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": " This paper presents a fixturing strategy for regrasping that does not require\na physical fixture. To regrasp an object in a gripper, a robot pushes the\nobject against external contact/s in the environment such that the external\ncontact keeps the object stationary while the fingers slide over the object. We\ncall this manipulation technique fixtureless fixturing. Exploiting the\nmechanics of pushing, we characterize a convex polyhedral set of pushes that\nresults in fixtureless fixturing. These pushes are robust against uncertainty\nin the object inertia, grasping force, and the friction at the contacts. We\npropose a sampling-based planner that uses the sets of robust pushes to rapidly\nbuild a tree of reachable grasps. A path in this tree is a pushing strategy,\npossibly involving pushes from different sides, to regrasp the object. We\ndemonstrate the experimental validity and robustness of the proposed\nmanipulation technique with different regrasp examples on a manipulation\nplatform. Such a fast and flexible regrasp planner facilitates versatile and\nflexible automation solutions.\n",
"title": "Regrasping by Fixtureless Fixturing"
} | null | null | [
"Computer Science"
]
| null | true | null | 601 | null | Validated | null | null |
null | {
"abstract": " Labeled Latent Dirichlet Allocation (LLDA) is an extension of the standard\nunsupervised Latent Dirichlet Allocation (LDA) algorithm, to address\nmulti-label learning tasks. Previous work has shown it to perform in par with\nother state-of-the-art multi-label methods. Nonetheless, with increasing label\nsets sizes LLDA encounters scalability issues. In this work, we introduce\nSubset LLDA, a simple variant of the standard LLDA algorithm, that not only can\neffectively scale up to problems with hundreds of thousands of labels but also\nimproves over the LLDA state-of-the-art. We conduct extensive experiments on\neight data sets, with label sets sizes ranging from hundreds to hundreds of\nthousands, comparing our proposed algorithm with the previously proposed LLDA\nalgorithms (Prior--LDA, Dep--LDA), as well as the state of the art in extreme\nmulti-label classification. The results show a steady advantage of our method\nover the other LLDA algorithms and competitive results compared to the extreme\nmulti-label classification algorithms.\n",
"title": "Subset Labeled LDA for Large-Scale Multi-Label Classification"
} | null | null | null | null | true | null | 602 | null | Default | null | null |
null | {
"abstract": " Multimedia Forensics allows to determine whether videos or images have been\ncaptured with the same device, and thus, eventually, by the same person.\nCurrently, the most promising technology to achieve this task, exploits the\nunique traces left by the camera sensor into the visual content. Anyway, image\nand video source identification are still treated separately from one another.\nThis approach is limited and anachronistic if we consider that most of the\nvisual media are today acquired using smartphones, that capture both images and\nvideos. In this paper we overcome this limitation by exploring a new approach\nthat allows to synergistically exploit images and videos to study the device\nfrom which they both come. Indeed, we prove it is possible to identify the\nsource of a digital video by exploiting a reference sensor pattern noise\ngenerated from still images taken by the same device of the query video. The\nproposed method provides comparable or even better performance, when compared\nto the current video identification strategies, where a reference pattern is\nestimated from video frames. We also show how this strategy can be effective\neven in case of in-camera digitally stabilized videos, where a non-stabilized\nreference is not available, by solving some state-of-the-art limitations. We\nexplore a possible direct application of this result, that is social media\nprofile linking, i.e. discovering relationships between two or more social\nmedia profiles by comparing the visual contents - images or videos - shared\ntherein.\n",
"title": "A Hybrid Approach to Video Source Identification"
} | null | null | null | null | true | null | 603 | null | Default | null | null |
null | {
"abstract": " We present an informal review of recent work on the asymptotics of\nApproximate Bayesian Computation (ABC). In particular we focus on how does the\nABC posterior, or point estimates obtained by ABC, behave in the limit as we\nhave more data? The results we review show that ABC can perform well in terms\nof point estimation, but standard implementations will over-estimate the\nuncertainty about the parameters. If we use the regression correction of\nBeaumont et al. then ABC can also accurately quantify this uncertainty. The\ntheoretical results also have practical implications for how to implement ABC.\n",
"title": "Asymptotics of ABC"
} | null | null | null | null | true | null | 604 | null | Default | null | null |
null | {
"abstract": " We consider the problem of learning sparse polymatrix games from observations\nof strategic interactions. We show that a polynomial time method based on\n$\\ell_{1,2}$-group regularized logistic regression recovers a game, whose Nash\nequilibria are the $\\epsilon$-Nash equilibria of the game from which the data\nwas generated (true game), in $\\mathcal{O}(m^4 d^4 \\log (pd))$ samples of\nstrategy profiles --- where $m$ is the maximum number of pure strategies of a\nplayer, $p$ is the number of players, and $d$ is the maximum degree of the game\ngraph. Under slightly more stringent separability conditions on the payoff\nmatrices of the true game, we show that our method learns a game with the exact\nsame Nash equilibria as the true game. We also show that $\\Omega(d \\log (pm))$\nsamples are necessary for any method to consistently recover a game, with the\nsame Nash-equilibria as the true game, from observations of strategic\ninteractions. We verify our theoretical results through simulation experiments.\n",
"title": "Learning Sparse Polymatrix Games in Polynomial Time and Sample Complexity"
} | null | null | null | null | true | null | 605 | null | Default | null | null |
null | {
"abstract": " The KdV equation can be derived in the shallow water limit of the Euler\nequations. Over the last few decades, this equation has been extended to\ninclude higher order effects. Although this equation has only one conservation\nlaw, exact periodic and solitonic solutions exist. Khare and Saxena\n\\cite{KhSa,KhSa14,KhSa15} demonstrated the possibility of generating new exact\nsolutions by combining known ones for several fundamental equations (e.g.,\nKorteweg - de Vries, Nonlinear Schrödinger). Here we find that this\nconstruction can be repeated for higher order, non-integrable extensions of\nthese equations. Contrary to many statements in the literature, there seems to\nbe no correlation between integrability and the number of nonlinear one\nvariable wave solutions.\n",
"title": "Superposition solutions to the extended KdV equation for water surface waves"
} | null | null | [
"Physics"
]
| null | true | null | 606 | null | Validated | null | null |
null | {
"abstract": " This paper proposes a new actor-critic-style algorithm called Dual\nActor-Critic or Dual-AC. It is derived in a principled way from the Lagrangian\ndual form of the Bellman optimality equation, which can be viewed as a\ntwo-player game between the actor and a critic-like function, which is named as\ndual critic. Compared to its actor-critic relatives, Dual-AC has the desired\nproperty that the actor and dual critic are updated cooperatively to optimize\nthe same objective function, providing a more transparent way for learning the\ncritic that is directly related to the objective function of the actor. We then\nprovide a concrete algorithm that can effectively solve the minimax\noptimization problem, using techniques of multi-step bootstrapping, path\nregularization, and stochastic dual ascent algorithm. We demonstrate that the\nproposed algorithm achieves the state-of-the-art performances across several\nbenchmarks.\n",
"title": "Boosting the Actor with Dual Critic"
} | null | null | null | null | true | null | 607 | null | Default | null | null |
null | {
"abstract": " Counting dominating sets in a graph $G$ is closely related to the\nneighborhood complex of $G$. We exploit this relation to prove that the number\nof dominating sets $d(G)$ of a graph is determined by the number of complete\nbipartite subgraphs of its complement. More precisely, we state the following.\nLet $G$ be a simple graph of order $n$ such that its complement has exactly\n$a(G)$ subgraphs isomorphic to $K_{2p,2q}$ and exactly $b(G)$ subgraphs\nisomorphic to $K_{2p+1,2q+1}$. Then $d(G) = 2^n -1 + 2[a(G)-b(G)]$. We also\nshow some new relations between the domination polynomial and the neighborhood\npolynomial of a graph.\n",
"title": "Counting Dominating Sets of Graphs"
} | null | null | null | null | true | null | 608 | null | Default | null | null |
null | {
"abstract": " High signal to noise ratio (SNR) consistency of model selection criteria in\nlinear regression models has attracted a lot of attention recently. However,\nmost of the existing literature on high SNR consistency deals with model order\nselection. Further, the limited literature available on the high SNR\nconsistency of subset selection procedures (SSPs) is applicable to linear\nregression with full rank measurement matrices only. Hence, the performance of\nSSPs used in underdetermined linear models (a.k.a compressive sensing (CS)\nalgorithms) at high SNR is largely unknown. This paper fills this gap by\nderiving necessary and sufficient conditions for the high SNR consistency of\npopular CS algorithms like $l_0$-minimization, basis pursuit de-noising or\nLASSO, orthogonal matching pursuit and Dantzig selector. Necessary conditions\nanalytically establish the high SNR inconsistency of CS algorithms when used\nwith the tuning parameters discussed in literature. Novel tuning parameters\nwith SNR adaptations are developed using the sufficient conditions and the\nchoice of SNR adaptations are discussed analytically using convergence rate\nanalysis. CS algorithms with the proposed tuning parameters are numerically\nshown to be high SNR consistent and outperform existing tuning parameters in\nthe moderate to high SNR regime.\n",
"title": "High SNR Consistent Compressive Sensing"
} | null | null | null | null | true | null | 609 | null | Default | null | null |
null | {
"abstract": " Reduction of communication and efficient partitioning are key issues for\nachieving scalability in hierarchical $N$-Body algorithms like FMM. In the\npresent work, we propose four independent strategies to improve partitioning\nand reduce communication. First of all, we show that the conventional wisdom of\nusing space-filling curve partitioning may not work well for boundary integral\nproblems, which constitute about 50% of FMM's application user base. We propose\nan alternative method which modifies orthogonal recursive bisection to solve\nthe cell-partition misalignment that has kept it from scaling previously.\nSecondly, we optimize the granularity of communication to find the optimal\nbalance between a bulk-synchronous collective communication of the local\nessential tree and an RDMA per task per cell. Finally, we take the dynamic\nsparse data exchange proposed by Hoefler et al. and extend it to a hierarchical\nsparse data exchange, which is demonstrated at scale to be faster than the MPI\nlibrary's MPI_Alltoallv that is commonly used.\n",
"title": "Communication Reducing Algorithms for Distributed Hierarchical N-Body Problems with Boundary Distributions"
} | null | null | null | null | true | null | 610 | null | Default | null | null |
null | {
"abstract": " In this paper, we focus on fully automatic traffic surveillance camera\ncalibration, which we use for speed measurement of passing vehicles. We improve\nover a recent state-of-the-art camera calibration method for traffic\nsurveillance based on two detected vanishing points. More importantly, we\npropose a novel automatic scene scale inference method. The method is based on\nmatching bounding boxes of rendered 3D models of vehicles with detected\nbounding boxes in the image. The proposed method can be used from arbitrary\nviewpoints, since it has no constraints on camera placement. We evaluate our\nmethod on the recent comprehensive dataset for speed measurement BrnoCompSpeed.\nExperiments show that our automatic camera calibration method by detection of\ntwo vanishing points reduces error by 50% (mean distance ratio error reduced\nfrom 0.18 to 0.09) compared to the previous state-of-the-art method. We also\nshow that our scene scale inference method is more precise, outperforming both\nstate-of-the-art automatic calibration method for speed measurement (error\nreduction by 86% -- 7.98km/h to 1.10km/h) and manual calibration (error\nreduction by 19% -- 1.35km/h to 1.10km/h). We also present qualitative results\nof the proposed automatic camera calibration method on video sequences obtained\nfrom real surveillance cameras in various places, and under different lighting\nconditions (night, dawn, day).\n",
"title": "Traffic Surveillance Camera Calibration by 3D Model Bounding Box Alignment for Accurate Vehicle Speed Measurement"
} | null | null | null | null | true | null | 611 | null | Default | null | null |
null | {
"abstract": " The success of autonomous systems will depend upon their ability to safely\nnavigate human-centric environments. This motivates the need for a real-time,\nprobabilistic forecasting algorithm for pedestrians, cyclists, and other agents\nsince these predictions will form a necessary step in assessing the risk of any\naction. This paper presents a novel approach to probabilistic forecasting for\npedestrians based on weighted sums of ordinary differential equations that are\nlearned from historical trajectory information within a fixed scene. The\nresulting algorithm is embarrassingly parallel and is able to work at real-time\nspeeds using a naive Python implementation. The quality of predicted locations\nof agents generated by the proposed algorithm is validated on a variety of\nexamples and considerably higher than existing state of the art approaches over\nlong time horizons.\n",
"title": "Technical Report for Real-Time Certified Probabilistic Pedestrian Forecasting"
} | null | null | null | null | true | null | 612 | null | Default | null | null |
null | {
"abstract": " This paper presents a distance-based discriminative framework for learning\nwith probability distributions. Instead of using kernel mean embeddings or\ngeneralized radial basis kernels, we introduce embeddings based on\ndissimilarity of distributions to some reference distributions denoted as\ntemplates. Our framework extends the theory of similarity of Balcan et al.\n(2008) to the population distribution case and we show that, for some learning\nproblems, some dissimilarity on distribution achieves low-error linear decision\nfunctions with high probability. Our key result is to prove that the theory\nalso holds for empirical distributions. Algorithmically, the proposed approach\nconsists in computing a mapping based on pairwise dissimilarity where learning\na linear decision function is amenable. Our experimental results show that the\nWasserstein distance embedding performs better than kernel mean embeddings and\ncomputing Wasserstein distance is far more tractable than estimating pairwise\nKullback-Leibler divergence of empirical distributions.\n",
