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": " Detect facial keypoints is a critical element in face recognition. However,\nthere is difficulty to catch keypoints on the face due to complex influences\nfrom original images, and there is no guidance to suitable algorithms. In this\npaper, we study different algorithms that can be applied to locate keyponits.\nSpecifically: our framework (1)prepare the data for further investigation\n(2)Using PCA and LBP to process the data (3) Apply different algorithms to\nanalysis data, including linear regression models, tree based model, neural\nnetwork and convolutional neural network, etc. Finally we will give our\nconclusion and further research topic. A comprehensive set of experiments on\ndataset demonstrates the effectiveness of our framework.\n",
"title": "Facial Keypoints Detection"
}
| null | null | null | null | true | null |
1401
| null |
Default
| null | null |
null |
{
"abstract": " One initial and essential question of magnetism is whether the magnetic\nproperties of a material are governed by localized moments or itinerant\nelectrons. Here we expose the case for the weakly ferromagnetic system\nFeGa$_{3-y}$Ge$_y$ wherein these two opposite models are reconciled, such that\nthe magnetic susceptibility is quantitatively explained by taking into account\nthe effects of spin-spin correlation. With the electron doping introduced by Ge\nsubstitution, the diamagnetic insulating parent compound FeGa$_3$ becomes a\nparamagnetic metal as early as at $ y=0.01 $, and turns into a weakly\nferromagnetic metal around the quantum critical point $ y=0.15 $. Within the\nferromagnetic regime of FeGa$_{3-y}$Ge$_y$, the magnetic properties are of a\nweakly itinerant ferromagnetic nature, located in the intermediate regime\nbetween the localized and the itinerant dominance. Our analysis implies a\npotential universality for all itinerant-electron ferromagnets.\n",
"title": "Transitions from a Kondo-like diamagnetic insulator into a modulated ferromagnetic metal in $\\bm{\\mathrm{FeGa}_{3-y}\\mathrm{Ge}_y}$"
}
| null | null |
[
"Physics"
] | null | true | null |
1402
| null |
Validated
| null | null |
null |
{
"abstract": " In this paper, we exhibit the tradeoffs between the (training) sample,\ncomputation and storage complexity for the problem of supervised classification\nusing signal subspace estimation. Our main tool is the use of tensor subspaces,\ni.e. subspaces with a Kronecker structure, for embedding the data into lower\ndimensions. Among the subspaces with a Kronecker structure, we show that using\nsubspaces with a hierarchical structure for representing data leads to improved\ntradeoffs. One of the main reasons for the improvement is that embedding data\ninto these hierarchical Kronecker structured subspaces prevents overfitting at\nhigher latent dimensions.\n",
"title": "Sample, computation vs storage tradeoffs for classification using tensor subspace models"
}
| null | null | null | null | true | null |
1403
| null |
Default
| null | null |
null |
{
"abstract": " Estimates of population size for hidden and hard-to-reach individuals are of\nparticular interest to health officials when health problems are concentrated\nin such populations. Efforts to derive these estimates are often frustrated by\na range of factors including social stigma or an association with illegal\nactivities that ordinarily preclude conventional survey strategies. This paper\nbuilds on and extends prior work that proposed a method to meet these\nchallenges. Here we describe a rigorous formalization of a one-step,\nnetwork-based population estimation procedure that can be employed under\nconditions of anonymity. The estimation procedure is designed to be implemented\nalongside currently accepted strategies for research with hidden populations.\nSimulation experiments are described that test the efficacy of the method\nacross a range of implementation conditions and hidden population sizes. The\nresults of these experiments show that reliable population estimates can be\nderived for hidden, networked population as large as 12,500 and perhaps larger\nfor one family of random graphs. As such, the method shows potential for\ncost-effective implementation health and disease surveillance officials\nconcerned with hidden populations. Limitations and future work are discussed in\nthe concluding section.\n",
"title": "One-step Estimation of Networked Population Size with Anonymity Using Respondent-Driven Capture-Recapture and Hashing"
}
| null | null | null | null | true | null |
1404
| null |
Default
| null | null |
null |
{
