreveals distinct modulations superimposed on this relationship in the vicinity\nof first and second-order mean motion resonances of adjacent and next-adjacent\nplanets. We discuss the impact of this structure and the implications on the\nstability of compact multi-planet systems. Applying the outcomes of our\nsimulations, we show that isolated systems of up to five Earth-mass planets can\nfit in the habitable zone of a Sun-like star without close encounters for at\nleast $10^9$ orbits.\n",
"title": "The stability of tightly-packed, evenly-spaced systems of Earth-mass planets orbiting a Sun-like star"
}
| null | null | null | null | true | null |
5133
| null |
Default
| null | null |
null |
{
"abstract": " We present GPUQT, a quantum transport code fully implemented on graphics\nprocessing units. Using this code, one can obtain intrinsic electronic\ntransport properties of large systems described by a real-space tight-binding\nHamiltonian together with one or more types of disorder. The DC Kubo\nconductivity is represented as a time integral of the velocity auto-correlation\nor a time derivative of the mean square displacement. Linear scaling (with\nrespect to the total number of orbitals in the system) computation time and\nmemory usage are achieved by using various numerical techniques, including\nsparse matrix-vector multiplication, random phase approximation of trace,\nChebyshev expansion of quantum evolution operator, and kernel polynomial method\nfor quantum resolution operator. We describe the inputs and outputs of GPUQT\nand give two examples to demonstrate its usage, paying attention to the\ninterpretations of the results.\n",
"title": "GPUQT: An efficient linear-scaling quantum transport code fully implemented on graphics processing units"
}
| null | null | null | null | true | null |
5134
| null |
Default
| null | null |
null |
{
"abstract": " This paper proposes and analyzes a new full-duplex (FD) cooperative cognitive\nradio network with wireless energy harvesting (EH). We consider that the\nsecondary receiver is equipped with a FD radio and acts as a FD hybrid access\npoint (HAP), which aims to collect information from its associated EH secondary\ntransmitter (ST) and relay the signals. The ST is assumed to be equipped with\nan EH unit and a rechargeable battery such that it can harvest and accumulate\nenergy from radio frequency (RF) signals transmitted by the primary transmitter\n(PT) and the HAP. We develop a novel cooperative spectrum sharing (CSS)\nprotocol for the considered system. In the proposed protocol, thanks to its FD\ncapability, the HAP can receive the PT's signals and transmit energy-bearing\nsignals to charge the ST simultaneously, or forward the PT's signals and\nreceive the ST's signals at the same time. We derive analytical expressions for\nthe achievable throughput of both primary and secondary links by characterizing\nthe dynamic charging/discharging behaviors of the ST battery as a finite-state\nMarkov chain. We present numerical results to validate our theoretical analysis\nand demonstrate the merits of the proposed protocol over its non-cooperative\ncounterpart.\n",
"title": "Full-Duplex Cooperative Cognitive Radio Networks with Wireless Energy Harvesting"
}
| null | null |
[
"Computer Science"
] | null | true | null |
5135
| null |
Validated
| null | null |
null |
{
"abstract": " In this study, a method to construct a full-colour volumetric display is\npresented using a commercially available inkjet printer. Photoreactive\nluminescence materials are minutely and automatically printed as the volume\nelements, and volumetric displays are constructed with high resolution using\neasy-to-fabricate means that exploit inkjet printing technologies. The results\nexperimentally demonstrate the first prototype of an inkjet printing-based\nvolumetric display composed of multiple layers of transparent films that yield\na full-colour three-dimensional (3D) image. Moreover, we propose a design\nalgorithm with 3D structures that provide multiple different 2D full-colour\npatterns when viewed from different directions and experimentally demonstrates\nprototypes. It is considered that these types of 3D volumetric structures and\ntheir fabrication methods based on widely deployed existing printing\ntechnologies can be utilised as novel information display devices and systems,\nincluding digital signage, media art, entertainment and security.\n",
"title": "Inkjet printing-based volumetric display projecting multiple full-colour 2D patterns"
}
| null | null | null | null | true | null |
5136
| null |
Default
| null | null |
null |
{
"abstract": " Strong gravitational lensing gives access to the total mass distribution of\ngalaxies. It can unveil a great deal of information about the lenses dark\nmatter content when combined with the study of the lenses light profile.\nHowever, gravitational lensing galaxies, by definition, appear surrounded by\npoint-like and diffuse lensed signal that is irrelevant to the lens flux.\nTherefore, the observer is most often restricted to studying the innermost\nportions of the galaxy, where classical fitting methods show some\ninstabilities. We aim at subtracting that lensed signal and at characterising\nsome lenses light profile by computing their shape parameters. Our objective is\nto evaluate the total integrated flux in an aperture the size of the Einstein\nring in order to obtain a robust estimate of the quantity of ordinary matter in\neach system. We are expanding the work we started in a previous paper that\nconsisted in subtracting point-like lensed images and in independently\nmeasuring each shape parameter. We improve it by designing a subtraction of the\ndiffuse lensed signal, based only on one simple hypothesis of symmetry. This\nextra step improves our study of the shape parameters and we refine it even\nmore by upgrading our half-light radius measurement. We also calculate the\nimpact of our specific image processing on the error bars. The diffuse lensed\nsignal subtraction makes it possible to study a larger portion of relevant\ngalactic flux, as the radius of the fitting region increases by on average\n17\\%. We retrieve new half-light radii values that are on average 11\\% smaller\nthan in our previous work, although the uncertainties overlap in most cases.\nThis shows that not taking the diffuse lensed signal into account may lead to a\nsignificant overestimate of the half-light radius. We are also able to measure\nthe flux within the Einstein radius and to compute secure error bars to all of\nour results.\n",
"title": "Analysis of luminosity distributions of strong lensing galaxies: subtraction of diffuse lensed signal"
}
| null | null | null | null | true | null |
5137
| null |
Default
| null | null |
null |
{
"abstract": " We introduce a new framework for estimating the support size of an unknown\ndistribution which improves upon known approximation-based techniques. Our main\ncontributions include describing a rigorous new weighted Chebyshev polynomial\napproximation method and introducing regularization terms into the problem\nformulation that provably improve the performance of state-of-the-art\napproximation-based approaches. In particular, we present both theoretical and\ncomputer simulation results that illustrate the utility and performance\nimprovements of our method. The theoretical analysis relies on jointly\noptimizing the bias and variance components of the risk, and combining new\nweighted minmax polynomial approximation techniques with discretized\nsemi-infinite programming solvers. Such a setting allows for casting the\nestimation problem as a linear program (LP) with a small number of variables\nand constraints that may be solved as efficiently as the original Chebyshev\napproximation-based problem. The described approach also applies to the support\ncoverage and entropy estimation problems. Our newly developed technique is\ntested on synthetic data and used to estimate the number of bacterial species\nin the human gut. On synthetic datasets, we observed up to five-fold\nimprovements in the value of the worst-case risk. For the bioinformatics\napplication, metagenomic data from the NIH Human Gut and the American Gut\nMicrobiome was combined and processed to obtain lists of bacterial taxonomies.\nThese were subsequently used to compute the bacterial species histograms and\nestimate the number of bacterial species in the human gut to roughly 2350, with\nthe species being represented by trillions of cells.\n",
"title": "Support Estimation via Regularized and Weighted Chebyshev Approximations"
}
| null | null | null | null | true | null |
5138
| null |
Default
| null | null |
null |
{
"abstract": " Two-photon superbunching of pseudothermal light is observed with single-mode\ncontinuous-wave laser light in a linear optical system. By adding more\ntwo-photon paths via three rotating ground glasses,g(2)(0) = 7.10 is\nexperimentally observed. The second-order temporal coherence function of\nsuperbunching pseudothermal light is theoretically and experimentally studied\nin detail. It is predicted that the degree of coherence of light can be\nincreased dramatically by adding more multi-photon paths. For instance, the\ndegree of the second- and third-order coherence of the superbunching\npseudothermal light with five rotating ground glasses can reach 32 and 7776,\nrespectively. The results are helpful to understand the physics of\nsuperbunching and to improve the visibility of thermal light ghost imaging.\n",
"title": "Two-photon superbunching of pseudothermal light in a Hanbury Brown-Twiss interferometer"
}
| null | null |
[
"Physics"
] | null | true | null |
5139
| null |
Validated
| null | null |
null |
{
"abstract": " The aim of this paper is to investigate the non-relativistic limit of\nintegrable quantum field theories with fermionic fields, such as the O(N)\nGross-Neveu model, the supersymmetric Sinh-Gordon and non-linear sigma models.\nThe non-relativistic limit of these theories is implemented by a double scaling\nlimit which consists of sending the speed of light c to infinity and rescaling\nat the same time the relevant coupling constant of the model in such a way to\nhave finite energy excitations. For the general purpose of mapping the space of\ncontinuous non-relativistic integrable models, this paper completes and\nintegrates the analysis done in Ref.[1] on the non-relativistic limit of purely\nbosonic theories.\n",