"title": "Distance Measure Machines"
} | null | null | null | null | true | null | 613 | null | Default | null | null |
null | {
"abstract": " Molecular reflections on usual wall surfaces can be statistically described\nby the Maxwell diffuse reflection model, which has been successfully applied in\nthe DSBGK simulations. We develop the DSBGK algorithm to implement the\nCercignani-Lampis-Lord (CLL) reflection model, which is widely applied to\npolished surfaces and used particularly in modeling space shuttles to predict\nthe heat and force loads exerted by the high-speed flows around the surfaces.\nWe also extend the DSBGK method to simulate gas mixtures and high contrast of\nnumber densities of different components can be handled at a cost of memory\nusage much lower than that needed by the DSMC simulations because the average\nnumbers of simulated molecules of different components per cell can be equal in\nthe DSBGK simulations.\n",
"title": "DSBGK Method to Incorporate the CLL Reflection Model and to Simulate Gas Mixtures"
} | null | null | null | null | true | null | 614 | null | Default | null | null |
null | {
"abstract": " Let $f(a,b,c,d)=\\sqrt{a^2+b^2}+\\sqrt{c^2+d^2}-\\sqrt{(a+c)^2+(b+d)^2}$, let\n$(a,b,c,d)$ stand for $a,b,c,d\\in\\mathbb Z_{\\geq 0}$ such that $ad-bc=1$.\nDefine \\begin{equation} \\label{eq_main} F(s) = \\sum_{(a,b,c,d)} f(a,b,c,d)^s.\n\\end{equation} In other words, we consider the sum of the powers of the\ntriangle inequality defects for the lattice parallelograms (in the first\nquadrant) of area one.\nWe prove that $F(s)$ converges when $s>1/2$ and diverges at $s=1/2$. We also\nprove $$\\sum\\limits_{\\substack{(a,b,c,d),\\\\ 1\\leq a\\leq b, 1\\leq c\\leq d}}\n\\frac{1}{(a+b)^2(c+d)^2(a+b+c+d)^2} = 1/24,$$ and show a general method to\nobtain such formulae. The method comes from the consideration of the tropical\nanalogue of the caustic curves, whose moduli give a complete set of continuous\ninvariants on the space of convex domains.\n",
"title": "Tropical formulae for summation over a part of SL(2, Z)"
} | null | null | null | null | true | null | 615 | null | Default | null | null |
null | {
"abstract": " We consider the task of generating draws from a Markov jump process (MJP)\nbetween two time points at which the process is known. Resulting draws are\ntypically termed bridges and the generation of such bridges plays a key role in\nsimulation-based inference algorithms for MJPs. The problem is challenging due\nto the intractability of the conditioned process, necessitating the use of\ncomputationally intensive methods such as weighted resampling or Markov chain\nMonte Carlo. An efficient implementation of such schemes requires an\napproximation of the intractable conditioned hazard/propensity function that is\nboth cheap and accurate. In this paper, we review some existing approaches to\nthis problem before outlining our novel contribution. Essentially, we leverage\nthe tractability of a Gaussian approximation of the MJP and suggest a\ncomputationally efficient implementation of the resulting conditioned hazard\napproximation. We compare and contrast our approach with existing methods using\nthree examples.\n",
"title": "Efficient sampling of conditioned Markov jump processes"
} | null | null | null | null | true | null | 616 | null | Default | null | null |
null | {
"abstract": " Using holography, we model experiments in which a 2+1D strange metal is\npumped by a laser pulse into a highly excited state, after which the time\nevolution of the optical conductivity is probed. We consider a finite-density\nstate with mildly broken translation invariance and excite it by oscillating\nelectric field pulses. At zero density, the optical conductivity would assume\nits thermalized value immediately after the pumping has ended. At finite\ndensity, pulses with significant DC components give rise to slow exponential\nrelaxation, governed by a vector quasinormal mode. In contrast, for\nhigh-frequency pulses the amplitude of the quasinormal mode is strongly\nsuppressed, so that the optical conductivity assumes its thermalized value\neffectively instantaneously. This surprising prediction may provide a stimulus\nfor taking up the challenge to realize these experiments in the laboratory.\nSuch experiments would test a crucial open question faced by applied\nholography: Are its predictions artefacts of the large $N$ limit or do they\nenjoy sufficient UV independence to hold at least qualitatively in real-world\nsystems?\n",
"title": "Holography and thermalization in optical pump-probe spectroscopy"
} | null | null | null | null | true | null | 617 | null | Default | null | null |
null | {
"abstract": " In this paper, we study random subsampling of Gaussian process regression,\none of the simplest approximation baselines, from a theoretical perspective.\nAlthough subsampling discards a large part of training data, we show provable\nguarantees on the accuracy of the predictive mean/variance and its\ngeneralization ability. For analysis, we consider embedding kernel matrices\ninto graphons, which encapsulate the difference of the sample size and enables\nus to evaluate the approximation and generalization errors in a unified manner.\nThe experimental results show that the subsampling approximation achieves a\nbetter trade-off regarding accuracy and runtime than the Nyström and random\nFourier expansion methods.\n",
"title": "On Random Subsampling of Gaussian Process Regression: A Graphon-Based Analysis"
} | null | null | null | null | true | null | 618 | null | Default | null | null |
null | {
"abstract": " Both hybrid automata and action languages are formalisms for describing the\nevolution of dynamic systems. This paper establishes a formal relationship\nbetween them. We show how to succinctly represent hybrid automata in an action\nlanguage which in turn is defined as a high-level notation for answer set\nprogramming modulo theories (ASPMT) --- an extension of answer set programs to\nthe first-order level similar to the way satisfiability modulo theories (SMT)\nextends propositional satisfiability (SAT). We first show how to represent\nlinear hybrid automata with convex invariants by an action language modulo\ntheories. A further translation into SMT allows for computing them using SMT\nsolvers that support arithmetic over reals. Next, we extend the representation\nto the general class of non-linear hybrid automata allowing even non-convex\ninvariants. We represent them by an action language modulo ODE (Ordinary\nDifferential Equations), which can be compiled into satisfiability modulo ODE.\nWe developed a prototype system cplus2aspmt based on these translations, which\nallows for a succinct representation of hybrid transition systems that can be\ncomputed effectively by the state-of-the-art SMT solver dReal.\n",
"title": "Representing Hybrid Automata by Action Language Modulo Theories"
} | null | null | null | null | true | null | 619 | null | Default | null | null |
null | {
"abstract": " In this paper, an enthalpy-based multiple-relaxation-time (MRT) lattice\nBoltzmann (LB) method is developed for solid-liquid phase change heat transfer\nin metal foams under local thermal non-equilibrium (LTNE) condition. The\nenthalpy-based MRT-LB method consists of three different MRT-LB models: one for\nflow field based on the generalized non-Darcy model, and the other two for\nphase change material (PCM) and metal foam temperature fields described by the\nLTNE model. The moving solid-liquid phase interface is implicitly tracked\nthrough the liquid fraction, which is simultaneously obtained when the energy\nequations of PCM and metal foam are solved. The present method has several\ndistinctive features. First, as compared with previous studies, the present\nmethod avoids the iteration procedure, thus it retains the inherent merits of\nthe standard LB method and is superior over the iteration method in terms of\naccuracy and computational efficiency. Second, a volumetric LB scheme instead\nof the bounce-back scheme is employed to realize the no-slip velocity condition\nin the interface and solid phase regions, which is consistent with the actual\nsituation. Last but not least, the MRT collision model is employed, and with\nadditional degrees of freedom, it has the ability to reduce the numerical\ndiffusion across phase interface induced by solid-liquid phase change.\nNumerical tests demonstrate that the present method can be served as an\naccurate and efficient numerical tool for studying metal foam enhanced\nsolid-liquid phase change heat transfer in latent heat storage. Finally,\ncomparisons and discussions are made to offer useful information for practical\napplications of the present method.\n",
"title": "An enthalpy-based multiple-relaxation-time lattice Boltzmann method for solid-liquid phase change heat transfer in metal foams"
} | null | null | null | null | true | null | 620 | null | Default | null | null |
null | {
"abstract": " Barchan dunes are crescentic shape dunes with horns pointing downstream. The\npresent paper reports the formation of subaqueous barchan dunes from initially\nconical heaps in a rectangular channel. Because the most unique feature of a\nbarchan dune is its horns, we associate the timescale for the appearance of\nhorns to the formation of a barchan dune. A granular heap initially conical was\nplaced on the bottom wall of a closed conduit and it was entrained by a water\nflow in turbulent regime. After a certain time, horns appear and grow, until an\nequilibrium length is reached. Our results show the existence of the timescales\n$0.5t_c$ and $2.5t_c$ for the appearance and equilibrium of horns,\nrespectively, where $t_c$ is a characteristic time that scales with the grains\ndiameter, gravity acceleration, densities of the fluid and grains, and shear\nand threshold velocities.\n",
"title": "Birth of a subaqueous barchan dune"
} | null | null | [
"Physics"
]
| null | true | null | 621 | null | Validated | null | null |
null | {
"abstract": " Isotonic regression is a standard problem in shape-constrained estimation\nwhere the goal is to estimate an unknown nondecreasing regression function $f$\nfrom independent pairs $(x_i, y_i)$ where $\\mathbb{E}[y_i]=f(x_i), i=1, \\ldots\nn$. While this problem is well understood both statistically and\ncomputationally, much less is known about its uncoupled counterpart where one\nis given only the unordered sets $\\{x_1, \\ldots, x_n\\}$ and $\\{y_1, \\ldots,\ny_n\\}$. In this work, we leverage tools from optimal transport theory to derive\nminimax rates under weak moments conditions on $y_i$ and to give an efficient\nalgorithm achieving optimal rates. Both upper and lower bounds employ\nmoment-matching arguments that are also pertinent to learning mixtures of\ndistributions and deconvolution.\n",
"title": "Uncoupled isotonic regression via minimum Wasserstein deconvolution"
} | null | null | null | null | true | null | 622 | null | Default | null | null |
null | {
"abstract": " This paper considers a time-inconsistent stopping problem in which the\ninconsistency arises from non-constant time preference rates. We show that the\nsmooth pasting principle, the main approach that has been used to construct\nexplicit solutions for conventional time-consistent optimal stopping problems,\nmay fail under time-inconsistency. Specifically, we prove that the smooth\npasting principle solves a time-inconsistent problem within the intra-personal\ngame theoretic framework if and only if a certain inequality on the model\nprimitives is satisfied. We show that the violation of this inequality can\nhappen even for very simple non-exponential discount functions. Moreover, we\ndemonstrate that the stopping problem does not admit any intra-personal\nequilibrium whenever the smooth pasting principle fails. The \"negative\" results\nin this paper caution blindly extending the classical approaches for\ntime-consistent stopping problems to their time-inconsistent counterparts.\n",
"title": "Failure of Smooth Pasting Principle and Nonexistence of Equilibrium Stopping Rules under Time-Inconsistency"
} | null | null | null | null | true | null | 623 | null | Default | null | null |
null | {
"abstract": " The Internet of Things (IoT) demands authentication systems which can provide\nboth security and usability. Recent research utilizes the rich sensing\ncapabilities of smart devices to build security schemes operating without human\ninteraction, such as zero-interaction pairing (ZIP) and zero-interaction\nauthentication (ZIA). Prior work proposed a number of ZIP and ZIA schemes and\nreported promising results. However, those schemes were often evaluated under\nconditions which do not reflect realistic IoT scenarios. In addition, drawing\nany comparison among the existing schemes is impossible due to the lack of a\ncommon public dataset and unavailability of scheme implementations.\nIn this paper, we address these challenges by conducting the first\nlarge-scale comparative study of ZIP and ZIA schemes, carried out under\nrealistic conditions. We collect and release the most comprehensive dataset in\nthe domain to date, containing over 4250 hours of audio recordings and 1\nbillion sensor readings from three different scenarios, and evaluate five\nstate-of-the-art schemes based on these data. Our study reveals that the\neffectiveness of the existing proposals is highly dependent on the scenario\nthey are used in. In particular, we show that these schemes are subject to\nerror rates between 0.6% and 52.8%.\n",
"title": "Perils of Zero-Interaction Security in the Internet of Things"
} | null | null | null | null | true | null | 624 | null | Default | null | null |
null | {
"abstract": " The increasing number of protein-based metamaterials demands reliable and\nefficient methods to study the physicochemical properties they may display. In\nthis regard, we develop a simulation strategy based on Molecular Dynamics (MD)\nthat addresses the geometric degrees of freedom of an auxetic two-dimensional\nprotein crystal. This model consists of a network of impenetrable rigid squares\nlinked through massless rigid rods, thus featuring a large number of both\nholonomic and nonholonomic constraints. Our MD methodology is optimized to\nstudy highly constrained systems and allows for the simulation of long-time\ndynamics with reasonably large timesteps. The data extracted from the\nsimulations shows a persistent motional interdependence among the protein\nsubunits in the crystal. We characterize the dynamical correlations featured by\nthese subunits and identify two regimes characterized by their locality or\nnonlocality, depending on the geometric parameters of the crystal. From the\nsame data, we also calculate the Poisson\\rq{}s (longitudinal to axial strain)\nratio of the crystal, and learn that, due to holonomic constraints (rigidness\nof the rod links), the crystal remains auxetic even after significant changes\nin the original geometry. The nonholonomic ones (collisions between subunits)\nincrease the number of inhomogeneous deformations of the crystal, thus driving\nit away from an isotropic response. Our work provides the first simulation of\nthe dynamics of protein crystals and offers insights into promising mechanical\nproperties afforded by these materials.\n",