"abstract": " Single ion solvation free energies are one of the most important properties\nof electrolyte solutions and yet there is ongoing debate about what these\nvalues are. Only the values for neutral ion pairs are known. Here, we use DFT\ninteraction potentials with molecular dynamics simulation (DFT-MD) combined\nwith a modified version of the quasi-chemical theory (QCT) to calculate these\nenergies for the lithium and fluoride ions. A method to correct for the error\nin the DFT functional is developed and very good agreement with the\nexperimental value for the lithium fluoride pair is obtained. Moreover, this\nmethod partitions the energies into physically intuitive terms such as surface\npotential, cavity and charging energies which are amenable to descriptions with\nreduced models. Our research suggests that lithium's solvation free energy is\ndominated by the free energetics of a charged hard sphere, whereas fluoride\nexhibits significant quantum mechanical behavior that cannot be simply\ndescribed with a reduced model.\n",
"title": "Real single ion solvation free energies with quantum mechanical simulation"
}
| null | null | null | null | true | null |
1405
| null |
Default
| null | null |
null |
{
"abstract": " We consider estimation of worker skills from worker-task interaction data\n(with unknown labels) for the single-coin crowd-sourcing binary classification\nmodel in symmetric noise. We define the (worker) interaction graph whose nodes\nare workers and an edge between two nodes indicates whether or not the two\nworkers participated in a common task. We show that skills are asymptotically\nidentifiable if and only if an appropriate limiting version of the interaction\ngraph is irreducible and has odd-cycles. We then formulate a weighted rank-one\noptimization problem to estimate skills based on observations on an\nirreducible, aperiodic interaction graph. We propose a gradient descent scheme\nand show that for such interaction graphs estimates converge asymptotically to\nthe global minimum. We characterize noise robustness of the gradient scheme in\nterms of spectral properties of signless Laplacians of the interaction graph.\nWe then demonstrate that a plug-in estimator based on the estimated skills\nachieves state-of-art performance on a number of real-world datasets. Our\nresults have implications for rank-one matrix completion problem in that\ngradient descent can provably recover $W \\times W$ rank-one matrices based on\n$W+1$ off-diagonal observations of a connected graph with a single odd-cycle.\n",
"title": "Crowdsourcing with Sparsely Interacting Workers"
}
| null | null | null | null | true | null |
1406
| null |
Default
| null | null |
null |
{
"abstract": " Recent deep learning based denoisers often outperform state-of-the-art\nconventional denoisers such as BM3D. They are typically trained to minimize the\nmean squared error (MSE) between the output of a deep neural network and the\nground truth image. In deep learning based denoisers, it is important to use\nhigh quality noiseless ground truth for high performance, but it is often\nchallenging or even infeasible to obtain such a clean image in application\nareas such as hyperspectral remote sensing and medical imaging. We propose a\nStein's Unbiased Risk Estimator (SURE) based method for training deep neural\nnetwork denoisers only with noisy images. We demonstrated that our SURE based\nmethod without ground truth was able to train deep neural network denoisers to\nyield performance close to deep learning denoisers trained with ground truth\nand to outperform state-of-the-art BM3D. Further improvements were achieved by\nincluding noisy test images for training denoiser networks using our proposed\nSURE based method.\n",
"title": "Training deep learning based denoisers without ground truth data"
}
| null | null | null | null | true | null |
1407
| null |
Default
| null | null |
null |
{
"abstract": " Here we consider some well-known facts in syntax from a physics perspective,\nwhich allows us to establish some remarkable equivalences. Specifically, we\nobserve that the operation MERGE put forward by N. Chomsky in 1995 can be\ninterpreted as a physical information coarse-graining. Thus, MERGE in\nlinguistics entails information renormalization in physics, according to\ndifferent time scales. We make this point mathematically formal in terms of\nlanguage models, i.e., probability distributions over word sequences, widely\nused in natural language processing as well as other ambits. In this setting,\nMERGE corresponds to a 3-index probability tensor implementing a\ncoarse-graining, akin to a probabilistic context-free grammar. The probability\nvectors of meaningful sentences are naturally given by stochastic tensor\nnetworks (TN) that are mostly loop-free, such as Tree Tensor Networks and\nMatrix Product States. These structures have short-ranged correlations in the\nsyntactic distance by construction and, because of the peculiarities of human\nlanguage, they are extremely efficient to manipulate computationally. We also\npropose how to obtain such language models from probability distributions of\ncertain TN quantum states, which we show to be efficiently preparable by a\nquantum computer. Moreover, using tools from entanglement theory, we use these\nquantum states to prove classical lower bounds on the perplexity of the\nprobability distribution for a set of words in a sentence. Implications of\nthese results are discussed in the ambits of theoretical and computational\nlinguistics, artificial intelligence, programming languages, RNA and protein\nsequencing, quantum many-body systems, and beyond. Our work shows how many of\nthe key linguistic ideas from the last century, including developments in\ncomputational linguistics, fit perfectly with known physical concepts linked to\nrenormalization.\n",