"title": "Non Relativistic Limit of Integrable QFT with fermionic excitations"
}
| null | null |
[
"Physics"
] | null | true | null |
5140
| null |
Validated
| null | null |
null |
{
"abstract": " The Benson-Solomon systems comprise the only known family of simple saturated\nfusion systems at the prime two that do not arise as the fusion system of any\nfinite group. We determine the automorphism groups and the possible almost\nsimple extensions of these systems and of their centric linking systems.\n",
"title": "Extensions of the Benson-Solomon fusion systems"
}
| null | null | null | null | true | null |
5141
| null |
Default
| null | null |
null |
{
"abstract": " We present a new paradigm for the simulation of arrays of Imaging Atmospheric\nCherenkov Telescopes (IACTs) which overcomes limitations of current approaches.\nUp to now, all major IACT experiments rely on the same Monte-Carlo simulation\nstrategy, using predefined observation and instrument settings. Simulations\nwith varying parameters are generated to provide better estimates of the\nInstrument Response Functions (IRFs) of different observations. However, a\nlarge fraction of the simulation configuration remains preserved, leading to\ncomplete negligence of all related influences. Additionally, the simulation\nscheme relies on interpolations between different array configurations, which\nare never fully reproducing the actual configuration for a given observation.\nInterpolations are usually performed on zenith angles, off-axis angles, array\nmultiplicity, and the optical response of the instrument. With the advent of\nhybrid systems consisting of a large number of IACTs with different sizes,\ntypes, and camera configurations, the complexity of the interpolation and the\nsize of the phase space becomes increasingly prohibitive. Going beyond the\nexisting approaches, we introduce a new simulation and analysis concept which\ntakes into account the actual observation conditions as well as individual\ntelescope configurations of each observation run of a given data set. These\nrun-wise simulations (RWS) thus exhibit considerably reduced systematic\nuncertainties compared to the existing approach, and are also more\ncomputationally efficient and simple. The RWS framework has been implemented in\nthe H.E.S.S. software and tested, and is already being exploited in science\nanalysis.\n",
"title": "Run-Wise Simulations for Imaging Atmospheric Cherenkov Telescope Arrays"
}
| null | null | null | null | true | null |
5142
| null |
Default
| null | null |
null |
{
"abstract": " In a poisoning attack against a learning algorithm, an adversary tampers with\na fraction of the training data $T$ with the goal of increasing the\nclassification error of the constructed hypothesis/model over the final test\ndistribution. In the distributed setting, $T$ might be gathered gradually from\n$m$ data providers $P_1,\\dots,P_m$ who generate and submit their shares of $T$\nin an online way.\nIn this work, we initiate a formal study of $(k,p)$-poisoning attacks in\nwhich an adversary controls $k\\in[n]$ of the parties, and even for each\ncorrupted party $P_i$, the adversary submits some poisoned data $T'_i$ on\nbehalf of $P_i$ that is still \"$(1-p)$-close\" to the correct data $T_i$ (e.g.,\n$1-p$ fraction of $T'_i$ is still honestly generated). For $k=m$, this model\nbecomes the traditional notion of poisoning, and for $p=1$ it coincides with\nthe standard notion of corruption in multi-party computation.\nWe prove that if there is an initial constant error for the generated\nhypothesis $h$, there is always a $(k,p)$-poisoning attacker who can decrease\nthe confidence of $h$ (to have a small error), or alternatively increase the\nerror of $h$, by $\\Omega(p \\cdot k/m)$. Our attacks can be implemented in\npolynomial time given samples from the correct data, and they use no wrong\nlabels if the original distributions are not noisy.\nAt a technical level, we prove a general lemma about biasing bounded\nfunctions $f(x_1,\\dots,x_n)\\in[0,1]$ through an attack model in which each\nblock $x_i$ might be controlled by an adversary with marginal probability $p$\nin an online way. When the probabilities are independent, this coincides with\nthe model of $p$-tampering attacks, thus we call our model generalized\n$p$-tampering. We prove the power of such attacks by incorporating ideas from\nthe context of coin-flipping attacks into the $p$-tampering model and\ngeneralize the results in both of these areas.\n",
"title": "Multi-party Poisoning through Generalized $p$-Tampering"
}
| null | null | null | null | true | null |
5143
| null |
Default
| null | null |
null |
{
"abstract": " We consider Jacobi matrices with eventually increasing sequences of diagonal\nand off-diagonal Jacobi parameters. We describe the asymptotic behavior of the\nsubordinate solution at the top of the essential spectrum, and the asymptotic\nbehavior of the spectral density at the top of the essential spectrum.\nIn particular, allowing on both diagonal and off-diagonal Jacobi parameters\nperturbations of the free case of the form $- \\sum_{j=1}^J c_j n^{-\\tau_j} +\no(n^{-\\tau_1-1})$ with $0 < \\tau_1 < \\tau_2 < \\dots < \\tau_J$ and $c_1>0$, we\nfind the asymptotic behavior of the $\\log$ of spectral density to order\n$O(\\log(2-x))$ as $x$ approaches $2$.\nApart from its intrinsic interest, the above results also allow us to\ndescribe the asymptotics of the spectral density for orthogonal polynomials on\nthe unit circle with real-valued Verblunsky coefficients of the same form.\n",
"title": "Spectral edge behavior for eventually monotone Jacobi and Verblunsky coefficients"
}
| null | null |
[
"Mathematics"
] | null | true | null |
5144
| null |
Validated
| null | null |
null |
{
"abstract": " Web video is often used as a source of data in various fields of study. While\nspecialized subsets of web video, mainly earmarked for dedicated purposes, are\noften analyzed in detail, there is little information available about the\nproperties of web video as a whole. In this paper we present insights gained\nfrom the analysis of the metadata associated with more than 120 million videos\nharvested from two popular web video platforms, vimeo and YouTube, in 2016 and\ncompare their properties with the ones found in commonly used video\ncollections. This comparison has revealed that existing collections do not (or\nno longer) properly reflect the properties of web video \"in the wild\".\n",
"title": "Web Video in Numbers - An Analysis of Web-Video Metadata"
}
| null | null | null | null | true | null |
5145
| null |
Default
| null | null |
null |
{
"abstract": " Given $n$ symmetric Bernoulli variables, what can be said about their\ncorrelation matrix viewed as a vector? We show that the set of those vectors\n$R(\\mathcal{B}_n)$ is a polytope and identify its vertices. Those extreme\npoints correspond to correlation vectors associated to the discrete uniform\ndistributions on diagonals of the cube $[0,1]^n$. We also show that the\npolytope is affinely isomorphic to a well-known cut polytope ${\\rm CUT}(n)$\nwhich is defined as a convex hull of the cut vectors in a complete graph with\nvertex set $\\{1,\\ldots,n\\}$. The isomorphism is obtained explicitly as\n$R(\\mathcal{B}_n)= {\\mathbf{1}}-2~{\\rm CUT}(n)$. As a corollary of this work,\nit is straightforward using linear programming to determine if a particular\ncorrelation matrix is realizable or not. Furthermore, a sampling method for\nmultivariate symmetric Bernoullis with given correlation is obtained. In some\ncases the method can also be used for general, not exclusively Bernoulli,\nmarginals.\n",
"title": "Bernoulli Correlations and Cut Polytopes"
}
| null | null | null | null | true | null |
5146
| null |
Default
| null | null |
null |
{
"abstract": " We show that the tensor product $A\\otimes B$ over $\\mathbb{C}$ of two $C^*\n$-algebras satisfying the \\textit{NCDL} conditions has again the same property.\nWe use this result to describe the $C^* $-algebra of the Heisenberg motion\ngroups $G_n = \\mathbb{T}^n \\ltimes \\mathbb{H}_n$ as algebra of operator fields\ndefined over the spectrum of $G_n $.\n",
"title": "Tensor products of NCDL-C*-algebras and the C*-algebra of the Heisenberg motion groups"
}
| null | null | null | null | true | null |
5147
| null |
Default
| null | null |
null |
{
"abstract": " As the intermediate level task connecting image captioning and object\ndetection, visual relationship detection started to catch researchers'\nattention because of its descriptive power and clear structure. It detects the\nobjects and captures their pair-wise interactions with a\nsubject-predicate-object triplet, e.g. person-ride-horse. In this paper, each\nvisual relationship is considered as a phrase with three components. We\nformulate the visual relationship detection as three inter-connected\nrecognition problems and propose a Visual Phrase guided Convolutional Neural\nNetwork (ViP-CNN) to address them simultaneously. In ViP-CNN, we present a\nPhrase-guided Message Passing Structure (PMPS) to establish the connection\namong relationship components and help the model consider the three problems\njointly. Corresponding non-maximum suppression method and model training\nstrategy are also proposed. Experimental results show that our ViP-CNN\noutperforms the state-of-art method both in speed and accuracy. We further\npretrain ViP-CNN on our cleansed Visual Genome Relationship dataset, which is\nfound to perform better than the pretraining on the ImageNet for this task.\n",
"title": "ViP-CNN: Visual Phrase Guided Convolutional Neural Network"
}
| null | null |
[
"Computer Science"
] | null | true | null |
5148
| null |
Validated
| null | null |
null |
{