"title": "Coarse-grained simulation of auxetic, two-dimensional crystal dynamics"
} | null | null | null | null | true | null | 625 | null | Default | null | null |
null | {
"abstract": " Recent advances in the field of network representation learning are mostly\nattributed to the application of the skip-gram model in the context of graphs.\nState-of-the-art analogues of skip-gram model in graphs define a notion of\nneighbourhood and aim to find the vector representation for a node, which\nmaximizes the likelihood of preserving this neighborhood.\nIn this paper, we take a drastic departure from the existing notion of\nneighbourhood of a node by utilizing the idea of coreness. More specifically,\nwe utilize the well-established idea that nodes with similar core numbers play\nequivalent roles in the network and hence induce a novel and an organic notion\nof neighbourhood. Based on this idea, we propose core2vec, a new algorithmic\nframework for learning low dimensional continuous feature mapping for a node.\nConsequently, the nodes having similar core numbers are relatively closer in\nthe vector space that we learn.\nWe further demonstrate the effectiveness of core2vec by comparing word\nsimilarity scores obtained by our method where the node representations are\ndrawn from standard word association graphs against scores computed by other\nstate-of-the-art network representation techniques like node2vec, DeepWalk and\nLINE. Our results always outperform these existing methods\n",
"title": "Core2Vec: A core-preserving feature learning framework for networks"
} | null | null | null | null | true | null | 626 | null | Default | null | null |
null | {
"abstract": " Numerous studies have been carried out to measure wind pressures around\ncircular cylinders since the early 20th century due to its engineering\nsignificance. Consequently, a large amount of wind pressure data sets have\naccumulated, which presents an excellent opportunity for using machine learning\n(ML) techniques to train models to predict wind pressures around circular\ncylinders. Wind pressures around smooth circular cylinders are a function of\nmainly the Reynolds number (Re), turbulence intensity (Ti) of the incident\nwind, and circumferential angle of the cylinder. Considering these three\nparameters as the inputs, this study trained two ML models to predict mean and\nfluctuating pressures respectively. Three machine learning algorithms including\ndecision tree regressor, random forest, and gradient boosting regression trees\n(GBRT) were tested. The GBRT models exhibited the best performance for\npredicting both mean and fluctuating pressures, and they are capable of making\naccurate predictions for Re ranging from 10^4 to 10^6 and Ti ranging from 0% to\n15%. It is believed that the GBRT models provide very efficient and economical\nalternative to traditional wind tunnel tests and computational fluid dynamic\nsimulations for determining wind pressures around smooth circular cylinders\nwithin the studied Re and Ti range.\n",
"title": "Predicting wind pressures around circular cylinders using machine learning techniques"
} | null | null | null | null | true | null | 627 | null | Default | null | null |
null | {
"abstract": " We study the problem of constructing a (near) uniform random proper\n$q$-coloring of a simple $k$-uniform hypergraph with $n$ vertices and maximum\ndegree $\\Delta$. (Proper in that no edge is mono-colored and simple in that two\nedges have maximum intersection of size one). We show that if $q\\geq\n\\max\\{C_k\\log n,500k^3\\Delta^{1/(k-1)}\\}$ then the Glauber Dynamics will become\nclose to uniform in $O(n\\log n)$ time, given a random (improper) start. This\nimproves on the results in Frieze and Melsted [5].\n",
"title": "Randomly coloring simple hypergraphs with fewer colors"
} | null | null | null | null | true | null | 628 | null | Default | null | null |
null | {
"abstract": " We begin by introducing the main ideas of the paper under discussion. We\ndiscuss some interesting issues regarding adaptive component-wise credible\nintervals. We then briefly touch upon the concepts of self-similarity and\nexcessive bias restriction. This is then followed by some comments on the\nextensive simulation study carried out in the paper.\n",
"title": "Contributed Discussion to Uncertainty Quantification for the Horseshoe by Stéphanie van der Pas, Botond Szabó and Aad van der Vaart"
} | null | null | null | null | true | null | 629 | null | Default | null | null |
null | {
"abstract": " Time series shapelets are discriminative sub-sequences and their similarity\nto time series can be used for time series classification. Initial shapelet\nextraction algorithms searched shapelets by complete enumeration of all\npossible data sub-sequences. Research on shapelets for univariate time series\nproposed a mechanism called shapelet learning which parameterizes the shapelets\nand learns them jointly with a prediction model in an optimization procedure.\nTrivial extension of this method to multivariate time series does not yield\nvery good results due to the presence of noisy channels which lead to\noverfitting. In this paper we propose a shapelet learning scheme for\nmultivariate time series in which we introduce channel masks to discount noisy\nchannels and serve as an implicit regularization.\n",
"title": "Channel masking for multivariate time series shapelets"
} | null | null | null | null | true | null | 630 | null | Default | null | null |
null | {
"abstract": " In this work, we present an experimental study of spin mediated enhanced\nnegative magnetoresistance in Ni80Fe20 (50 nm)/p-Si (350 nm) bilayer. The\nresistance measurement shows a reduction of ~2.5% for the bilayer specimen as\ncompared to 1.3% for Ni80Fe20 (50 nm) on oxide specimen for an out-of-plane\napplied magnetic field of 3T. In the Ni80Fe20-only film, the negative\nmagnetoresistance behavior is attributed to anisotropic magnetoresistance. We\npropose that spin polarization due to spin-Hall effect is the underlying cause\nof the enhanced negative magnetoresistance observed in the bilayer. Silicon has\nweak spin orbit coupling so spin Hall magnetoresistance measurement is not\nfeasible. We use V2{\\omega} and V3{\\omega} measurement as a function of\nmagnetic field and angular rotation of magnetic field in direction normal to\nelectric current to elucidate the spin-Hall effect. The angular rotation of\nmagnetic field shows a sinusoidal behavior for both V2{\\omega} and V3{\\omega},\nwhich is attributed to the spin phonon interactions resulting from the\nspin-Hall effect mediated spin polarization. We propose that the spin\npolarization leads to a decrease in hole-phonon scattering resulting in\nenhanced negative magnetoresistance.\n",
"title": "Spin mediated enhanced negative magnetoresistance in Ni80Fe20 and p-silicon bilayer"
} | null | null | [
"Physics"
]
| null | true | null | 631 | null | Validated | null | null |
null | {
"abstract": " Recent work on the representation of functions on sets has considered the use\nof summation in a latent space to enforce permutation invariance. In\nparticular, it has been conjectured that the dimension of this latent space may\nremain fixed as the cardinality of the sets under consideration increases.\nHowever, we demonstrate that the analysis leading to this conjecture requires\nmappings which are highly discontinuous and argue that this is only of limited\npractical use. Motivated by this observation, we prove that an implementation\nof this model via continuous mappings (as provided by e.g. neural networks or\nGaussian processes) actually imposes a constraint on the dimensionality of the\nlatent space. Practical universal function representation for set inputs can\nonly be achieved with a latent dimension at least the size of the maximum\nnumber of input elements.\n",
"title": "On the Limitations of Representing Functions on Sets"
} | null | null | null | null | true | null | 632 | null | Default | null | null |
null | {
"abstract": " Measurement error in observational datasets can lead to systematic bias in\ninferences based on these datasets. As studies based on observational data are\nincreasingly used to inform decisions with real-world impact, it is critical\nthat we develop a robust set of techniques for analyzing and adjusting for\nthese biases. In this paper we present a method for estimating the distribution\nof an outcome given a binary exposure that is subject to underreporting. Our\nmethod is based on a missing data view of the measurement error problem, where\nthe true exposure is treated as a latent variable that is marginalized out of a\njoint model. We prove three different conditions under which the outcome\ndistribution can still be identified from data containing only error-prone\nobservations of the exposure. We demonstrate this method on synthetic data and\nanalyze its sensitivity to near violations of the identifiability conditions.\nFinally, we use this method to estimate the effects of maternal smoking and\nopioid use during pregnancy on childhood obesity, two import problems from\npublic health. Using the proposed method, we estimate these effects using only\nsubject-reported drug use data and substantially refine the range of estimates\ngenerated by a sensitivity analysis-based approach. Further, the estimates\nproduced by our method are consistent with existing literature on both the\neffects of maternal smoking and the rate at which subjects underreport smoking.\n",
"title": "Learning Models from Data with Measurement Error: Tackling Underreporting"
} | null | null | null | null | true | null | 633 | null | Default | null | null |
null | {
"abstract": " An extremely simple, description of Karmarkar's algorithm with very few\ntechnical terms is given.\n",
"title": "A simple introduction to Karmarkar's Algorithm for Linear Programming"
} | null | null | null | null | true | null | 634 | null | Default | null | null |
null | {
"abstract": " We consider a wireless sensor network that uses inductive near-field coupling\nfor wireless powering or communication, or for both. The severely limited range\nof an inductively coupled source-destination pair can be improved using\nresonant relay devices, which are purely passive in nature. Utilization of such\nmagneto-inductive relays has only been studied for regular network topologies,\nallowing simplified assumptions on the mutual antenna couplings. In this work\nwe present an analysis of magneto-inductive passive relaying in arbitrarily\narranged networks. We find that the resulting channel has characteristics\nsimilar to multipath fading: the channel power gain is governed by a\nnon-coherent sum of phasors, resulting in increased frequency selectivity. We\npropose and study two strategies to increase the channel power gain of random\nrelay networks: i) deactivation of individual relays by open-circuit switching\nand ii) frequency tuning. The presented results show that both methods improve\nthe utilization of available passive relays, leading to reliable and\nsignificant performance gains.\n",
"title": "Magneto-inductive Passive Relaying in Arbitrarily Arranged Networks"
} | null | null | [
"Computer Science"
]
| null | true | null | 635 | null | Validated | null | null |
null | {
"abstract": " Transfer operators such as the Perron--Frobenius or Koopman operator play an\nimportant role in the global analysis of complex dynamical systems. The\neigenfunctions of these operators can be used to detect metastable sets, to\nproject the dynamics onto the dominant slow processes, or to separate\nsuperimposed signals. We extend transfer operator theory to reproducing kernel\nHilbert spaces and show that these operators are related to Hilbert space\nrepresentations of conditional distributions, known as conditional mean\nembeddings in the machine learning community. Moreover, numerical methods to\ncompute empirical estimates of these embeddings are akin to data-driven methods\nfor the approximation of transfer operators such as extended dynamic mode\ndecomposition and its variants. One main benefit of the presented kernel-based\napproaches is that these methods can be applied to any domain where a\nsimilarity measure given by a kernel is available. We illustrate the results\nwith the aid of guiding examples and highlight potential applications in\nmolecular dynamics as well as video and text data analysis.\n",
"title": "Eigendecompositions of Transfer Operators in Reproducing Kernel Hilbert Spaces"
} | null | null | null | null | true | null | 636 | null | Default | null | null |
null | {
"abstract": " In this paper we consider the three-dimensional Schrödinger operator with\na $\\delta$-interaction of strength $\\alpha > 0$ supported on an unbounded\nsurface parametrized by the mapping $\\mathbb{R}^2\\ni x\\mapsto (x,\\beta f(x))$,\nwhere $\\beta \\in [0,\\infty)$ and $f\\colon \\mathbb{R}^2\\rightarrow\\mathbb{R}$,\n$f\\not\\equiv 0$, is a $C^2$-smooth, compactly supported function. The surface\nsupporting the interaction can be viewed as a local deformation of the plane.\nIt is known that the essential spectrum of this Schrödinger operator\ncoincides with $[-\\frac14\\alpha^2,+\\infty)$. We prove that for all sufficiently\nsmall $\\beta > 0$ its discrete spectrum is non-empty and consists of a unique\nsimple eigenvalue. Moreover, we obtain an asymptotic expansion of this\neigenvalue in the limit $\\beta \\rightarrow 0+$. In particular, this eigenvalue\ntends to $-\\frac14\\alpha^2$ exponentially fast as $\\beta\\rightarrow 0+$.\n",
"title": "Asymptotics of the bound state induced by $δ$-interaction supported on a weakly deformed plane"
} | null | null | null | null | true | null | 637 | null | Default | null | null |
null | {
"abstract": " In this paper, a comparative study was conducted between complex networks\nrepresenting origin and destination survey data. Similarities were found\nbetween the characteristics of the networks of Brazilian cities with networks\nof foreign cities. Power laws were found in the distributions of edge weights\nand this scale - free behavior can occur due to the economic characteristics of\nthe cities.\n",
"title": "Análise comparativa de pesquisas de origens e destinos: uma abordagem baseada em Redes Complexas"
} | null | null | null | null | true | null | 638 | null | Default | null | null |
null | {
"abstract": " Tension-network (`tensegrity') robots encounter many control challenges as\narticulated soft robots, due to the structures' high-dimensional nonlinear\ndynamics. Control approaches have been developed which use the inverse\nkinematics of tensegrity structures, either for open-loop control or as\nequilibrium inputs for closed-loop controllers. However, current formulations\nof the tensegrity inverse kinematics problem are limited in robotics\napplications: first, they can lead to higher than needed cable tensions, and\nsecond, may lack solutions when applied to robots with high node-to-cable\nratios. This work provides progress in both directions. To address the first\nlimitation, the objective function for the inverse kinematics optimization\nproblem is modified to produce cable tensions as low or lower than before, thus\nreducing the load on the robots' motors. For the second, a reformulation of the\nstatic equilibrium constraint is proposed, which produces solutions independent\nof the number of nodes within each rigid body. Simulation results using the\nsecond reformulation on a specific tensegrity spine robot show reasonable\nopen-loop control results, whereas the previous formulation could not produce\nany solution.\n",
"title": "Inverse Kinematics for Control of Tensegrity Soft Robots: Existence and Optimality of Solutions"
} | null | null | null | null | true | null | 639 | null | Default | null | null |
null | {
"abstract": " The statistical behaviour of the smallest eigenvalue has important\nimplications for systems which can be modeled using a Wishart-Laguerre\nensemble, the regular one or the fixed trace one. For example, the density of\nthe smallest eigenvalue of the Wishart-Laguerre ensemble plays a crucial role\nin characterizing multiple channel telecommunication systems. Similarly, in the\nquantum entanglement problem, the smallest eigenvalue of the fixed trace\nensemble carries information regarding the nature of entanglement.\nFor real Wishart-Laguerre matrices, there exists an elegant recurrence scheme\nsuggested by Edelman to directly obtain the exact expression for the smallest\neigenvalue density. In the case of complex Wishart-Laguerre matrices, for\nfinding exact and explicit expressions for the smallest eigenvalue density,\nexisting results based on determinants become impractical when the determinants\ninvolve large-size matrices. In this work, we derive a recurrence scheme for\nthe complex case which is analogous to that of Edelman's for the real case.\nThis is used to obtain exact results for the smallest eigenvalue density for\nboth the regular, and the fixed trace complex Wishart-Laguerre ensembles. We\nvalidate our analytical results using Monte Carlo simulations. We also study\nscaled Wishart-Laguerre ensemble and investigate its efficacy in approximating\nthe fixed-trace ensemble. Eventually, we apply our result for the fixed-trace\nensemble to investigate the behaviour of the smallest eigenvalue in the\nparadigmatic system of coupled kicked tops.\n",
"title": "Smallest eigenvalue density for regular or fixed-trace complex Wishart-Laguerre ensemble and entanglement in coupled kicked tops"
} | null | null | [
"Physics",
"Mathematics",
"Statistics"
]
| null | true | null | 640 | null | Validated | null | null |
null | {
"abstract": " Dam breach models are commonly used to predict outflow hydrographs of\npotentially failing dams and are key ingredients for evaluating flood risk. In\nthis paper a new dam breach modeling framework is introduced that shall improve\nthe reliability of hydrograph predictions of homogeneous earthen embankment\ndams. Striving for a small number of parameters, the simplified physics-based\nmodel describes the processes of failing embankment dams by breach enlargement,\ndriven by progressive surface erosion. Therein the erosion rate of dam material\nis modeled by empirical sediment transport formulations. Embedding the model\ninto a Bayesian multilevel framework allows for quantitative analysis of\ndifferent categories of uncertainties. To this end, data available in\nliterature of observed peak discharge and final breach width of historical dam\nfailures was used to perform model inversion by applying Markov Chain Monte\nCarlo simulation. Prior knowledge is mainly based on non-informative\ndistribution functions. The resulting posterior distribution shows that the\nmain source of uncertainty is a correlated subset of parameters, consisting of\nthe residual error term and the epistemic term quantifying the breach erosion\nrate. The prediction intervals of peak discharge and final breach width are\ncongruent with values known from literature. To finally predict the outflow\nhydrograph for real case applications, an alternative residual model was\nformulated that assumes perfect data and a perfect model. The fully\nprobabilistic fashion of hydrograph prediction has the potential to improve the\nadequate risk management of downstream flooding.\n",
"title": "Development of probabilistic dam breach model using Bayesian inference"
} | null | null | null | null | true | null | 641 | null | Default | null | null |
null | {
"abstract": " We study the near-infrared properties of 690 Mira candidates in the central\nregion of the Large Magellanic Cloud, based on time-series observations at\nJHKs. We use densely-sampled I-band observations from the OGLE project to\ngenerate template light curves in the near infrared and derive robust mean\nmagnitudes at those wavelengths. We obtain near-infrared Period-Luminosity\nrelations for Oxygen-rich Miras with a scatter as low as 0.12 mag at Ks. We\nstudy the Period-Luminosity-Color relations and the color excesses of\nCarbon-rich Miras, which show evidence for a substantially different reddening\nlaw.\n",
"title": "Large Magellanic Cloud Near-Infrared Synoptic Survey. V. Period-Luminosity Relations of Miras"
} | null | null | null | null | true | null | 642 | null | Default | null | null |
null | {
"abstract": " There is an inherent need for autonomous cars, drones, and other robots to\nhave a notion of how their environment behaves and to anticipate changes in the\nnear future. In this work, we focus on anticipating future appearance given the\ncurrent frame of a video. Existing work focuses on either predicting the future\nappearance as the next frame of a video, or predicting future motion as optical\nflow or motion trajectories starting from a single video frame. This work\nstretches the ability of CNNs (Convolutional Neural Networks) to predict an\nanticipation of appearance at an arbitrarily given future time, not necessarily\nthe next video frame. We condition our predicted future appearance on a\ncontinuous time variable that allows us to anticipate future frames at a given\ntemporal distance, directly from the input video frame. We show that CNNs can\nlearn an intrinsic representation of typical appearance changes over time and\nsuccessfully generate realistic predictions at a deliberate time difference in\nthe near future.\n",
"title": "One-Step Time-Dependent Future Video Frame Prediction with a Convolutional Encoder-Decoder Neural Network"
} | null | null | null | null | true | null | 643 | null | Default | null | null |
null | {
"abstract": " We consider the problem of dynamic spectrum access for network utility\nmaximization in multichannel wireless networks. The shared bandwidth is divided\ninto K orthogonal channels. In the beginning of each time slot, each user\nselects a channel and transmits a packet with a certain transmission\nprobability. After each time slot, each user that has transmitted a packet\nreceives a local observation indicating whether its packet was successfully\ndelivered or not (i.e., ACK signal). The objective is a multi-user strategy for\naccessing the spectrum that maximizes a certain network utility in a\ndistributed manner without online coordination or message exchanges between\nusers. Obtaining an optimal solution for the spectrum access problem is\ncomputationally expensive in general due to the large state space and partial\nobservability of the states. To tackle this problem, we develop a novel\ndistributed dynamic spectrum access algorithm based on deep multi-user\nreinforcement leaning. Specifically, at each time slot, each user maps its\ncurrent state to spectrum access actions based on a trained deep-Q network used\nto maximize the objective function. Game theoretic analysis of the system\ndynamics is developed for establishing design principles for the implementation\nof the algorithm. Experimental results demonstrate strong performance of the\nalgorithm.\n",
"title": "Deep Multi-User Reinforcement Learning for Distributed Dynamic Spectrum Access"
} | null | null | [
"Computer Science"
]
| null | true | null | 644 | null | Validated | null | null |
null | {
"abstract": " We design a new myopic strategy for a wide class of sequential design of\nexperiment (DOE) problems, where the goal is to collect data in order to to\nfulfil a certain problem specific goal. Our approach, Myopic Posterior Sampling\n(MPS), is inspired by the classical posterior (Thompson) sampling algorithm for\nmulti-armed bandits and leverages the flexibility of probabilistic programming\nand approximate Bayesian inference to address a broad set of problems.\nEmpirically, this general-purpose strategy is competitive with more specialised\nmethods in a wide array of DOE tasks, and more importantly, enables addressing\ncomplex DOE goals where no existing method seems applicable. On the theoretical\nside, we leverage ideas from adaptive submodularity and reinforcement learning\nto derive conditions under which MPS achieves sublinear regret against natural\nbenchmark policies.\n",
"title": "Myopic Bayesian Design of Experiments via Posterior Sampling and Probabilistic Programming"
} | null | null | null | null | true | null | 645 | null | Default | null | null |
null | {
"abstract": " Deep convolutional neural networks have liberated its extraordinary power on\nvarious tasks. However, it is still very challenging to deploy state-of-the-art\nmodels into real-world applications due to their high computational complexity.\nHow can we design a compact and effective network without massive experiments\nand expert knowledge? In this paper, we propose a simple and effective\nframework to learn and prune deep models in an end-to-end manner. In our\nframework, a new type of parameter -- scaling factor is first introduced to\nscale the outputs of specific structures, such as neurons, groups or residual\nblocks. Then we add sparsity regularizations on these factors, and solve this\noptimization problem by a modified stochastic Accelerated Proximal Gradient\n(APG) method. By forcing some of the factors to zero, we can safely remove the\ncorresponding structures, thus prune the unimportant parts of a CNN. Comparing\nwith other structure selection methods that may need thousands of trials or\niterative fine-tuning, our method is trained fully end-to-end in one training\npass without bells and whistles. We evaluate our method, Sparse Structure\nSelection with several state-of-the-art CNNs, and demonstrate very promising\nresults with adaptive depth and width selection.\n",
"title": "Data-Driven Sparse Structure Selection for Deep Neural Networks"
} | null | null | null | null | true | null | 646 | null | Default | null | null |
null | {
"abstract": " Numerical simulations of the G.O. Roberts dynamo are presented. Dynamos both\nwith and without a significant mean field are obtained. Exact bounds are\nderived for the total energy which conform with the Kolmogorov phenomenology of\nturbulence. Best fits to numerical data show the same functional dependences as\nthe inequalities obtained from optimum theory.\n",
"title": "Scaling laws and bounds for the turbulent G.O. Roberts dynamo"
} | null | null | [
"Physics"
]
| null | true | null | 647 | null | Validated | null | null |
null | {
"abstract": " Using a projection-based decoupling of the Fokker-Planck equation, control\nstrategies that allow to speed up the convergence to the stationary\ndistribution are investigated. By means of an operator theoretic framework for\na bilinear control system, two different feedback control laws are proposed.\nProjected Riccati and Lyapunov equations are derived and properties of the\nassociated solutions are given. The well-posedness of the closed loop systems\nis shown and local and global stabilization results, respectively, are\nobtained. An essential tool in the construction of the controls is the choice\nof appropriate control shape functions. Results for a two dimensional double\nwell potential illustrate the theoretical findings in a numerical setup.\n",
"title": "Control Strategies for the Fokker-Planck Equation"
} | null | null | null | null | true | null | 648 | null | Default | null | null |
null | {
"abstract": " We offer a generalization of a formula of Popov involving the Von Mangoldt\nfunction. Some commentary on its relation to other results in analytic number\ntheory is mentioned as well as an analogue involving the m$\\ddot{o}$bius\nfunction.\n",
"title": "On Popov's formula involving the Von Mangoldt function"
} | null | null | [
"Mathematics"
]
| null | true | null | 649 | null | Validated | null | null |
null | {
"abstract": " In this paper, we show that any compact manifold that carries a\nSL(n;R)-foliation is fibered on the circle S^1.\n",
"title": "On fibering compact manifold over the circle"
} | null | null | null | null | true | null | 650 | null | Default | null | null |
null | {
"abstract": " Given the importance of crystal symmetry for the emergence of topological\nquantum states, we have studied, as exemplified in NbNiTe2, the interplay of\ncrystal symmetry, atomic displacements (lattice vibration), band degeneracy,\nand band topology. For NbNiTe2 structure in space group 53 (Pmna) - having an\ninversion center arising from two glide planes and one mirror plane with a\n2-fold rotation and screw axis - a full gap opening exists between two band\nmanifolds near the Fermi energy. Upon atomic displacements by optical phonons,\nthe symmetry lowers to space group 28 (Pma2), eliminating one glide plane along\nc, the associated rotation and screw axis, and the inversion center. As a\nresult, twenty Weyl points emerge, including four type-II Weyl points in the\nG-X direction at the boundary between a pair of adjacent electron and hole\nbands. Thus, optical phonons may offer control of the transition to a Weyl\nfermion state.\n",
"title": "Phonon-Induced Topological Transition to a Type-II Weyl Semimetal"
} | null | null | [
"Physics"
]
| null | true | null | 651 | null | Validated | null | null |
null | {
"abstract": " Several social, medical, engineering and biological challenges rely on\ndiscovering the functionality of networks from their structure and node\nmetadata, when it is available. For example, in chemoinformatics one might want\nto detect whether a molecule is toxic based on structure and atomic types, or\ndiscover the research field of a scientific collaboration network. Existing\ntechniques rely on counting or measuring structural patterns that are known to\nshow large variations from network to network, such as the number of triangles,\nor the assortativity of node metadata. We introduce the concept of multi-hop\nassortativity, that captures the similarity of the nodes situated at the\nextremities of a randomly selected path of a given length. We show that\nmulti-hop assortativity unifies various existing concepts and offers a\nversatile family of 'fingerprints' to characterize networks. These fingerprints\nallow in turn to recover the functionalities of a network, with the help of the\nmachine learning toolbox. Our method is evaluated empirically on established\nsocial and chemoinformatic network benchmarks. Results reveal that our\nassortativity based features are competitive providing highly accurate results\noften outperforming state of the art methods for the network classification\ntask.\n",