"abstract": " We propose the notion of Haantjes algebra, which consists of an assignment of\na family of fields of operators over a differentiable manifold, with vanishing\nHaantjes torsion and satisfying suitable compatibility conditions among each\nothers. Haantjes algebras naturally generalize several known interesting\ngeometric structures, arising in Riemannian geometry and in the theory of\nintegrable systems. At the same time, they play a crucial role in the theory of\ndiagonalization of operators on differentiable manifolds.\nWhenever the elements of an Haantjes algebra are semisimple and commute, we\nshall prove that there exists a set of local coordinates where all operators\ncan be diagonalized simultaneously. Moreover, in the non-semisimple case, they\nacquire simultaneously a block-diagonal form.\n",
"title": "Haantjes Algebras and Diagonalization"
}
| null | null | null | null | true | null |
5149
| null |
Default
| null | null |
null |
{
"abstract": " This paper considers a multipair amplify-and-forward massive MIMO relaying\nsystem with one-bit ADCs and one-bit DACs at the relay. The channel state\ninformation is estimated via pilot training, and then utilized by the relay to\nperform simple maximum-ratio combining/maximum-ratio transmission processing.\nLeveraging on the Bussgang decomposition, an exact achievable rate is derived\nfor the system with correlated quantization noise. Based on this, a closed-form\nasymptotic approximation for the achievable rate is presented, thereby enabling\nefficient evaluation of the impact of key parameters on the system performance.\nFurthermore, power scaling laws are characterized to study the potential energy\nefficiency associated with deploying massive one-bit antenna arrays at the\nrelay. In addition, a power allocation strategy is designed to compensate for\nthe rate degradation caused by the coarse quantization. Our results suggest\nthat the quality of the channel estimates depends on the specific orthogonal\npilot sequences that are used, contrary to unquantized systems where any set of\northogonal pilot sequences gives the same result. Moreover, the sum rate gap\nbetween the double-quantized relay system and an ideal non-quantized system is\na moderate factor of $4/\\pi^2$ in the low power regime.\n",
"title": "Multipair Massive MIMO Relaying Systems with One-Bit ADCs and DACs"
}
| null | null | null | null | true | null |
5150
| null |
Default
| null | null |
null |
{
"abstract": " Exoplanet research is carried out at the limits of the capabilities of\ncurrent telescopes and instruments. The studied signals are weak, and often\nembedded in complex systematics from instrumental, telluric, and astrophysical\nsources. Combining repeated observations of periodic events, simultaneous\nobservations with multiple telescopes, different observation techniques, and\nexisting information from theory and prior research can help to disentangle the\nsystematics from the planetary signals, and offers synergistic advantages over\nanalysing observations separately. Bayesian inference provides a\nself-consistent statistical framework that addresses both the necessity for\ncomplex systematics models, and the need to combine prior information and\nheterogeneous observations. This chapter offers a brief introduction to\nBayesian inference in the context of exoplanet research, with focus on time\nseries analysis, and finishes with an overview of a set of freely available\nprogramming libraries.\n",
"title": "Bayesian Methods for Exoplanet Science"
}
| null | null | null | null | true | null |
5151
| null |
Default
| null | null |
null |
{
"abstract": " Understanding the influence of features in machine learning is crucial to\ninterpreting models and selecting the best features for classification. In this\nwork we propose the use of principles from coalitional game theory to reason\nabout importance of features. In particular, we propose the use of the Banzhaf\npower index as a measure of influence of features on the outcome of a\nclassifier. We show that features having Banzhaf power index of zero can be\nlosslessly pruned without damage to classifier accuracy. Computing the power\nindices does not require having access to data samples. However, if samples are\navailable, the indices can be empirically estimated. We compute Banzhaf power\nindices for a neural network classifier on real-life data, and compare the\nresults with gradient-based feature saliency, and coefficients of a logistic\nregression model with $L_1$ regularization.\n",
"title": "Feature importance scores and lossless feature pruning using Banzhaf power indices"
}
| null | null | null | null | true | null |
5152
| null |
Default
| null | null |
null |
{
"abstract": " Wikipedia is the largest existing knowledge repository that is growing on a\ngenuine crowdsourcing support. While the English Wikipedia is the most\nextensive and the most researched one with over five million articles,\ncomparatively little is known about the behavior and growth of the remaining\n283 smaller Wikipedias, the smallest of which, Afar, has only one article. Here\nwe use a subset of this data, consisting of 14962 different articles, each of\nwhich exists in 26 different languages, from Arabic to Ukrainian. We study the\ngrowth of Wikipedias in these languages over a time span of 15 years. We show\nthat, while an average article follows a random path from one language to\nanother, there exist six well-defined clusters of Wikipedias that share common\ngrowth patterns. The make-up of these clusters is remarkably robust against the\nmethod used for their determination, as we verify via four different clustering\nmethods. Interestingly, the identified Wikipedia clusters have little\ncorrelation with language families and groups. Rather, the growth of Wikipedia\nacross different languages is governed by different factors, ranging from\nsimilarities in culture to information literacy.\n",
"title": "Robust clustering of languages across Wikipedia growth"
}
| null | null | null | null | true | null |
5153
| null |
Default
| null | null |
null |
{
"abstract": " Adopting two independent approaches (a) Lorentz-invariance of physical laws\nand (b) local phase invariance of quantum field theory applied to the Dirac\nLagrangian for massive electrically neutral Dirac particles, we rediscovered\nthe fundamental field equations of Heaviside Gravity (HG) of 1893 and\nMaxwellian Gravity (MG), which look different from each other due to a sign\ndifference in some terms of their respective field equations. However, they are\nshown to represent two mathematical representations of a single physical theory\nof vector gravity that we name here as Heaviside-Maxwellian Gravity (HMG), in\nwhich the speed of gravitational waves in vacuum is uniquely found to be equal\nto the speed of light in vacuum. We also corrected a sign error in Heaviside's\nspeculative gravitational analogue of the Lorentz force law. This spin-1 HMG is\nshown to produce attractive force between like masses under static condition,\ncontrary to the prevalent view of field theorists. Galileo's law of\nuniversality of free fall is a consequence of HMG, without any initial\nassumption of the equality of gravitational mass with velocity-dependent mass.\nWe also note a new set of Lorentz-Maxwell's equations having the same physical\neffects as the standard set - a byproduct of our present study.\n",
"title": "Attractive Heaviside-Maxwellian (Vector) Gravity from Special Relativity and Quantum Field Theory"
}
| null | null | null | null | true | null |
5154
| null |
Default
| null | null |
null |
{
"abstract": " We study the effect of critical pairing fluctuations on the electronic\nproperties in the normal state of a clean superconductor in three dimensions.\nUsing a functional renormalization group approach to take the non-Gaussian\nnature of critical fluctuations into account, we show microscopically that in\nthe BCS regime, where the inverse coherence length is much smaller than the\nFermi wavevector, critical pairing fluctuations give rise to a non-analytic\ncontribution to the quasi-particle damping of order $ T_c \\sqrt{Gi} \\ln ( 80 /\nGi )$, where the Ginzburg-Levanyuk number $Gi$ is a dimensionless measure for\nthe width of the critical region. As a consequence, there is a temperature\nwindow above $T_c$ where the quasiparticle damping due to critical pairing\nfluctuations can be larger than the usual $T^2$-Fermi liquid damping due to\nnon-critical scattering processes. On the other hand, in the strong coupling\nregime where $Gi$ is of order unity, we find that the quasiparticle damping due\nto critical pairing fluctuations is proportional to the temperature. Moreover,\nwe show that in the vicinity of the critical temperature $T_c$ the electronic\ndensity of states exhibits a fluctuation-induced pseudogap. We also use\nfunctional renormalization group methods to derive and classify various types\nof processes induced by the pairing interaction in Fermi systems close to the\nsuperconducting instability.\n",
"title": "Critical pairing fluctuations in the normal state of a superconductor: pseudogap and quasi-particle damping"
}
| null | null | null | null | true | null |
5155
| null |
Default
| null | null |
null |
{
"abstract": " In this paper, we propose a low-rank coordinate descent approach to\nstructured semidefinite programming with diagonal constraints. The approach,\nwhich we call the Mixing method, is extremely simple to implement, has no free\nparameters, and typically attains an order of magnitude or better improvement\nin optimization performance over the current state of the art. We show that the\nmethod is strictly decreasing, converges to a critical point, and further that\nfor sufficient rank all non-optimal critical points are unstable. Moreover, we\nprove that with a step size, the Mixing method converges to the global optimum\nof the semidefinite program almost surely in a locally linear rate under random\ninitialization. This is the first low-rank semidefinite programming method that\nhas been shown to achieve a global optimum on the spherical manifold without\nassumption. We apply our algorithm to two related domains: solving the maximum\ncut semidefinite relaxation, and solving a maximum satisfiability relaxation\n(we also briefly consider additional applications such as learning word\nembeddings). In all settings, we demonstrate substantial improvement over the\nexisting state of the art along various dimensions, and in total, this work\nexpands the scope and scale of problems that can be solved using semidefinite\nprogramming methods.\n",