"title": "Multi-hop assortativities for networks classification"
} | null | null | null | null | true | null | 652 | null | Default | null | null |
null | {
"abstract": " This paper is concerned with the online estimation of a nonlinear dynamic\nsystem from a series of noisy measurements. The focus is on cases wherein\noutliers are present in-between normal noises. We assume that the outliers\nfollow an unknown generating mechanism which deviates from that of normal\nnoises, and then model the outliers using a Bayesian nonparametric model called\nDirichlet process mixture (DPM). A sequential particle-based algorithm is\nderived for posterior inference for the outlier model as well as the state of\nthe system to be estimated. The resulting algorithm is termed DPM based robust\nPF (DPM-RPF). The nonparametric feature makes this algorithm allow the data to\n\"speak for itself\" to determine the complexity and structure of the outlier\nmodel. Simulation results show that it performs remarkably better than two\nstate-of-the-art methods especially when outliers appear frequently along time.\n",
"title": "A Bayesian Nonparametrics based Robust Particle Filter Algorithm"
} | null | null | null | null | true | null | 653 | null | Default | null | null |
null | {
"abstract": " Computed tomography (CT) examinations are commonly used to predict lung\nnodule malignancy in patients, which are shown to improve noninvasive early\ndiagnosis of lung cancer. It remains challenging for computational approaches\nto achieve performance comparable to experienced radiologists. Here we present\nNoduleX, a systematic approach to predict lung nodule malignancy from CT data,\nbased on deep learning convolutional neural networks (CNN). For training and\nvalidation, we analyze >1000 lung nodules in images from the LIDC/IDRI cohort.\nAll nodules were identified and classified by four experienced thoracic\nradiologists who participated in the LIDC project. NoduleX achieves high\naccuracy for nodule malignancy classification, with an AUC of ~0.99. This is\ncommensurate with the analysis of the dataset by experienced radiologists. Our\napproach, NoduleX, provides an effective framework for highly accurate nodule\nmalignancy prediction with the model trained on a large patient population. Our\nresults are replicable with software available at\nthis http URL.\n",
"title": "Highly accurate model for prediction of lung nodule malignancy with CT scans"
} | null | null | null | null | true | null | 654 | null | Default | null | null |
null | {
"abstract": " The turbulent Rayleigh--Taylor system in a rotating reference frame is\ninvestigated by direct numerical simulations within the Oberbeck-Boussinesq\napproximation. On the basis of theoretical arguments, supported by our\nsimulations, we show that the Rossby number decreases in time, and therefore\nthe Coriolis force becomes more important as the system evolves and produces\nmany effects on Rayleigh--Taylor turbulence. We find that rotation reduces the\nintensity of turbulent velocity fluctuations and therefore the growth rate of\nthe temperature mixing layer. Moreover, in presence of rotation the conversion\nof potential energy into turbulent kinetic energy is found to be less effective\nand the efficiency of the heat transfer is reduced. Finally, during the\nevolution of the mixing layer we observe the development of a\ncyclone-anticyclone asymmetry.\n",
"title": "Rotating Rayleigh-Taylor turbulence"
} | null | null | [
"Physics"
]
| null | true | null | 655 | null | Validated | null | null |
null | {
"abstract": " In the animal world, the competition between individuals belonging to\ndifferent species for a resource often requires the cooperation of several\nindividuals in groups. This paper proposes a generalization of the Hawk-Dove\nGame for an arbitrary number of agents: the N-person Hawk-Dove Game. In this\nmodel, doves exemplify the cooperative behavior without intraspecies conflict,\nwhile hawks represent the aggressive behavior. In the absence of hawks, doves\nshare the resource equally and avoid conflict, but having hawks around lead to\ndoves escaping without fighting. Conversely, hawks fight for the resource at\nthe cost of getting injured. Nevertheless, if doves are present in sufficient\nnumber to expel the hawks, they can aggregate to protect the resource, and thus\navoid being plundered by hawks. We derive and numerically solve an exact\nequation for the evolution of the system in both finite and infinite well-mixed\npopulations, finding the conditions for stable coexistence between both\nspecies. Furthermore, by varying the different parameters, we found a scenario\nof bifurcations that leads the system from dominating hawks and coexistence to\nbi-stability, multiple interior equilibria and dominating doves.\n",
"title": "Evolutionary dynamics of N-person Hawk-Dove games"
} | null | null | null | null | true | null | 656 | null | Default | null | null |
null | {
"abstract": " With approximately half of the world's population at risk of contracting\ndengue, this mosquito-borne disease is of global concern. International\ntravellers significantly contribute to dengue's rapid and large-scale spread by\nimporting the disease from endemic into non-endemic countries. To prevent\nfuture outbreaks and dengue from establishing in non-endemic countries,\nknowledge about the arrival time and location of infected travellers is\ncrucial. We propose a network model that predicts the monthly number of dengue\ninfected air passengers arriving at any given airport. We consider\ninternational air travel volumes, monthly dengue incidence rates and temporal\ninfection dynamics. Our findings shed light onto dengue importation routes and\nreveal country-specific reporting rates that have been until now largely\nunknown.\n",
"title": "A global model for predicting the arrival of imported dengue infections"
} | null | null | null | null | true | null | 657 | null | Default | null | null |
null | {
"abstract": " The best summary of a long video differs among different people due to its\nhighly subjective nature. Even for the same person, the best summary may change\nwith time or mood. In this paper, we introduce the task of generating\ncustomized video summaries through simple text. First, we train a deep\narchitecture to effectively learn semantic embeddings of video frames by\nleveraging the abundance of image-caption data via a progressive and residual\nmanner. Given a user-specific text description, our algorithm is able to select\nsemantically relevant video segments and produce a temporally aligned video\nsummary. In order to evaluate our textually customized video summaries, we\nconduct experimental comparison with baseline methods that utilize ground-truth\ninformation. Despite the challenging baselines, our method still manages to\nshow comparable or even exceeding performance. We also show that our method is\nable to generate semantically diverse video summaries by only utilizing the\nlearned visual embeddings.\n",
"title": "Contextually Customized Video Summaries via Natural Language"
} | null | null | null | null | true | null | 658 | null | Default | null | null |
null | {
"abstract": " Recently, heavily doped semiconductors are emerging as an alternate for low\nloss plasmonic materials. InN, belonging to the group III nitrides, possesses\nthe unique property of surface electron accumulation (SEA) which provides two\ndimensional electron gas (2DEG) system. In this report, we demonstrated the\nsurface plasmon properties of InN nanoparticles originating from SEA using the\nreal space mapping of the surface plasmon fields for the first time. The SEA is\nconfirmed by Raman studies which are further corroborated by photoluminescence\nand photoemission spectroscopic studies. The frequency of 2DEG corresponding to\nSEA is found to be in the THz region. The periodic fringes are observed in the\nnear-field scanning optical microscopic images of InN nanostructures. The\nobserved fringes are attributed to the interference of propagated and back\nreflected surface plasmon polaritons (SPPs). The observation of SPPs is solely\nattributed to the 2DEG corresponding to the SEA of InN. In addition, resonance\nkind of behavior with the enhancement of the near-field intensity is observed\nin the near-field images of InN nanostructures. Observation of SPPs indicates\nthat InN with SEA can be a promising THz plasmonic material for the light\nconfinement.\n",
"title": "Observation of surface plasmon polaritons in 2D electron gas of surface electron accumulation in InN nanostructures"
} | null | null | null | null | true | null | 659 | null | Default | null | null |
null | {
"abstract": " It is shown that using beam splitters with non-equal wave vectors results in\na new recoil diagram which is qualitatively different from the well-known\ndiagram associated with the Mach-Zehnder atom interferometer. We predict a new\nasymmetric Mach-Zehnder atom interferometer (AMZAI) and study it when one uses\na Raman beam splitter. The main feature is that the phase of AMZAI contains a\nquantum part proportional to the recoil frequency. A response sensitive only to\nthe quantum phase was found. A new technique to measure the recoil frequency\nand fine structure constant is proposed and studied outside of the Raman-Nath\napproximation.\n",
"title": "Asymmetric Mach-Zehnder atom interferometers"
} | null | null | null | null | true | null | 660 | null | Default | null | null |
null | {
"abstract": " In this paper, we consider a partial information two-person zero-sum\nstochastic differential game problem where the system is governed by a backward\nstochastic differential equation driven by Teugels martingales associated with\na Lévy process and an independent Brownian motion. One sufficient (a\nverification theorem) and one necessary conditions for the existence of optimal\ncontrols are proved. To illustrate the general results, a linear quadratic\nstochastic differential game problem is discussed.\n",
"title": "Partial Information Stochastic Differential Games for Backward Stochastic Systems Driven By Lévy Processes"
} | null | null | null | null | true | null | 661 | null | Default | null | null |
null | {
"abstract": " In recent years, research has been done on applying Recurrent Neural Networks\n(RNNs) as recommender systems. Results have been promising, especially in the\nsession-based setting where RNNs have been shown to outperform state-of-the-art\nmodels. In many of these experiments, the RNN could potentially improve the\nrecommendations by utilizing information about the user's past sessions, in\naddition to its own interactions in the current session. A problem for\nsession-based recommendation, is how to produce accurate recommendations at the\nstart of a session, before the system has learned much about the user's current\ninterests. We propose a novel approach that extends a RNN recommender to be\nable to process the user's recent sessions, in order to improve\nrecommendations. This is done by using a second RNN to learn from recent\nsessions, and predict the user's interest in the current session. By feeding\nthis information to the original RNN, it is able to improve its\nrecommendations. Our experiments on two different datasets show that the\nproposed approach can significantly improve recommendations throughout the\nsessions, compared to a single RNN working only on the current session. The\nproposed model especially improves recommendations at the start of sessions,\nand is therefore able to deal with the cold start problem within sessions.\n",
"title": "Inter-Session Modeling for Session-Based Recommendation"
} | null | null | null | null | true | null | 662 | null | Default | null | null |
null | {
"abstract": " Shock wave interactions with defects, such as pores, are known to play a key\nrole in the chemical initiation of energetic materials. The shock response of\nhexanitrostilbene is studied through a combination of large scale reactive\nmolecular dynamics and mesoscale hydrodynamic simulations. In order to extend\nour simulation capability at the mesoscale to include weak shock conditions (<\n6 GPa), atomistic simulations of pore collapse are used to define a strain rate\ndependent strength model. Comparing these simulation methods allows us to\nimpose physically-reasonable constraints on the mesoscale model parameters. In\ndoing so, we have been able to study shock waves interacting with pores as a\nfunction of this viscoplastic material response. We find that the pore collapse\nbehavior of weak shocks is characteristically different to that of strong\nshocks.\n",
"title": "Multiscale Modeling of Shock Wave Localization in Porous Energetic Material"
} | null | null | null | null | true | null | 663 | null | Default | null | null |
null | {
"abstract": " We propose factor models for the cross-section of daily cryptoasset returns\nand provide source code for data downloads, computing risk factors and\nbacktesting them out-of-sample. In \"cryptoassets\" we include all\ncryptocurrencies and a host of various other digital assets (coins and tokens)\nfor which exchange market data is available. Based on our empirical analysis,\nwe identify the leading factor that appears to strongly contribute into daily\ncryptoasset returns. Our results suggest that cross-sectional statistical\narbitrage trading may be possible for cryptoassets subject to efficient\nexecutions and shorting.\n",
"title": "Cryptoasset Factor Models"
} | null | null | [
"Quantitative Finance"
]
| null | true | null | 664 | null | Validated | null | null |
null | {
"abstract": " Many signals on Cartesian product graphs appear in the real world, such as\ndigital images, sensor observation time series, and movie ratings on Netflix.\nThese signals are \"multi-dimensional\" and have directional characteristics\nalong each factor graph. However, the existing graph Fourier transform does not\ndistinguish these directions, and assigns 1-D spectra to signals on product\ngraphs. Further, these spectra are often multi-valued at some frequencies. Our\nmain result is a multi-dimensional graph Fourier transform that solves such\nproblems associated with the conventional GFT. Using algebraic properties of\nCartesian products, the proposed transform rearranges 1-D spectra obtained by\nthe conventional GFT into the multi-dimensional frequency domain, of which each\ndimension represents a directional frequency along each factor graph. Thus, the\nmulti-dimensional graph Fourier transform enables directional frequency\nanalysis, in addition to frequency analysis with the conventional GFT.\nMoreover, this rearrangement resolves the multi-valuedness of spectra in some\ncases. The multi-dimensional graph Fourier transform is a foundation of novel\nfilterings and stationarities that utilize dimensional information of graph\nsignals, which are also discussed in this study. The proposed methods are\napplicable to a wide variety of data that can be regarded as signals on\nCartesian product graphs. This study also notes that multivariate graph signals\ncan be regarded as 2-D univariate graph signals. This correspondence provides\nnatural definitions of the multivariate graph Fourier transform and the\nmultivariate stationarity based on their 2-D univariate versions.\n",
"title": "Multi-dimensional Graph Fourier Transform"
} | null | null | null | null | true | null | 665 | null | Default | null | null |
null | {
"abstract": " The double exponential formula was introduced for calculating definite\nintegrals with singular point oscillation functions and Fourier-integrals. The\ndouble exponential transformation is not only useful for numerical computations\nbut it is also used in different methods of Sinc theory. In this paper we use\ndouble exponential transformation for calculating particular improper\nintegrals. By improving integral estimates having singular final points. By\ncomparison between double exponential transformations and single exponential\ntransformations it is proved that the error margin of double exponential\ntransformations is smaller. Finally Fourier-integral and double exponential\ntransformations are discussed.\n",
"title": "Criteria for the Application of Double Exponential Transformation"
} | null | null | null | null | true | null | 666 | null | Default | null | null |
null | {
"abstract": " Strain engineering has attracted great attention, particularly for epitaxial\nfilms grown on a different substrate. Residual strains of SiC have been widely\nemployed to form ultra-high frequency and high Q factor resonators. However, to\ndate the highest residual strain of SiC was reported to be limited to\napproximately 0.6%. Large strains induced into SiC could lead to several\ninteresting physical phenomena, as well as significant improvement of resonant\nfrequencies. We report an unprecedented nano strain-amplifier structure with an\nultra-high residual strain up to 8% utilizing the natural residual stress\nbetween epitaxial 3C SiC and Si. In addition, the applied strain can be tuned\nby changing the dimensions of the amplifier structure. The possibility of\nintroducing such a controllable and ultra-high strain will open the door to\ninvestigating the physics of SiC in large strain regimes, and the development\nof ultra sensitive mechanical sensors.\n",
"title": "Ultra-high strain in epitaxial silicon carbide nanostructures utilizing residual stress amplification"
} | null | null | null | null | true | null | 667 | null | Default | null | null |
null | {
"abstract": " A complex system can be represented and analyzed as a network, where nodes\nrepresent the units of the network and edges represent connections between\nthose units. For example, a brain network represents neurons as nodes and axons\nbetween neurons as edges. In many networks, some nodes have a\ndisproportionately high number of edges. These nodes also have many edges\nbetween each other, and are referred to as the rich club. In many different\nnetworks, the nodes of this club are assumed to support global network\nintegration. However, another set of nodes potentially exhibits a connectivity\nstructure that is more advantageous to global network integration. Here, in a\nmyriad of different biological and man-made networks, we discover the diverse\nclub--a set of nodes that have edges diversely distributed across the network.\nThe diverse club exhibits, to a greater extent than the rich club, properties\nconsistent with an integrative network function--these nodes are more highly\ninterconnected and their edges are more critical for efficient global\nintegration. Moreover, we present a generative evolutionary network model that\nproduces networks with a diverse club but not a rich club, thus demonstrating\nthat these two clubs potentially evolved via distinct selection pressures.\nGiven the variety of different networks that we analyzed--the c. elegans, the\nmacaque brain, the human brain, the United States power grid, and global air\ntraffic--the diverse club appears to be ubiquitous in complex networks. These\nresults warrant the distinction and analysis of two critical clubs of nodes in\nall complex systems.\n",
"title": "The Diverse Club: The Integrative Core of Complex Networks"
} | null | null | [
"Physics"
]
| null | true | null | 668 | null | Validated | null | null |
null | {
"abstract": " Neural network based generative models with discriminative components are a\npowerful approach for semi-supervised learning. However, these techniques a)\ncannot account for model uncertainty in the estimation of the model's\ndiscriminative component and b) lack flexibility to capture complex stochastic\npatterns in the label generation process. To avoid these problems, we first\npropose to use a discriminative component with stochastic inputs for increased\nnoise flexibility. We show how an efficient Gibbs sampling procedure can\nmarginalize the stochastic inputs when inferring missing labels in this model.\nFollowing this, we extend the discriminative component to be fully Bayesian and\nproduce estimates of uncertainty in its parameter values. This opens the door\nfor semi-supervised Bayesian active learning.\n",
"title": "Bayesian Semisupervised Learning with Deep Generative Models"
} | null | null | null | null | true | null | 669 | null | Default | null | null |
null | {
"abstract": " Detection of interactions between treatment effects and patient descriptors\nin clinical trials is critical for optimizing the drug development process. The\nincreasing volume of data accumulated in clinical trials provides a unique\nopportunity to discover new biomarkers and further the goal of personalized\nmedicine, but it also requires innovative robust biomarker detection methods\ncapable of detecting non-linear, and sometimes weak, signals. We propose a set\nof novel univariate statistical tests, based on the theory of random walks,\nwhich are able to capture non-linear and non-monotonic covariate-treatment\ninteractions. We also propose a novel combined test, which leverages the power\nof all of our proposed univariate tests into a single general-case tool. We\npresent results for both synthetic trials as well as real-world clinical\ntrials, where we compare our method with state-of-the-art techniques and\ndemonstrate the utility and robustness of our approach.\n",
"title": "Robust Detection of Covariate-Treatment Interactions in Clinical Trials"
} | null | null | null | null | true | null | 670 | null | Default | null | null |
null | {
"abstract": " The limitations in performance of the present RICH system in the LHCb\nexperiment are given by the natural chromatic dispersion of the gaseous\nCherenkov radiator, the aberrations of the optical system and the pixel size of\nthe photon detectors. Moreover, the overall PID performance can be affected by\nhigh detector occupancy as the pattern recognition becomes more difficult with\nhigh particle multiplicities. This paper shows a way to improve performance by\nsystematically addressing each of the previously mentioned limitations. These\nideas are applied in the present and future upgrade phases of the LHCb\nexperiment. Although applied to specific circumstances, they are used as a\nparadigm on what is achievable in the development and realisation of high\nprecision RICH detectors.\n",
"title": "The Future of RICH Detectors through the Light of the LHCb RICH"
} | null | null | null | null | true | null | 671 | null | Default | null | null |
null | {
"abstract": " Given a klt singularity $x\\in (X, D)$, we show that a quasi-monomial\nvaluation $v$ with a finitely generated associated graded ring is the minimizer\nof the normalized volume function $\\widehat{\\rm vol}_{(X,D),x}$, if and only if\n$v$ induces a degeneration to a K-semistable log Fano cone singularity.\nMoreover, such a minimizer is unique among all quasi-monomial valuations up to\nrescaling. As a consequence, we prove that for a klt singularity $x\\in X$ on\nthe Gromov-Hausdorff limit of Kähler-Einstein Fano manifolds, the\nintermediate K-semistable cone associated to its metric tangent cone is\nuniquely determined by the algebraic structure of $x\\in X$, hence confirming a\nconjecture by Donaldson-Sun.\n",
"title": "Stability of Valuations: Higher Rational Rank"
} | null | null | null | null | true | null | 672 | null | Default | null | null |
null | {
"abstract": " Condensed-matter analogs of the Higgs boson in particle physics allow\ninsights into its behavior in different symmetries and dimensionalities.\nEvidence for the Higgs mode has been reported in a number of different\nsettings, including ultracold atomic gases, disordered superconductors, and\ndimerized quantum magnets. However, decay processes of the Higgs mode (which\nare eminently important in particle physics) have not yet been studied in\ncondensed matter due to the lack of a suitable material system coupled to a\ndirect experimental probe. A quantitative understanding of these processes is\nparticularly important for low-dimensional systems where the Higgs mode decays\nrapidly and has remained elusive to most experimental probes. Here, we discover\nand study the Higgs mode in a two-dimensional antiferromagnet using\nspin-polarized inelastic neutron scattering. Our spin-wave spectra of\nCa$_2$RuO$_4$ directly reveal a well-defined, dispersive Higgs mode, which\nquickly decays into transverse Goldstone modes at the antiferromagnetic\nordering wavevector. Through a complete mapping of the transverse modes in the\nreciprocal space, we uniquely specify the minimal model Hamiltonian and\ndescribe the decay process. We thus establish a novel condensed matter platform\nfor research on the dynamics of the Higgs mode.\n",
"title": "Higgs mode and its decay in a two dimensional antiferromagnet"
} | null | null | null | null | true | null | 673 | null | Default | null | null |
null | {
"abstract": " Well-known for its simplicity and effectiveness in classification, AdaBoost,\nhowever, suffers from overfitting when class-conditional distributions have\nsignificant overlap. Moreover, it is very sensitive to noise that appears in\nthe labels. This article tackles the above limitations simultaneously via\noptimizing a modified loss function (i.e., the conditional risk). The proposed\napproach has the following two advantages. (1) It is able to directly take into\naccount label uncertainty with an associated label confidence. (2) It\nintroduces a \"trustworthiness\" measure on training samples via the Bayesian\nrisk rule, and hence the resulting classifier tends to have finite sample\nperformance that is superior to that of the original AdaBoost when there is a\nlarge overlap between class conditional distributions. Theoretical properties\nof the proposed method are investigated. Extensive experimental results using\nsynthetic data and real-world data sets from UCI machine learning repository\nare provided. The empirical study shows the high competitiveness of the\nproposed method in predication accuracy and robustness when compared with the\noriginal AdaBoost and several existing robust AdaBoost algorithms.\n",
"title": "Robust and Efficient Boosting Method using the Conditional Risk"
} | null | null | null | null | true | null | 674 | null | Default | null | null |
null | {
"abstract": " We study the problem of sparsity constrained $M$-estimation with arbitrary\ncorruptions to both {\\em explanatory and response} variables in the\nhigh-dimensional regime, where the number of variables $d$ is larger than the\nsample size $n$. Our main contribution is a highly efficient gradient-based\noptimization algorithm that we call Trimmed Hard Thresholding -- a robust\nvariant of Iterative Hard Thresholding (IHT) by using trimmed mean in gradient\ncomputations. Our algorithm can deal with a wide class of sparsity constrained\n$M$-estimation problems, and we can tolerate a nearly dimension independent\nfraction of arbitrarily corrupted samples. More specifically, when the\ncorrupted fraction satisfies $\\epsilon \\lesssim {1} /\\left({\\sqrt{k} \\log\n(nd)}\\right)$, where $k$ is the sparsity of the parameter, we obtain accurate\nestimation and model selection guarantees with optimal sample complexity.\nFurthermore, we extend our algorithm to sparse Gaussian graphical model\n(precision matrix) estimation via a neighborhood selection approach. We\ndemonstrate the effectiveness of robust estimation in sparse linear, logistic\nregression, and sparse precision matrix estimation on synthetic and real-world\nUS equities data.\n",
"title": "High Dimensional Robust Estimation of Sparse Models via Trimmed Hard Thresholding"
} | null | null | null | null | true | null | 675 | null | Default | null | null |
null | {
"abstract": " We present a communication- and data-sensitive formulation of ADER-DG for\nhyperbolic differential equation systems. Sensitive here has multiple flavours:\nFirst, the formulation reduces the persistent memory footprint. This reduces\npressure on the memory subsystem. Second, the formulation realises the\nunderlying predictor-corrector scheme with single-touch semantics, i.e., each\ndegree of freedom is read on average only once per time step from the main\nmemory. This reduces communication through the memory controllers. Third, the\nformulation breaks up the tight coupling of the explicit time stepping's\nalgorithmic steps to mesh traversals. This averages out data access peaks.\nDifferent operations and algorithmic steps are ran on different grid entities.\nFinally, the formulation hides distributed memory data transfer behind the\ncomputation aligned with the mesh traversal. This reduces pressure on the\nmachine interconnects. All techniques applied by our formulation are elaborated\nby means of a rigorous task formalism. They break up ADER-DG's tight causal\ncoupling of compute steps and can be generalised to other predictor-corrector\nschemes.\n",
"title": "Stop talking to me -- a communication-avoiding ADER-DG realisation"
} | null | null | null | null | true | null | 676 | null | Default | null | null |
null | {
"abstract": " This paper outlines a methodology for Bayesian multimodel uncertainty\nquantification (UQ) and propagation and presents an investigation into the\neffect of prior probabilities on the resulting uncertainties. The UQ\nmethodology is adapted from the information-theoretic method previously\npresented by the authors (Zhang and Shields, 2018) to a fully Bayesian\nconstruction that enables greater flexibility in quantifying uncertainty in\nprobability model form. Being Bayesian in nature and rooted in UQ from small\ndatasets, prior probabilities in both probability model form and model\nparameters are shown to have a significant impact on quantified uncertainties\nand, consequently, on the uncertainties propagated through a physics-based\nmodel. These effects are specifically investigated for a simplified plate\nbuckling problem with uncertainties in material properties derived from a small\nnumber of experiments using noninformative priors and priors derived from past\nstudies of varying appropriateness. It is illustrated that prior probabilities\ncan have a significant impact on multimodel UQ for small datasets and\ninappropriate (but seemingly reasonable) priors may even have lingering effects\nthat bias probabilities even for large datasets. When applied to uncertainty\npropagation, this may result in probability bounds on response quantities that\ndo not include the true probabilities.\n",