"title": "The Mixing method: low-rank coordinate descent for semidefinite programming with diagonal constraints"
}
| null | null | null | null | true | null |
5156
| null |
Default
| null | null |
null |
{
"abstract": " The Fisher information metric is an important foundation of information\ngeometry, wherein it allows us to approximate the local geometry of a\nprobability distribution. Recurrent neural networks such as the\nSequence-to-Sequence (Seq2Seq) networks that have lately been used to yield\nstate-of-the-art performance on speech translation or image captioning have so\nfar ignored the geometry of the latent embedding, that they iteratively learn.\nWe propose the information geometric Seq2Seq (GeoSeq2Seq) network which\nabridges the gap between deep recurrent neural networks and information\ngeometry. Specifically, the latent embedding offered by a recurrent network is\nencoded as a Fisher kernel of a parametric Gaussian Mixture Model, a formalism\ncommon in computer vision. We utilise such a network to predict the shortest\nroutes between two nodes of a graph by learning the adjacency matrix using the\nGeoSeq2Seq formalism; our results show that for such a problem the\nprobabilistic representation of the latent embedding supersedes the\nnon-probabilistic embedding by 10-15\\%.\n",
"title": "GeoSeq2Seq: Information Geometric Sequence-to-Sequence Networks"
}
| null | null | null | null | true | null |
5157
| null |
Default
| null | null |
null |
{
"abstract": " Given a network, the statistical ensemble of its graph-Voronoi diagrams with\nrandomly chosen cell centers exhibits properties convertible into information\non the network's large scale structures. We define a node-pair level measure\ncalled {\\it Voronoi cohesion} which describes the probability for sharing the\nsame Voronoi cell, when randomly choosing $g$ centers in the network. This\nmeasure provides information based on the global context (the network in its\nentirety) a type of information that is not carried by other similarity\nmeasures. We explore the mathematical background of this phenomenon and several\nof its potential applications. A special focus is laid on the possibilities and\nlimitations pertaining to the exploitation of the phenomenon for community\ndetection purposes.\n",
"title": "Stochastic graph Voronoi tessellation reveals community structure"
}
| null | null | null | null | true | null |
5158
| null |
Default
| null | null |
null |
{
"abstract": " The statistics of the smallest eigenvalue of Wishart-Laguerre ensemble is\nimportant from several perspectives. The smallest eigenvalue density is\ntypically expressible in terms of determinants or Pfaffians. These results are\nof utmost significance in understanding the spectral behavior of\nWishart-Laguerre ensembles and, among other things, unveil the underlying\nuniversality aspects in the asymptotic limits. However, obtaining exact and\nexplicit expressions by expanding determinants or Pfaffians becomes impractical\nif large dimension matrices are involved. For the real matrices ($\\beta=1$)\nEdelman has provided an efficient recurrence scheme to work out exact and\nexplicit results for the smallest eigenvalue density which does not involve\ndeterminants or matrices. Very recently, an analogous recurrence scheme has\nbeen obtained for the complex matrices ($\\beta=2$). In the present work we\nextend this to $\\beta$-Wishart-Laguerre ensembles for the case when exponent\n$\\alpha$ in the associated Laguerre weight function, $\\lambda^\\alpha\ne^{-\\beta\\lambda/2}$, is a non-negative integer, while $\\beta$ is positive\nreal. This also gives access to the smallest eigenvalue density of fixed trace\n$\\beta$-Wishart-Laguerre ensemble, as well as moments for both cases. Moreover,\ncomparison with earlier results for the smallest eigenvalue density in terms of\ncertain hypergeometric function of matrix argument results in an effective way\nof evaluating these explicitly. Exact evaluations for large values of $n$ (the\nmatrix dimension) and $\\alpha$ also enable us to compare with Tracy-Widom\ndensity and large deviation results of Katzav and Castillo. We also use our\nresult to obtain the density of the largest of the proper delay times which are\neigenvalues of the Wigner-Smith matrix and are relevant to the problem of\nquantum chaotic scattering.\n",
"title": "Recursion for the smallest eigenvalue density of $β$-Wishart-Laguerre ensemble"
}
| null | null | null | null | true | null |
5159
| null |
Default
| null | null |
null |
{
"abstract": " Advances in image processing and computer vision in the latest years have\nbrought about the use of visual features in artwork recommendation. Recent\nworks have shown that visual features obtained from pre-trained deep neural\nnetworks (DNNs) perform very well for recommending digital art. Other recent\nworks have shown that explicit visual features (EVF) based on attractiveness\ncan perform well in preference prediction tasks, but no previous work has\ncompared DNN features versus specific attractiveness-based visual features\n(e.g. brightness, texture) in terms of recommendation performance. In this\nwork, we study and compare the performance of DNN and EVF features for the\npurpose of physical artwork recommendation using transactional data from\nUGallery, an online store of physical paintings. In addition, we perform an\nexploratory analysis to understand if DNN embedded features have some relation\nwith certain EVF. Our results show that DNN features outperform EVF, that\ncertain EVF features are more suited for physical artwork recommendation and,\nfinally, we show evidence that certain neurons in the DNN might be partially\nencoding visual features such as brightness, providing an opportunity for\nexplaining recommendations based on visual neural models.\n",
"title": "Comparing Neural and Attractiveness-based Visual Features for Artwork Recommendation"
}
| null | null |
[
"Computer Science"
] | null | true | null |
5160
| null |
Validated
| null | null |
null |
{
"abstract": " We propose DeepMapping, a novel registration framework using deep neural\nnetworks (DNNs) as auxiliary functions to align multiple point clouds from\nscratch to a globally consistent frame. We use DNNs to model the highly\nnon-convex mapping process that traditionally involves hand-crafted data\nassociation, sensor pose initialization, and global refinement. Our key novelty\nis that properly defining unsupervised losses to \"train\" these DNNs through\nback-propagation is equivalent to solving the underlying registration problem,\nyet enables fewer dependencies on good initialization as required by ICP. Our\nframework contains two DNNs: a localization network that estimates the poses\nfor input point clouds, and a map network that models the scene structure by\nestimating the occupancy status of global coordinates. This allows us to\nconvert the registration problem to a binary occupancy classification, which\ncan be solved efficiently using gradient-based optimization. We further show\nthat DeepMapping can be readily extended to address the problem of Lidar SLAM\nby imposing geometric constraints between consecutive point clouds. Experiments\nare conducted on both simulated and real datasets. Qualitative and quantitative\ncomparisons demonstrate that DeepMapping often enables more robust and accurate\nglobal registration of multiple point clouds than existing techniques. Our code\nis available at this http URL.\n",
"title": "DeepMapping: Unsupervised Map Estimation From Multiple Point Clouds"
}
| null | null | null | null | true | null |
5161
| null |
Default
| null | null |
null |
{
"abstract": " The first systematic comparison between Swarm-C accelerometer-derived\nthermospheric density and both empirical and physics-based model results using\nmultiple model performance metrics is presented. This comparison is performed\nat the satellite's high temporal 10-s resolution, which provides a meaningful\nevaluation of the models' fidelity for orbit prediction and other space weather\nforecasting applications. The comparison against the physical model is\ninfluenced by the specification of the lower atmospheric forcing, the\nhigh-latitude ionospheric plasma convection, and solar activity. Some insights\ninto the model response to thermosphere-driving mechanisms are obtained through\na machine learning exercise. The results of this analysis show that the\nshort-timescale variations observed by Swarm-C during periods of high solar and\ngeomagnetic activity were better captured by the physics-based model than the\nempirical models. It is concluded that Swarm-C data agree well with the\nclimatologies inherent within the models and are, therefore, a useful data set\nfor further model validation and scientific research.\n",
"title": "The First Comparison Between Swarm-C Accelerometer-Derived Thermospheric Densities and Physical and Empirical Model Estimates"
}
| null | null | null | null | true | null |
5162
| null |
Default
| null | null |
null |
{
"abstract": " We consider model-based clustering methods for continuous, correlated data\nthat account for external information available in the presence of mixed-type\nfixed covariates by proposing the MoEClust suite of models. These allow\ncovariates influence the component weights and/or component densities by\nmodelling the parameters of the mixture as functions of the covariates. A\nfamiliar range of constrained eigen-decomposition parameterisations of the\ncomponent covariance matrices are also accommodated. This paper thus addresses\nthe equivalent aims of including covariates in Gaussian Parsimonious Clustering\nModels and incorporating parsimonious covariance structures into the Gaussian\nmixture of experts framework. The MoEClust models demonstrate significant\nimprovement from both perspectives in applications to univariate and\nmultivariate data sets.\n",
"title": "Gaussian Parsimonious Clustering Models with Covariates"
}
| null | null |
[
"Statistics"
] | null | true | null |
5163
| null |
Validated
| null | null |
null |
{