"title": "The effect of prior probabilities on quantification and propagation of imprecise probabilities resulting from small datasets"
} | null | null | null | null | true | null | 677 | null | Default | null | null |
null | {
"abstract": " Failing to distinguish between a sheepdog and a skyscraper should be worse\nand penalized more than failing to distinguish between a sheepdog and a poodle;\nafter all, sheepdogs and poodles are both breeds of dogs. However, existing\nmetrics of failure (so-called \"loss\" or \"win\") used in textual or visual\nclassification/recognition via neural networks seldom view a sheepdog as more\nsimilar to a poodle than to a skyscraper. We define a metric that, inter alia,\ncan penalize failure to distinguish between a sheepdog and a skyscraper more\nthan failure to distinguish between a sheepdog and a poodle. Unlike previously\nemployed possibilities, this metric is based on an ultrametric tree associated\nwith any given tree organization into a semantically meaningful hierarchy of a\nclassifier's classes.\n",
"title": "Hierarchical loss for classification"
} | null | null | null | null | true | null | 678 | null | Default | null | null |
null | {
"abstract": " Achieving the goals in the title (and others) relies on a cardinality-wise\nscanning of the ideals of the poset. Specifically, the relevant numbers\nattached to the k+1 element ideals are inferred from the corresponding numbers\nof the k-element (order) ideals. Crucial in all of this is a compressed\nrepresentation (using wildcards) of the ideal lattice. The whole scheme invites\ndistributed computation.\n",
"title": "An efficient data structure for counting all linear extensions of a poset, calculating its jump number, and the likes"
} | null | null | null | null | true | null | 679 | null | Default | null | null |
null | {
"abstract": " We present a scalable, black box, perception-in-the-loop technique to find\nadversarial examples for deep neural network classifiers. Black box means that\nour procedure only has input-output access to the classifier, and not to the\ninternal structure, parameters, or intermediate confidence values.\nPerception-in-the-loop means that the notion of proximity between inputs can be\ndirectly queried from human participants rather than an arbitrarily chosen\nmetric. Our technique is based on covariance matrix adaptation evolution\nstrategy (CMA-ES), a black box optimization approach. CMA-ES explores the\nsearch space iteratively in a black box manner, by generating populations of\ncandidates according to a distribution, choosing the best candidates according\nto a cost function, and updating the posterior distribution to favor the best\ncandidates. We run CMA-ES using human participants to provide the fitness\nfunction, using the insight that the choice of best candidates in CMA-ES can be\nnaturally modeled as a perception task: pick the top $k$ inputs perceptually\nclosest to a fixed input. We empirically demonstrate that finding adversarial\nexamples is feasible using small populations and few iterations. We compare the\nperformance of CMA-ES on the MNIST benchmark with other black-box approaches\nusing $L_p$ norms as a cost function, and show that it performs favorably both\nin terms of success in finding adversarial examples and in minimizing the\ndistance between the original and the adversarial input. In experiments on the\nMNIST, CIFAR10, and GTSRB benchmarks, we demonstrate that CMA-ES can find\nperceptually similar adversarial inputs with a small number of iterations and\nsmall population sizes when using perception-in-the-loop. Finally, we show that\nnetworks trained specifically to be robust against $L_\\infty$ norm can still be\nsusceptible to perceptually similar adversarial examples.\n",
"title": "Perception-in-the-Loop Adversarial Examples"
} | null | null | null | null | true | null | 680 | null | Default | null | null |
null | {
"abstract": " This paper presents a novel generative model to synthesize fluid simulations\nfrom a set of reduced parameters. A convolutional neural network is trained on\na collection of discrete, parameterizable fluid simulation velocity fields. Due\nto the capability of deep learning architectures to learn representative\nfeatures of the data, our generative model is able to accurately approximate\nthe training data set, while providing plausible interpolated in-betweens. The\nproposed generative model is optimized for fluids by a novel loss function that\nguarantees divergence-free velocity fields at all times. In addition, we\ndemonstrate that we can handle complex parameterizations in reduced spaces, and\nadvance simulations in time by integrating in the latent space with a second\nnetwork. Our method models a wide variety of fluid behaviors, thus enabling\napplications such as fast construction of simulations, interpolation of fluids\nwith different parameters, time re-sampling, latent space simulations, and\ncompression of fluid simulation data. Reconstructed velocity fields are\ngenerated up to 700x faster than traditional CPU solvers, while achieving\ncompression rates of over 1300x.\n",
"title": "Deep Fluids: A Generative Network for Parameterized Fluid Simulations"
} | null | null | null | null | true | null | 681 | null | Default | null | null |
null | {
"abstract": " The two-stage least-squares (2SLS) estimator is known to be biased when its\nfirst-stage fit is poor. I show that better first-stage prediction can\nalleviate this bias. In a two-stage linear regression model with Normal noise,\nI consider shrinkage in the estimation of the first-stage instrumental variable\ncoefficients. For at least four instrumental variables and a single endogenous\nregressor, I establish that the standard 2SLS estimator is dominated with\nrespect to bias. The dominating IV estimator applies James-Stein type shrinkage\nin a first-stage high-dimensional Normal-means problem followed by a\ncontrol-function approach in the second stage. It preserves invariances of the\nstructural instrumental variable equations.\n",
"title": "Bias Reduction in Instrumental Variable Estimation through First-Stage Shrinkage"
} | null | null | null | null | true | null | 682 | null | Default | null | null |
null | {
"abstract": " An unsupervised learning classification model is described. It achieves\nclassification error probability competitive with that of popular supervised\nlearning classifiers such as SVM or kNN. The model is based on the incremental\nexecution of small step shift and rotation operations upon selected\ndiscriminative hyperplanes at the arrival of input samples. When applied, in\nconjunction with a selected feature extractor, to a subset of the ImageNet\ndataset benchmark, it yields 6.2 % Top 3 probability of error; this exceeds by\nmerely about 2 % the result achieved by (supervised) k-Nearest Neighbor, both\nusing same feature extractor. This result may also be contrasted with popular\nunsupervised learning schemes such as k-Means which is shown to be practically\nuseless on same dataset.\n",
"title": "An Unsupervised Learning Classifier with Competitive Error Performance"
} | null | null | null | null | true | null | 683 | null | Default | null | null |
null | {
"abstract": " We investigate the predictability of several range-based stock volatility\nestimators, and compare them to the standard close-to-close estimator which is\nmost commonly acknowledged as the volatility. The patterns of volatility\nchanges are analyzed using LSTM recurrent neural networks, which are a state of\nthe art method of sequence learning. We implement the analysis on all current\nconstituents of the Dow Jones Industrial Average index, and report averaged\nevaluation results. We find that changes in the values of range-based\nestimators are more predictable than that of the estimator using daily closing\nvalues only.\n",
"title": "Exploring the predictability of range-based volatility estimators using RNNs"
} | null | null | null | null | true | null | 684 | null | Default | null | null |
null | {
"abstract": " Correlated random walks (CRW) have been used for a long time as a null model\nfor animal's random search movement in two dimensions (2D). An increasing\nnumber of studies focus on animals' movement in three dimensions (3D), but the\nkey properties of CRW, such as the way the mean squared displacement is related\nto the path length, are well known only in 1D and 2D. In this paper I derive\nsuch properties for 3D CRW, in a consistent way with the expression of these\nproperties in 2D. This should allow 3D CRW to act as a null model when\nanalyzing actual 3D movements similarly to what is done in 2D\n",
"title": "Mean squared displacement and sinuosity of three-dimensional random search movements"
} | null | null | null | null | true | null | 685 | null | Default | null | null |
null | {
"abstract": " This paper presents a novel context-based approach for pedestrian motion\nprediction in crowded, urban intersections, with the additional flexibility of\nprediction in similar, but new, environments. Previously, Chen et. al. combined\nMarkovian-based and clustering-based approaches to learn motion primitives in a\ngrid-based world and subsequently predict pedestrian trajectories by modeling\nthe transition between learned primitives as a Gaussian Process (GP). This work\nextends that prior approach by incorporating semantic features from the\nenvironment (relative distance to curbside and status of pedestrian traffic\nlights) in the GP formulation for more accurate predictions of pedestrian\ntrajectories over the same timescale. We evaluate the new approach on\nreal-world data collected using one of the vehicles in the MIT Mobility On\nDemand fleet. The results show 12.5% improvement in prediction accuracy and a\n2.65 times reduction in Area Under the Curve (AUC), which is used as a metric\nto quantify the span of predicted set of trajectories, such that a lower AUC\ncorresponds to a higher level of confidence in the future direction of\npedestrian motion.\n",
"title": "Context-Aware Pedestrian Motion Prediction In Urban Intersections"
} | null | null | null | null | true | null | 686 | null | Default | null | null |
null | {
"abstract": " We present E NERGY N ET , a new framework for analyzing and building\nartificial neural network architectures. Our approach adaptively learns the\nstructure of the networks in an unsupervised manner. The methodology is based\nupon the theoretical guarantees of the energy function of restricted Boltzmann\nmachines (RBM) of infinite number of nodes. We present experimental results to\nshow that the final network adapts to the complexity of a given problem.\n",
"title": "EnergyNet: Energy-based Adaptive Structural Learning of Artificial Neural Network Architectures"
} | null | null | null | null | true | null | 687 | null | Default | null | null |
null | {
"abstract": " Finding the dense regions of a graph and relations among them is a\nfundamental problem in network analysis. Core and truss decompositions reveal\ndense subgraphs with hierarchical relations. The incremental nature of\nalgorithms for computing these decompositions and the need for global\ninformation at each step of the algorithm hinders scalable parallelization and\napproximations since the densest regions are not revealed until the end. In a\nprevious work, Lu et al. proposed to iteratively compute the $h$-indices of\nneighbor vertex degrees to obtain the core numbers and prove that the\nconvergence is obtained after a finite number of iterations. This work\ngeneralizes the iterative $h$-index computation for truss decomposition as well\nas nucleus decomposition which leverages higher-order structures to generalize\ncore and truss decompositions. In addition, we prove convergence bounds on the\nnumber of iterations. We present a framework of local algorithms to obtain the\ncore, truss, and nucleus decompositions. Our algorithms are local, parallel,\noffer high scalability, and enable approximations to explore time and quality\ntrade-offs. Our shared-memory implementation verifies the efficiency,\nscalability, and effectiveness of our local algorithms on real-world networks.\n",
"title": "Local Algorithms for Hierarchical Dense Subgraph Discovery"
} | null | null | [
"Computer Science"
]
| null | true | null | 688 | null | Validated | null | null |
null | {
"abstract": " We propose a robust gesture-based communication pipeline for divers to\ninstruct an Autonomous Underwater Vehicle (AUV) to assist them in performing\nhigh-risk tasks and helping in case of emergency. A gesture communication\nlanguage (CADDIAN) is developed, based on consolidated and standardized diver\ngestures, including an alphabet, syntax and semantics, ensuring a logical\nconsistency. A hierarchical classification approach is introduced for hand\ngesture recognition based on stereo imagery and multi-descriptor aggregation to\nspecifically cope with underwater image artifacts, e.g. light backscatter or\ncolor attenuation. Once the classification task is finished, a syntax check is\nperformed to filter out invalid command sequences sent by the diver or\ngenerated by errors in the classifier. Throughout this process, the diver\nreceives constant feedback from an underwater tablet to acknowledge or abort\nthe mission at any time. The objective is to prevent the AUV from executing\nunnecessary, infeasible or potentially harmful motions. Experimental results\nunder different environmental conditions in archaeological exploration and\nbridge inspection applications show that the system performs well in the field.\n",
"title": "Robust Gesture-Based Communication for Underwater Human-Robot Interaction in the context of Search and Rescue Diver Missions"
} | null | null | null | null | true | null | 689 | null | Default | null | null |
null | {