"abstract": " We provide requirements on effectively enumerable topological spaces which\nguarantee that the Rice-Shapiro theorem holds for the computable elements of\nthese spaces. We show that the relaxation of these requirements leads to the\nclasses of effectively enumerable topological spaces where the Rice-Shapiro\ntheorem does not hold. We propose two constructions that generate effectively\nenumerable topological spaces with particular properties from wn--families and\ncomputable trees without computable infinite paths. Using them we propose\nexamples that give a flavor of this class.\n",
"title": "The Rice-Shapiro theorem in Computable Topology"
}
| null | null | null | null | true | null |
5164
| null |
Default
| null | null |
null |
{
"abstract": " This paper studies the problem of multivariate linear regression where a\nportion of the observations is grossly corrupted or is missing, and the\nmagnitudes and locations of such occurrences are unknown in priori. To deal\nwith this problem, we propose a new approach by explicitly consider the error\nsource as well as its sparseness nature. An interesting property of our\napproach lies in its ability of allowing individual regression output elements\nor tasks to possess their unique noise levels. Moreover, despite working with a\nnon-smooth optimization problem, our approach still guarantees to converge to\nits optimal solution. Experiments on synthetic data demonstrate the\ncompetitiveness of our approach compared with existing multivariate regression\nmodels. In addition, empirically our approach has been validated with very\npromising results on two exemplar real-world applications: The first concerns\nthe prediction of \\textit{Big-Five} personality based on user behaviors at\nsocial network sites (SNSs), while the second is 3D human hand pose estimation\nfrom depth images. The implementation of our approach and comparison methods as\nwell as the involved datasets are made publicly available in support of the\nopen-source and reproducible research initiatives.\n",
"title": "Multivariate Regression with Grossly Corrupted Observations: A Robust Approach and its Applications"
}
| null | null | null | null | true | null |
5165
| null |
Default
| null | null |
null |
{
"abstract": " Knowing where people live is a fundamental component of many decision making\nprocesses such as urban development, infectious disease containment, evacuation\nplanning, risk management, conservation planning, and more. While bottom-up,\nsurvey driven censuses can provide a comprehensive view into the population\nlandscape of a country, they are expensive to realize, are infrequently\nperformed, and only provide population counts over broad areas. Population\ndisaggregation techniques and population projection methods individually\naddress these shortcomings, but also have shortcomings of their own. To jointly\nanswer the questions of \"where do people live\" and \"how many people live\nthere,\" we propose a deep learning model for creating high-resolution\npopulation estimations from satellite imagery. Specifically, we train\nconvolutional neural networks to predict population in the USA at a\n$0.01^{\\circ} \\times 0.01^{\\circ}$ resolution grid from 1-year composite\nLandsat imagery. We validate these models in two ways: quantitatively, by\ncomparing our model's grid cell estimates aggregated at a county-level to\nseveral US Census county-level population projections, and qualitatively, by\ndirectly interpreting the model's predictions in terms of the satellite image\ninputs. We find that aggregating our model's estimates gives comparable results\nto the Census county-level population projections and that the predictions made\nby our model can be directly interpreted, which give it advantages over\ntraditional population disaggregation methods. In general, our model is an\nexample of how machine learning techniques can be an effective tool for\nextracting information from inherently unstructured, remotely sensed data to\nprovide effective solutions to social problems.\n",
"title": "A Deep Learning Approach for Population Estimation from Satellite Imagery"
}
| null | null | null | null | true | null |
5166
| null |
Default
| null | null |
null |
{
"abstract": " We measure trends in the diffusion of misinformation on Facebook and Twitter\nbetween January 2015 and July 2018. We focus on stories from 570 sites that\nhave been identified as producers of false stories. Interactions with these\nsites on both Facebook and Twitter rose steadily through the end of 2016.\nInteractions then fell sharply on Facebook while they continued to rise on\nTwitter, with the ratio of Facebook engagements to Twitter shares falling by\napproximately 60 percent. We see no similar pattern for other news, business,\nor culture sites, where interactions have been relatively stable over time and\nhave followed similar trends on the two platforms both before and after the\nelection.\n",
"title": "Trends in the Diffusion of Misinformation on Social Media"
}
| null | null | null | null | true | null |
5167
| null |
Default
| null | null |
null |
{
"abstract": " We describe the SemEval task of extracting keyphrases and relations between\nthem from scientific documents, which is crucial for understanding which\npublications describe which processes, tasks and materials. Although this was a\nnew task, we had a total of 26 submissions across 3 evaluation scenarios. We\nexpect the task and the findings reported in this paper to be relevant for\nresearchers working on understanding scientific content, as well as the broader\nknowledge base population and information extraction communities.\n",
"title": "SemEval 2017 Task 10: ScienceIE - Extracting Keyphrases and Relations from Scientific Publications"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
5168
| null |
Validated
| null | null |
null |
{
"abstract": " We consider a two-dimensional nonlinear Schrödinger equation with\nconcentrated nonlinearity. In both the focusing and defocusing case we prove\nlocal well-posedness, i.e., existence and uniqueness of the solution for short\ntimes, as well as energy and mass conservation. In addition, we prove that this\nimplies global existence in the defocusing case, irrespective of the power of\nthe nonlinearity, while in the focusing case blowing-up solutions may arise.\n",
"title": "Well-posedness of the Two-dimensional Nonlinear Schrödinger Equation with Concentrated Nonlinearity"
}
| null | null | null | null | true | null |
5169
| null |
Default
| null | null |
null |
{
"abstract": " In the late 1980s, Premet conjectured that the nilpotent variety of any\nfinite dimensional restricted Lie algebra over an algebraically closed field of\ncharacteristic $p>0$ is irreducible. This conjecture remains open, but it is\nknown to hold for a large class of simple restricted Lie algebras, e.g. for Lie\nalgebras of connected reductive algebraic groups, and for Cartan series $W, S$\nand $H$. In this paper, with the assumption that $p>3$, we confirm this\nconjecture for the minimal $p$-envelope $W(1;n)_p$ of the Zassenhaus algebra\n$W(1;n)$ for all $n\\geq 2$.\n",
"title": "The nilpotent variety of $W(1;n)_{p}$ is irreducible"
}
| null | null |
[
"Mathematics"
] | null | true | null |
5170
| null |
Validated
| null | null |
null |
{
"abstract": " We report on the existence and stability of freely moving solitons in a\nspatially inhomogeneous Bose- Einstein condensate with helicoidal spin-orbit\n(SO) coupling. In spite of the periodically varying parameters, the system\nallows for the existence of stable propagating solitons. Such states are found\nin the rotating frame, where the helicoidal SO coupling is reduced to a\nhomogeneous one. In the absence of the Zeeman splitting, the coupled\nGross-Pitaevskii equations describing localized states feature many properties\nof the integrable systems. In particular, four-parametric families of solitons\ncan be obtained in the exact form. Such solitons interact elastically. Zeeman\nsplitting still allows for the existence of two families of moving solitons,\nbut makes collisions of solitons inelastic.\n",
"title": "Solitons in Bose-Einstein Condensates with Helicoidal Spin-Orbit Coupling"
}
| null | null | null | null | true | null |
5171
| null |
Default
| null | null |
null |
{
"abstract": " In the following text we prove that for all finite $p\\geq0$ there exists a\ntopological graph $X$ such that $\\{p,p+1,p+2,\\ldots\\}\\cup\\{+\\infty\\}$ is the\ncollection of all possible heights for transformation groups with phase space\n$X$. Moreover for all topological graph $X$ with $p$ as height of\ntransformation group $(Homeo(X),X)$, $\\{p,p+1,p+2,\\ldots\\}\\cup\\{+\\infty\\}$\nagain is the collection of all possible heights for transformation groups with\nphase space $X$.\n",
"title": "Possible heights of graph transformation groups"
}
| null | null | null | null | true | null |
5172
| null |
Default
| null | null |
null |
{
"abstract": " Building machines that can understand text like humans is an AI-complete\nproblem. A great deal of research has already gone into this, with astounding\nresults, allowing everyday people to discuss with their telephones, or have\ntheir reading materials analysed and classified by computers. A prerequisite\nfor processing text semantics, common to the above examples, is having some\ncomputational representation of text as an abstract object. Operations on this\nrepresentation practically correspond to making semantic inferences, and by\nextension simulating understanding text. The complexity and granularity of\nsemantic processing that can be realised is constrained by the mathematical and\ncomputational robustness, expressiveness, and rigour of the tools used.\nThis dissertation contributes a series of such tools, diverse in their\nmathematical formulation, but common in their application to model semantic\ninferences when machines process text. These tools are principally expressed in\nnine distinct models that capture aspects of semantic dependence in highly\ninterpretable and non-complex ways. This dissertation further reflects on\npresent and future problems with the current research paradigm in this area,\nand makes recommendations on how to overcome them.\nThe amalgamation of the body of work presented in this dissertation advances\nthe complexity and granularity of semantic inferences that can be made\nautomatically by machines.\n",
"title": "Dependencies: Formalising Semantic Catenae for Information Retrieval"
}
| null | null | null | null | true | null |
5173
| null |
Default
| null | null |
null |
{
"abstract": " Remote sensing image processing is so important in geo-sciences. Images which\nare obtained by different types of sensors might initially be unrecognizable.\nTo make an acceptable visual perception in the images, some pre-processing\nsteps (for removing noises and etc) are preformed which they affect the\nanalysis of images. There are different types of processing according to the\ntypes of remote sensing images. The method that we are going to introduce in\nthis paper is to use virtual colors to colorize the gray-scale images of\nsatellite sensors. This approach helps us to have a better analysis on a sample\nsingle-band image which has been taken by Landsat-8 (OLI) sensor (as a\nmulti-band sensor with natural color bands, its images' natural color can be\ncompared to synthetic color by our approach). A good feature of this method is\nthe original image reversibility in order to keep the suitable resolution of\noutput images.\n",
"title": "A New Pseudo-color Technique Based on Intensity Information Protection for Passive Sensor Imagery"
}
| null | null |
[
"Computer Science"
] | null | true | null |
5174
| null |
Validated
| null | null |
null |
{
"abstract": " The independent control of two magnetic electrodes and spin-coherent\ntransport in magnetic tunnel junctions are strictly required for tunneling\nmagnetoresistance, while junctions with only one ferromagnetic electrode\nexhibit tunneling anisotropic magnetoresistance dependent on the anisotropic\ndensity of states with no room temperature performance so far. Here we report\nan alternative approach to obtaining tunneling anisotropic magnetoresistance in\nalfa-FeRh-based junctions driven by the magnetic phase transition of alfa-FeRh\nand resultantly large variation of the density of states in the vicinity of MgO\ntunneling barrier, referred to as phase transition tunneling anisotropic\nmagnetoresistance. The junctions with only one alfa-FeRh magnetic electrode\nshow a magnetoresistance ratio up to 20% at room temperature. Both the polarity\nand magnitude of the phase transition tunneling anisotropic magnetoresistance\ncan be modulated by interfacial engineering at the alfa-FeRh/MgO interface.\nBesides the fundamental significance, our finding might add a different\ndimension to magnetic random access memory and antiferromagnet spintronics.\n",
"title": "Tunneling anisotropic magnetoresistance driven by magnetic phase transition"
}
| null | null | null | null | true | null |
5175
| null |
Default
| null | null |
null |
{
"abstract": " We apply a reinforcement learning algorithm to show how smart particles can\nlearn approximately optimal strategies to navigate in complex flows. In this\npaper we consider microswimmers in a paradigmatic three-dimensional case given\nby a stationary superposition of two Arnold-Beltrami-Childress flows with\nchaotic advection along streamlines. In such a flow, we study the evolution of\npoint-like particles which can decide in which direction to swim, while keeping\nthe velocity amplitude constant. We show that it is sufficient to endow the\nswimmers with a very restricted set of actions (six fixed swimming directions\nin our case) to have enough freedom to find efficient strategies to move upward\nand escape local fluid traps. The key ingredient is the\nlearning-from-experience structure of the algorithm, which assigns positive or\nnegative rewards depending on whether the taken action is, or is not,\nprofitable for the predetermined goal in the long term horizon. This is another\nexample supporting the efficiency of the reinforcement learning approach to\nlearn how to accomplish difficult tasks in complex fluid environments.\n",
"title": "Finding Efficient Swimming Strategies in a Three Dimensional Chaotic Flow by Reinforcement Learning"
}
| null | null | null | null | true | null |
5176
| null |
Default
| null | null |
null |
{
"abstract": " We propose and analyze a variational wave function for a\npopulation-imbalanced one-dimensional Fermi gas that allows for\nFulde-Ferrell-Larkin-Ovchinnikov (FFLO) type pairing correlations among the two\nfermion species, while also accounting for the harmonic confining potential. In\nthe strongly interacting regime, we find large spatial oscillations of the\norder parameter, indicative of an FFLO state. The obtained density profiles\nversus imbalance are consistent with recent experimental results as well as\nwith theoretical calculations based on combining Bethe ansatz with the local\ndensity approximation. Although we find no signature of the FFLO state in the\ndensities of the two fermion species, we show that the oscillations of the\norder parameter appear in density-density correlations, both in-situ and after\nfree expansion. Furthermore, above a critical polarization, the value of which\ndepends on the interaction, we find the unpaired Fermi-gas state to be\nenergetically more favorable.\n",
"title": "Trapped imbalanced fermionic superfluids in one dimension: A variational approach"
}
| null | null | null | null | true | null |
5177
| null |
Default
| null | null |
null |
{
"abstract": " We evaluated the prospects of quantifying the parameterized post-Newtonian\nparameter beta and solar quadrupole moment J2 with observations of near-Earth\nasteroids with large orbital precession rates (9 to 27 arcsec century$^{-1}$).\nWe considered existing optical and radar astrometry, as well as radar\nastrometry that can realistically be obtained with the Arecibo planetary radar\nin the next five years. Our sensitivity calculations relied on a traditional\ncovariance analysis and Monte Carlo simulations. We found that independent\nestimates of beta and J2 can be obtained with precisions of $6\\times10^{-4}$\nand $3\\times10^{-8}$, respectively. Because we assumed rather conservative\nobservational uncertainties, as is the usual practice when reporting radar\nastrometry, it is likely that the actual precision will be closer to\n$2\\times10^{-4}$ and $10^{-8}$, respectively. A purely dynamical determination\nof solar oblateness with asteroid radar astronomy may therefore rival the\nhelioseismology determination.\n",
"title": "Prospects of dynamical determination of General Relativity parameter beta and solar quadrupole moment J2 with asteroid radar astronomy"
}
| null | null | null | null | true | null |
5178
| null |
Default
| null | null |
null |
{
"abstract": " Each time a learner in a self-paced online course is trying to answer an\nassessment question, it takes some time to submit the answer, and if multiple\nattempts are allowed and the first answer was incorrect, it takes some time to\nsubmit the second attempt, and so on. Here we study the distribution of such\n\"response times\". We find that the log-normal statistical model for such times,\npreviously suggested in the literature, holds for online courses qualitatively.\nUsers who, according to this model, tend to take longer on submits are more\nlikely to complete the course, have a higher level of engagement and achieve a\nhigher grade. This finding can be the basis for designing interventions in\nonline courses, such as MOOCs, which would encourage some users to slow down.\n",
"title": "Modelling and Using Response Times in Online Courses"
}
| null | null | null | null | true | null |
5179
| null |
Default
| null | null |
null |
{
"abstract": " We give a concise presentation of the Univalent Foundations of mathematics\noutlining the main ideas, followed by a discussion of the UniMath library of\nformalized mathematics implementing the ideas of the Univalent Foundations\n(section 1), and the challenges one faces in attempting to design a large-scale\nlibrary of formalized mathematics (section 2). This leads us to a general\ndiscussion about the links between architecture and mathematics where a meeting\nof minds is revealed between architects and mathematicians (section 3). On the\nway our odyssey from the foundations to the \"horizon\" of mathematics will lead\nus to meet the mathematicians David Hilbert and Nicolas Bourbaki as well as the\narchitect Christopher Alexander.\n",
"title": "Univalent Foundations and the UniMath Library"
}
| null | null | null | null | true | null |
5180
| null |
Default
| null | null |
null |
{
"abstract": " This paper addresses the problem of synchronizing orthogonal matrices over\ndirected graphs. For synchronized transformations (or matrices), composite\ntransformations over loops equal the identity. We formulate the synchronization\nproblem as a least-squares optimization problem with nonlinear constraints. The\nsynchronization problem appears as one of the key components in applications\nranging from 3D-localization to image registration. The main contributions of\nthis work can be summarized as the introduction of two novel algorithms; one\nfor symmetric graphs and one for graphs that are possibly asymmetric. Under\ngeneral conditions, the former has guaranteed convergence to the solution of a\nspectral relaxation to the synchronization problem. The latter is stable for\nsmall step sizes when the graph is quasi-strongly connected. The proposed\nmethods are verified in numerical simulations.\n",
"title": "Distributed methods for synchronization of orthogonal matrices over graphs"
}
| null | null | null | null | true | null |
5181
| null |
Default
| null | null |
null |
{
"abstract": " Threshold theorem is probably the most important development of mathematical\nepidemic modelling. Unfortunately, some models may not behave according to the\nthreshold. In this paper, we will focus on the final outcome of SIR model with\ndemography. The behaviour of the model approached by deteministic and\nstochastic models will be introduced, mainly using simulations. Furthermore, we\nwill also investigate the dynamic of susceptibles in population in absence of\ninfective. We have successfully showed that both deterministic and stochastic\nmodels performed similar results when $R_0 \\leq 1$. That is, the disease-free\nstage in the epidemic. But when $R_0 > 1$, the deterministic and stochastic\napproaches had different interpretations.\n",
"title": "Stochastic Model of SIR Epidemic Modelling"
}
| null | null | null | null | true | null |