"abstract": " Unique among alkali-doped $\\textit {A}$$_3$C$_{60}$ fullerene compounds, the\nA15 and fcc forms of Cs$_3$C$_{60}$ exhibit superconducting states varying\nunder hydrostatic pressure with highest transition temperatures at $T_\\textrm\n{C}$$^\\textrm {meas}$ = 38.3 and 35.2 K, respectively. Herein it is argued that\nthese two compounds under pressure represent the optimal materials of the\n$\\textit {A}$$_3$C$_{60}$ family, and that the C$_{60}$-associated\nsuperconductivity is mediated through Coulombic interactions with charges on\nthe alkalis. A derivation of the interlayer Coulombic pairing model of\nhigh-$T_\\textrm {C}$ superconductivity employing non-planar geometry is\nintroduced, generalizing the picture of two interacting layers to an\ninteraction between charge reservoirs located on the C$_{60}$ and alkali ions.\nThe optimal transition temperature follows the algebraic expression, $T_\\textrm\n{C0}$ = (12.474 nm$^2$ K)/$\\ell$${\\zeta}$, where $\\ell$ relates to the mean\nspacing between interacting surface charges on the C$_{60}$ and ${\\zeta}$ is\nthe average radial distance between the C$_{60}$ surface and the neighboring Cs\nions. Values of $T_\\textrm {C0}$ for the measured cation stoichiometries of\nCs$_{3-\\textrm{x}}$C$_{60}$ with x $\\approx$ 0 are found to be 38.19 and 36.88\nK for the A15 and fcc forms, respectively, with the dichotomy in transition\ntemperature reflecting the larger ${\\zeta}$ and structural disorder in the fcc\nform. In the A15 form, modeled interacting charges and Coulomb potential\ne$^2$/${\\zeta}$ are shown to agree quantitatively with findings from\nnuclear-spin relaxation and mid-infrared optical conductivity. In the fcc form,\nsuppression of $T_\\textrm {C}$$^\\textrm {meas}$ below $T_\\textrm {C0}$ is\nascribed to native structural disorder. Phononic effects in conjunction with\nCoulombic pairing are discussed.\n",
"title": "High-$T_\\textrm {C}$ superconductivity in Cs$_3$C$_{60}$ compounds governed by local Cs-C$_{60}$ Coulomb interactions"
} | null | null | null | null | true | null | 690 | null | Default | null | null |
null | {
"abstract": " Improving the performance of superconducting qubits and resonators generally\nresults from a combination of materials and fabrication process improvements\nand design modifications that reduce device sensitivity to residual losses. One\ninstance of this approach is to use trenching into the device substrate in\ncombination with superconductors and dielectrics with low intrinsic losses to\nimprove quality factors and coherence times. Here we demonstrate titanium\nnitride coplanar waveguide resonators with mean quality factors exceeding two\nmillion and controlled trenching reaching 2.2 $\\mu$m into the silicon\nsubstrate. Additionally, we measure sets of resonators with a range of sizes\nand trench depths and compare these results with finite-element simulations to\ndemonstrate quantitative agreement with a model of interface dielectric loss.\nWe then apply this analysis to determine the extent to which trenching can\nimprove resonator performance.\n",
"title": "Analysis and mitigation of interface losses in trenched superconducting coplanar waveguide resonators"
} | null | null | null | null | true | null | 691 | null | Default | null | null |
null | {
"abstract": " This paper will detail changes in the operational paradigm of the Fermi\nNational Accelerator Laboratory (FNAL) magnetron $H^{-}$ ion source due to\nupgrades in the accelerator system. Prior to November of 2012 the $H^{-}$ ions\nfor High Energy Physics (HEP) experiments were extracted at ~18 keV vertically\ndownward into a 90 degree bending magnet and accelerated through a\nCockcroft-Walton accelerating column to 750 keV. Following the upgrade in the\nfall of 2012 the $H^{-}$ ions are now directly extracted from a magnetron at 35\nkeV and accelerated to 750 keV by a Radio Frequency Quadrupole (RFQ). This\nchange in extraction energy as well as the orientation of the ion source\nrequired not only a redesign of the ion source, but an updated understanding of\nits operation at these new values. Discussed in detail are the changes to the\nion source timing, arc discharge current, hydrogen gas pressure, and cesium\ndelivery system that were needed to maintain consistent operation at >99%\nuptime for HEP, with an increased ion source lifetime of over 9 months.\n",
"title": "Recent Operation of the FNAL Magnetron $H^{-}$ Ion Source"
} | null | null | null | null | true | null | 692 | null | Default | null | null |
null | {
"abstract": " We consider the withdrawal of a ball from a fluid reservoir to understand the\nlongevity of the connection between that ball and the fluid it breaks away\nfrom, at intermediate Reynolds numbers. Scaling arguments based on the\nprocesses observed as the ball interacts with the fluid surface were applied to\nthe `pinch-off time', when the ball breaks its connection with the fluid from\nwhich it has been withdrawn, measured experimentally. At the lowest Reynolds\nnumbers tested, pinch-off occurs in a `surface seal' close to the reservoir\nsurface, where at larger Reynolds numbers pinch-off occurs in an `ejecta seal'\nclose to the ball. Our scaling analysis shows that the connection between ball\nand fluid is controlled by the fluid film draining from the ball as it\ncontinues to be winched away from the fluid reservoir. The draining flow itself\ndepends on the amount of fluid coating the ball on exit from the reservoir. We\nconsider the possibilities that this coating was created through: a surface\ntension driven Landau Levitch Derjaguin wetting of the surface; a\nvisco-inertial quick coating; or alternatively through the inertia of the fluid\nmoving with the ball through the reservoir. We show that although the pinch-off\nmechanism is controlled by viscosity, the coating mechanism is governed by a\ndifferent length and timescale, dictated by the inertial added mass of the ball\nwhen submersed.\n",
"title": "A Ball Breaking Away from a Fluid"
} | null | null | null | null | true | null | 693 | null | Default | null | null |
null | {
"abstract": " We disentangle all the individual degrees of freedom in the quantum impurity\nproblem to deconstruct the Kondo singlet, both in real and energy space, by\nstudying the contribution of each individual free electron eigenstate. This is\na problem of two spins coupled to a bath, where the bath is formed by the\nremaining conduction electrons. Being a mixed state, we resort to the\n\"concurrence\" to quantify entanglement. We identify \"projected natural\norbitals\" that allow us to individualize a single-particle electronic wave\nfunction that is responsible of more than $90\\%$ of the impurity screening. In\nthe weak coupling regime, the impurity is entangled to an electron at the Fermi\nlevel, while in the strong coupling regime, the impurity counterintuitively\nentangles mostly with the high energy electrons and disentangles completely\nfrom the low-energy states carving a \"hole\" around the Fermi level. This\nenables one to use concurrence as a pseudo order parameter to compute the\ncharacteristic \"size\" of the Kondo cloud, beyond which electrons are are weakly\ncorrelated to the impurity and are dominated by the physics of the boundary.\n",
"title": "Unveiling the internal entanglement structure of the Kondo singlet"
} | null | null | null | null | true | null | 694 | null | Default | null | null |
null | {
"abstract": " Motivated by the recently proposed parallel orbital-updating approach in real\nspace method, we propose a parallel orbital-updating based plane-wave basis\nmethod for electronic structure calculations, for solving the corresponding\neigenvalue problems. In addition, we propose two new modified parallel\norbital-updating methods. Compared to the traditional plane-wave methods, our\nmethods allow for two-level parallelization, which is particularly interesting\nfor large scale parallelization. Numerical experiments show that these new\nmethods are more reliable and efficient for large scale calculations on modern\nsupercomputers\n",
"title": "A parallel orbital-updating based plane-wave basis method for electronic structure calculations"
} | null | null | [
"Physics",
"Mathematics"
]
| null | true | null | 695 | null | Validated | null | null |
null | {
"abstract": " The particular type of four-kink multi-solitons (or quadrons) adiabatic\ndynamics of the sine-Gordon equation in a model with two identical point\nattracting impurities has been studied. This model can be used for describing\nmagnetization localized waves in multilayer ferromagnet. The quadrons structure\nand properties has been numerically investigated. The cases of both large and\nsmall distances between impurities has been viewed. The dependence of the\nlocalized in impurity region nonlinear high-amplitude waves frequencies on the\ndistance between the impurities has been found. For an analytical description\nof two bound localized on impurities nonlinear waves dynamics, using\nperturbation theory, the system of differential equations for harmonic\noscillators with elastic link has been found. The analytical model\nqualitatively describes the results of the sine-Gordon equation numerical\nsimulation.\n",
"title": "Dynamics of the multi-soliton waves in the sine-Gordon model with two identical point impurities"
} | null | null | [
"Physics"
]
| null | true | null | 696 | null | Validated | null | null |
null | {
"abstract": " Estimates of the Hubble constant, $H_0$, from the distance ladder and the\ncosmic microwave background (CMB) differ at the $\\sim$3-$\\sigma$ level,\nindicating a potential issue with the standard $\\Lambda$CDM cosmology.\nInterpreting this tension correctly requires a model comparison calculation\ndepending on not only the traditional `$n$-$\\sigma$' mismatch but also the\ntails of the likelihoods. Determining the form of the tails of the local $H_0$\nlikelihood is impossible with the standard Gaussian least-squares\napproximation, as it requires using non-Gaussian distributions to faithfully\nrepresent anchor likelihoods and model outliers in the Cepheid and supernova\n(SN) populations, and simultaneous fitting of the full distance-ladder dataset\nto correctly propagate uncertainties. We have developed a Bayesian hierarchical\nmodel that describes the full distance ladder, from nearby geometric anchors\nthrough Cepheids to Hubble-Flow SNe. This model does not rely on any\ndistributions being Gaussian, allowing outliers to be modeled and obviating the\nneed for arbitrary data cuts. Sampling from the $\\sim$3000-parameter joint\nposterior using Hamiltonian Monte Carlo, we find $H_0$ = (72.72 $\\pm$ 1.67)\n${\\rm km\\,s^{-1}\\,Mpc^{-1}}$ when applied to the outlier-cleaned Riess et al.\n(2016) data, and ($73.15 \\pm 1.78$) ${\\rm km\\,s^{-1}\\,Mpc^{-1}}$ with SN\noutliers reintroduced. Our high-fidelity sampling of the low-$H_0$ tail of the\ndistance-ladder likelihood allows us to apply Bayesian model comparison to\nassess the evidence for deviation from $\\Lambda$CDM. We set up this comparison\nto yield a lower limit on the odds of the underlying model being $\\Lambda$CDM\ngiven the distance-ladder and Planck XIII (2016) CMB data. The odds against\n$\\Lambda$CDM are at worst 10:1 or 7:1, depending on whether the SNe outliers\nare cut or modeled, or 60:1 if an approximation to the Planck Int. XLVI (2016)\nlikelihood is used.\n",
"title": "Clarifying the Hubble constant tension with a Bayesian hierarchical model of the local distance ladder"
} | null | null | [
"Physics"
]
| null | true | null | 697 | null | Validated | null | null |
null | {
"abstract": " A multi-user multi-armed bandit (MAB) framework is used to develop algorithms\nfor uncoordinated spectrum access. The number of users is assumed to be unknown\nto each user. A stochastic setting is first considered, where the rewards on a\nchannel are the same for each user. In contrast to prior work, it is assumed\nthat the number of users can possibly exceed the number of channels, and that\nrewards can be non-zero even under collisions. The proposed algorithm consists\nof an estimation phase and an allocation phase. It is shown that if every user\nadopts the algorithm, the system wide regret is constant with time with high\nprobability. The regret guarantees hold for any number of users and channels,\nin particular, even when the number of users is less than the number of\nchannels. Next, an adversarial multi-user MAB framework is considered, where\nthe rewards on the channels are user-dependent. It is assumed that the number\nof users is less than the number of channels, and that the users receive zero\nreward on collision. The proposed algorithm combines the Exp3.P algorithm\ndeveloped in prior work for single user adversarial bandits with a collision\nresolution mechanism to achieve sub-linear regret. It is shown that if every\nuser employs the proposed algorithm, the system wide regret is of the order\n$O(T^\\frac{3}{4})$ over a horizon of time $T$. The algorithms in both\nstochastic and adversarial scenarios are extended to the dynamic case where the\nnumber of users in the system evolves over time and are shown to lead to\nsub-linear regret.\n",
"title": "Multi-User Multi-Armed Bandits for Uncoordinated Spectrum Access"
} | null | null | null | null | true | null | 698 | null | Default | null | null |
null | {
"abstract": " In this paper, we analyze the effects of contact models on contact-implicit\ntrajectory optimization for manipulation. We consider three different\napproaches: (1) a contact model that is based on complementarity constraints,\n(2) a smooth contact model, and our proposed method (3) a variable smooth\ncontact model. We compare these models in simulation in terms of physical\naccuracy, quality of motions, and computation time. In each case, the\noptimization process is initialized by setting all torque variables to zero,\nnamely, without a meaningful initial guess. For simulations, we consider a\npushing task with varying complexity for a 7 degrees-of-freedom robot arm. Our\nresults demonstrate that the optimization based on the proposed variable smooth\ncontact model provides a good trade-off between the physical fidelity and\nquality of motions at the cost of increased computation time.\n",
"title": "A Comparative Analysis of Contact Models in Trajectory Optimization for Manipulation"
} | null | null | null | null | true | null | 699 | null | Default | null | null |
null | {
"abstract": " The challenge of assigning importance to individual neurons in a network is\nof interest when interpreting deep learning models. In recent work, Dhamdhere\net al. proposed Total Conductance, a \"natural refinement of Integrated\nGradients\" for attributing importance to internal neurons. Unfortunately, the\nauthors found that calculating conductance in tensorflow required the addition\nof several custom gradient operators and did not scale well. In this work, we\nshow that the formula for Total Conductance is mathematically equivalent to\nPath Integrated Gradients computed on a hidden layer in the network. We provide\na scalable implementation of Total Conductance using standard tensorflow\ngradient operators that we call Neuron Integrated Gradients. We compare Neuron\nIntegrated Gradients to DeepLIFT, a pre-existing computationally efficient\napproach that is applicable to calculating internal neuron importance. We find\nthat DeepLIFT produces strong empirical results and is faster to compute, but\nbecause it lacks the theoretical properties of Neuron Integrated Gradients, it\nmay not always be preferred in practice. Colab notebook reproducing results:\nthis http URL\n",
"title": "Computationally Efficient Measures of Internal Neuron Importance"
} | null | null | null | null | true | null | 700 | null | Default | null | null |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.