5182
| null |
Default
| null | null |
null |
{
"abstract": " An evolutionary model for emergence of diversity in language is developed. We\ninvestigated the effects of two real life observations, namely, people prefer\npeople that they communicate with well, and people interact with people that\nare physically close to each other. Clearly these groups are relatively small\ncompared to the entire population. We restrict selection of the teachers from\nsuch small groups, called imitation sets, around parents. Then the child learns\nlanguage from a teacher selected within the imitation set of her parent. As a\nresult, there are subcommunities with their own languages developed. Within\nsubcommunity comprehension is found to be high. The number of languages is\nrelated to the relative size of imitation set by a power law.\n",
"title": "Parent Oriented Teacher Selection Causes Language Diversity"
}
| null | null | null | null | true | null |
5183
| null |
Default
| null | null |
null |
{
"abstract": " Random walks are at the heart of many existing network embedding methods.\nHowever, such algorithms have many limitations that arise from the use of\nrandom walks, e.g., the features resulting from these methods are unable to\ntransfer to new nodes and graphs as they are tied to vertex identity. In this\nwork, we introduce the Role2Vec framework which uses the flexible notion of\nattributed random walks, and serves as a basis for generalizing existing\nmethods such as DeepWalk, node2vec, and many others that leverage random walks.\nOur proposed framework enables these methods to be more widely applicable for\nboth transductive and inductive learning as well as for use on graphs with\nattributes (if available). This is achieved by learning functions that\ngeneralize to new nodes and graphs. We show that our proposed framework is\neffective with an average AUC improvement of 16.55% while requiring on average\n853x less space than existing methods on a variety of graphs.\n",
"title": "Learning Role-based Graph Embeddings"
}
| null | null | null | null | true | null |
5184
| null |
Default
| null | null |
null |
{
"abstract": " Ultracold atomic physics experiments offer a nearly ideal context for the\ninvestigation of quantum systems far from equilibrium. We describe three\nrelated emerging directions of research into extreme non-equilibrium phenomena\nin atom traps: quantum emulation of ultrafast atom-light interactions, coherent\nphasonic spectroscopy in tunable quasicrystals, and realization of Floquet\nmatter in strongly-driven lattice systems. We show that all three should enable\nquantum emulation in parameter regimes inaccessible in solid-state experiments,\nfacilitating a complementary approach to open problems in non-equilibrium\ncondensed matter.\n",
"title": "Quantum Emulation of Extreme Non-equilibrium Phenomena with Trapped Atoms"
}
| null | null | null | null | true | null |
5185
| null |
Default
| null | null |
null |
{
"abstract": " We consider solving convex-concave saddle point problems. We focus on two\nvariants of gradient decent-ascent algorithms, Extra-gradient (EG) and\nOptimistic Gradient (OGDA) methods, and show that they admit a unified analysis\nas approximations of the classical proximal point method for solving\nsaddle-point problems. This viewpoint enables us to generalize EG (in terms of\nextrapolation steps) and OGDA (in terms of parameters) and obtain new\nconvergence rate results for these algorithms for the bilinear case as well as\nthe strongly convex-concave case.\n",
"title": "A Unified Analysis of Extra-gradient and Optimistic Gradient Methods for Saddle Point Problems: Proximal Point Approach"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
5186
| null |
Validated
| null | null |
null |
{
"abstract": " Power system dynamic state estimation is essential to monitoring and\ncontrolling power system stability. Kalman filtering approaches are predominant\nin estimation of synchronous machine dynamic states (i.e. rotor angle and rotor\nspeed). This paper proposes an adaptive multi-step prediction (AMSP) approach\nto improve the extended Kalman filter s (EKF) performance in estimating the\ndynamic states of a synchronous machine. The proposed approach consists of\nthree major steps. First, two indexes are defined to quantify the non-linearity\nlevels of the state transition function and measurement function, respectively.\nSecond, based on the non-linearity indexes, a multi prediction factor (Mp) is\ndefined to determine the number of prediction steps. And finally, to mitigate\nthe non-linearity impact on dynamic state estimation (DSE) accuracy, the\nprediction step repeats a few time based on Mp before performing the correction\nstep. The two-area four-machine system is used to evaluate the effectiveness of\nthe proposed AMSP approach. It is shown through the Monte-Carlo method that a\ngood trade-off between estimation accuracy and computational time can be\nachieved effectively through the proposed AMSP approach.\n",
"title": "Adaptive Multi-Step Prediction based EKF to Power System Dynamic State Estimation"
}
| null | null | null | null | true | null |
5187
| null |
Default
| null | null |
null |
{
"abstract": " A number of statistical estimation problems can be addressed by semidefinite\nprograms (SDP). While SDPs are solvable in polynomial time using interior point\nmethods, in practice generic SDP solvers do not scale well to high-dimensional\nproblems. In order to cope with this problem, Burer and Monteiro proposed a\nnon-convex rank-constrained formulation, which has good performance in practice\nbut is still poorly understood theoretically.\nIn this paper we study the rank-constrained version of SDPs arising in MaxCut\nand in synchronization problems. We establish a Grothendieck-type inequality\nthat proves that all the local maxima and dangerous saddle points are within a\nsmall multiplicative gap from the global maximum. We use this structural\ninformation to prove that SDPs can be solved within a known accuracy, by\napplying the Riemannian trust-region method to this non-convex problem, while\nconstraining the rank to be of order one. For the MaxCut problem, our\ninequality implies that any local maximizer of the rank-constrained SDP\nprovides a $ (1 - 1/(k-1)) \\times 0.878$ approximation of the MaxCut, when the\nrank is fixed to $k$.\nWe then apply our results to data matrices generated according to the\nGaussian ${\\mathbb Z}_2$ synchronization problem, and the two-groups stochastic\nblock model with large bounded degree. We prove that the error achieved by\nlocal maximizers undergoes a phase transition at the same threshold as for\ninformation-theoretically optimal methods.\n",
"title": "Solving SDPs for synchronization and MaxCut problems via the Grothendieck inequality"
}
| null | null | null | null | true | null |
5188
| null |
Default
| null | null |
null |
{
"abstract": " We study the existence and stability of stationary solutions of\nPoisson-Nernst- Planck equations with steric effects (PNP-steric equations)\nwith two counter-charged species. These equations describe steady current\nthrough open ionic channels quite well. The current levels in open ionic\nchannels are known to switch between `open' or `closed' states in a spontaneous\nstochastic process called gating, suggesting that their governing equations\nshould give rise to multiple stationary solutions that enable such multi-stable\nbehavior. We show that within a range of parameters, steric effects give rise\nto multiple stationary solutions that are smooth. These solutions, however, are\nall unstable under PNP-steric dynamics. Following these findings, we introduce\na novel PNP-Cahn-Hilliard model, and show that it admits multiple stationary\nsolutions that are smooth and stable. The various branches of stationary\nsolutions and their stability are mapped utilizing bifurcation analysis and\nnumerical continuation methods.\n",
"title": "Poisson-Nernst-Planck equations with steric effects - non-convexity and multiple stationary solutions"
}
| null | null | null | null | true | null |
5189
| null |
Default
| null | null |
null |
{
"abstract": " In this paper we present a new algorithm for compressive sensing that makes\nuse of binary measurement matrices and achieves exact recovery of ultra sparse\nvectors, in a single pass and without any iterations. Due to its noniterative\nnature, our algorithm is hundreds of times faster than $\\ell_1$-norm\nminimization, and methods based on expander graphs, both of which require\nmultiple iterations. Our algorithm can accommodate nearly sparse vectors, in\nwhich case it recovers index set of the largest components, and can also\naccommodate burst noise measurements. Compared to compressive sensing methods\nthat are guaranteed to achieve exact recovery of all sparse vectors, our method\nrequires fewer measurements However, methods that achieve statistical recovery,\nthat is, recovery of almost all but not all sparse vectors, can require fewer\nmeasurements than our method.\n",
"title": "A Fast Noniterative Algorithm for Compressive Sensing Using Binary Measurement Matrices"
}
| null | null | null | null | true | null |
5190
| null |
Default
| null | null |
null |
{
"abstract": " An important goal common to domain adaptation and causal inference is to make\naccurate predictions when the distributions for the source (or training)\ndomain(s) and target (or test) domain(s) differ. In many cases, these different\ndistributions can be modeled as different contexts of a single underlying\nsystem, in which each distribution corresponds to a different perturbation of\nthe system, or in causal terms, an intervention. We focus on a class of such\ncausal domain adaptation problems, where data for one or more source domains\nare given, and the task is to predict the distribution of a certain target\nvariable from measurements of other variables in one or more target domains. We\npropose an approach for solving these problems that exploits causal inference\nand does not rely on prior knowledge of the causal graph, the type of\ninterventions or the intervention targets. We demonstrate our approach by\nevaluating a possible implementation on simulated and real world data.\n",
"title": "Domain Adaptation by Using Causal Inference to Predict Invariant Conditional Distributions"
}
| null | null | null | null | true | null |
5191
| null |
Default
| null | null |
null |
{
"abstract": " In this paper we focus on the problem of finding the optimal weights of the\nshallowest of neural networks consisting of a single Rectified Linear Unit\n(ReLU). These functions are of the form $\\mathbf{x}\\rightarrow\n\\max(0,\\langle\\mathbf{w},\\mathbf{x}\\rangle)$ with $\\mathbf{w}\\in\\mathbb{R}^d$\ndenoting the weight vector. We focus on a planted model where the inputs are\nchosen i.i.d. from a Gaussian distribution and the labels are generated\naccording to a planted weight vector. We first show that mini-batch stochastic\ngradient descent when suitably initialized, converges at a geometric rate to\nthe planted model with a number of samples that is optimal up to numerical\nconstants. Next we focus on a parallel implementation where in each iteration\nthe mini-batch gradient is calculated in a distributed manner across multiple\nprocessors and then broadcast to a master or all other processors. To reduce\nthe communication cost in this setting we utilize a Quanitzed Stochastic\nGradient Scheme (QSGD) where the partial gradients are quantized. Perhaps\nunexpectedly, we show that QSGD maintains the fast convergence of SGD to a\nglobally optimal model while significantly reducing the communication cost. We\nfurther corroborate our numerical findings via various experiments including\ndistributed implementations over Amazon EC2.\n",
"title": "Fitting ReLUs via SGD and Quantized SGD"
}
| null | null | null | null | true | null |
5192
| null |
Default
| null | null |
null |
{
"abstract": " An alternative proof is given of the existence of greatest lower bounds in\nthe imbalance order of binary maximal instantaneous codes of a given size.\nThese codes are viewed as maximal antichains of a given size in the infinite\nbinary tree of 0-1 words. The proof proposed makes use of a single balancing\noperation instead of expansion and contraction as in the original proof of the\nexistence of glb.\n",
"title": "The meet operation in the imbalance lattice of maximal instantaneous codes: alternative proof of existence"
}
| null | null | null | null | true | null |
5193
| null |
Default
| null | null |
null |
{
"abstract": " We study approximations of the partition function of dense graphical models.\nPartition functions of graphical models play a fundamental role is statistical\nphysics, in statistics and in machine learning. Two of the main methods for\napproximating the partition function are Markov Chain Monte Carlo and\nVariational Methods. An impressive body of work in mathematics, physics and\ntheoretical computer science provides conditions under which Markov Chain Monte\nCarlo methods converge in polynomial time. These methods often lead to\npolynomial time approximation algorithms for the partition function in cases\nwhere the underlying model exhibits correlation decay. There are very few\ntheoretical guarantees for the performance of variational methods. One\nexception is recent results by Risteski (2016) who considered dense graphical\nmodels and showed that using variational methods, it is possible to find an\n$O(\\epsilon n)$ additive approximation to the log partition function in time\n$n^{O(1/\\epsilon^2)}$ even in a regime where correlation decay does not hold.\nWe show that under essentially the same conditions, an $O(\\epsilon n)$\nadditive approximation of the log partition function can be found in constant\ntime, independent of $n$. In particular, our results cover dense Ising and\nPotts models as well as dense graphical models with $k$-wise interaction. They\nalso apply for low threshold rank models.\n",
"title": "Approximating Partition Functions in Constant Time"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
5194
| null |
Validated
| null | null |
null |
{
"abstract": " In this paper, we study Hyers-Ulam stability for integral equation of\nVolterra type in time scale setting. Moreover we study the stability of the\nconsidered equation in Hyers-Ulam-Rassias sense. Our technique depends on\nsuccessive approximation method, and we use time scale variant of induction\nprinciple to show that equation (1.1) is stable on unbounded domains in\nHyers-Ulam-Rassias sense.\n",
"title": "Stability of a Volterra Integral Equation on Time Scales"
}
| null | null |
[
"Mathematics"
] | null | true | null |
5195
| null |
Validated
| null | null |
null |
{
"abstract": " $\\omega$ Centauri (NGC 5139) hosts hundreds of pulsating variable stars of\ndifferent types, thus representing a treasure trove for studies of their\ncorresponding period-luminosity (PL) relations. Our goal in this study is to\nobtain the PL relations for RR Lyrae, and SX Phoenicis stars in the field of\nthe cluster, based on high-quality, well-sampled light curves in the\nnear-infrared (IR). $\\omega$ Centauri was observed using VIRCAM mounted on\nVISTA. A total of 42 epochs in $J$ and 100 epochs in $K_{\\rm S}$ were obtained,\nspanning 352 days. Point-spread function photometry was performed using DoPhot\nand DAOPHOT in the outer and inner regions of the cluster, respectively. Based\non the comprehensive catalogue of near-IR light curves thus secured, PL\nrelations were obtained for the different types of pulsators in the cluster,\nboth in the $J$ and $K_{\\rm S}$ bands. This includes the first PL relations in\nthe near-IR for fundamental-mode SX Phoenicis stars. The near-IR magnitudes and\nperiods of Type II Cepheids and RR Lyrae stars were used to derive an updated\ntrue distance modulus to the cluster, with a resulting value of $(m-M)_0 =\n13.708 \\pm 0.035 \\pm 0.10$ mag, where the error bars correspond to the adopted\nstatistical and systematic errors, respectively. Adding the errors in\nquadrature, this is equivalent to a heliocentric distance of $5.52\\pm 0.27$\nkpc.\n",
"title": "Near-IR period-luminosity relations for pulsating stars in $ω$ Centauri (NGC 5139)"
}
| null | null | null | null | true | null |
5196
| null |
Default
| null | null |
null |
{
"abstract": " We introduce pseudo-deterministic interactive proofs (psdAM): interactive\nproof systems for search problems where the verifier is guaranteed with high\nprobability to output the same output on different executions. As in the case\nwith classical interactive proofs, the verifier is a probabilistic polynomial\ntime algorithm interacting with an untrusted powerful prover.\nWe view pseudo-deterministic interactive proofs as an extension of the study\nof pseudo-deterministic randomized polynomial time algorithms: the goal of the\nlatter is to find canonical solutions to search problems whereas the goal of\nthe former is to prove that a solution to a search problem is canonical to a\nprobabilistic polynomial time verifier. Alternatively, one may think of the\npowerful prover as aiding the probabilistic polynomial time verifier to find\ncanonical solutions to search problems, with high probability over the\nrandomness of the verifier. The challenge is that pseudo-determinism should\nhold not only with respect to the randomness, but also with respect to the\nprover: a malicious prover should not be able to cause the verifier to output a\nsolution other than the unique canonical one.\n",
"title": "Pseudo-deterministic Proofs"
}
| null | null | null | null | true | null |
5197
| null |
Default
| null | null |
null |
{
"abstract": " Erosion and deposition during flow through porous media can lead to large\nerosive bursts that manifest as jumps in permeability and pressure loss. Here\nwe reveal that the cause of these bursts is the re-opening of clogged pores\nwhen the pressure difference between two opposite sites of the pore surpasses a\ncertain threshold. We perform numerical simulations of flow through porous\nmedia and compare our predictions to experimental results, recovering with\nexcellent agreement shape and power-law distribution of pressure loss jumps,\nand the behavior of the permeability jumps as function of particle\nconcentration. Furthermore, we find that erosive bursts only occur for pressure\ngradient thresholds within the range of two critical values, independent on how\nthe flow is driven. Our findings provide a better understanding of sudden sand\nproduction in oil wells and breakthrough in filtration.\n",
"title": "The Mechanism behind Erosive Bursts in Porous Media"
}
| null | null |
[
"Physics"
] | null | true | null |
5198
| null |
Validated
| null | null |
null |
{
"abstract": " We study the maximum likelihood degree (ML degree) of toric varieties, known\nas discrete exponential models in statistics. By introducing scaling\ncoefficients to the monomial parameterization of the toric variety, one can\nchange the ML degree. We show that the ML degree is equal to the degree of the\ntoric variety for generic scalings, while it drops if and only if the scaling\nvector is in the locus of the principal $A$-determinant. We also illustrate how\nto compute the ML estimate of a toric variety numerically via homotopy\ncontinuation from a scaled toric variety with low ML degree. Throughout, we\ninclude examples motivated by algebraic geometry and statistics. We compute the\nML degree of rational normal scrolls and a large class of Veronese-type\nvarieties. In addition, we investigate the ML degree of scaled Segre varieties,\nhierarchical loglinear models, and graphical models.\n",
"title": "The Maximum Likelihood Degree of Toric Varieties"
}
| null | null |
[
"Mathematics",
"Statistics"
] | null | true | null |
5199
| null |
Validated
| null | null |
null |
{
"abstract": " We study the incompressible limit of a pressure correction MAC scheme [3] for\nthe unstationary compressible barotropic Navier-Stokes equations. Provided the\ninitial data are well-prepared, the solution of the numerical scheme converges,\nas the Mach number tends to zero, towards the solution of the classical\npressure correction inf-sup stable MAC scheme for the incompressible\nNavier-Stokes equations.\n",
"title": "Low Mach number limit of a pressure correction MAC scheme for compressible barotropic flows"
}
| null | null | null | null | true | null |
5200
| null |
Default
| null | null |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.