"title": "Language Design and Renormalization"
}
| null | null | null | null | true | null |
1408
| null |
Default
| null | null |
null |
{
"abstract": " We study the moduli space of stable sheaves of Euler characteristic 2,\nsupported on curves of arithmetic genus 2 contained in a smooth quadric\nsurface. We show that this moduli space is rational. We compute its Betti\nnumbers and we give a classification of the stable sheaves involving locally\nfree resolutions.\n",
"title": "On the geometry of the moduli space of sheaves supported on curves of genus two in a quadric surface"
}
| null | null | null | null | true | null |
1409
| null |
Default
| null | null |
null |
{
"abstract": " The development of efficient (heuristic) algorithms for practical\ncombinatorial optimization problems is costly, so we want to automatically\nlearn them instead. We show the feasibility of this approach on the important\nTravelling Salesman Problem (TSP). We learn a heuristic algorithm that uses a\nNeural Network policy to construct a tour. As an alternative to the Pointer\nNetwork, our model is based entirely on (graph) attention layers and is\ninvariant to the input order of the nodes. We train the model efficiently using\nREINFORCE with a simple and robust baseline based on a deterministic (greedy)\nrollout of the best policy so far. We significantly improve over results from\nprevious works that consider learned heuristics for the TSP, reducing the\noptimality gap for a single tour construction from 1.51% to 0.32% for instances\nwith 20 nodes, from 4.59% to 1.71% for 50 nodes and from 6.89% to 4.43% for 100\nnodes. Additionally, we improve over a recent Reinforcement Learning framework\nfor two variants of the Vehicle Routing Problem (VRP).\n",
"title": "Attention Solves Your TSP, Approximately"
}
| null | null | null | null | true | null |
1410
| null |
Default
| null | null |
null |
{
"abstract": " We study a demand response problem from utility (also referred to as\noperator)'s perspective with realistic settings, in which the utility faces\nuncertainty and limited communication. Specifically, the utility does not know\nthe cost function of consumers and cannot have multiple rounds of information\nexchange with consumers. We formulate an optimization problem for the utility\nto minimize its operational cost considering time-varying demand response\ntargets and responses of consumers. We develop a joint online learning and\npricing algorithm. In each time slot, the utility sends out a price signal to\nall consumers and estimates the cost functions of consumers based on their\nnoisy responses. We measure the performance of our algorithm using regret\nanalysis and show that our online algorithm achieves logarithmic regret with\nrespect to the operating horizon. In addition, our algorithm employs linear\nregression to estimate the aggregate response of consumers, making it easy to\nimplement in practice. Simulation experiments validate the theoretic results\nand show that the performance gap between our algorithm and the offline\noptimality decays quickly.\n",
"title": "A Distributed Online Pricing Strategy for Demand Response Programs"
}
| null | null | null | null | true | null |
1411
| null |
Default
| null | null |
null |
{
"abstract": " This paper presents a triangular lattice photonic crystal fiber with very\nhigh nonlinear coefficient. Finite element method (FEM) is used to scrutinize\ndifferent optical properties of proposed highly nonlinear photonic crystal\nfiber (HNL-PCF). The HNL-PCF exhibits a high nonlinearity up to $10\\times10^{4}\nW^{-1}km^{-1}$ over the wavelength of 1500 nm to 1700 nm. Moreover, proposed\nHNL-PCF shows a very low confinement loss of $10^{-3} dB/km$ at 1550 nm\nwavelength. Furthermore, chromatic dispersion, dispersion slope, effective area\netc. are also analyzed thoroughly. The proposed fiber will be a suitable\ncandidate for broadband dispersion compensation, sensor devices and\nsupercontinuum generation.\n",
"title": "Highly Nonlinear and Low Confinement Loss Photonic Crystal Fiber Using GaP Slot Core"
}
| null | null | null | null | true | null |
1412
| null |
Default
| null | null |
null |
{
"abstract": " The Epicurean Philosophy is commonly thought as simplistic and hedonistic.\nHere I discuss how this is a misconception and explore its link to\nReinforcement Learning. Based on the letters of Epicurus, I construct an\nobjective function for hedonism which turns out to be equivalent of the\nReinforcement Learning objective function when omitting the discount factor. I\nthen discuss how Plato and Aristotle 's views that can be also loosely linked\nto Reinforcement Learning, as well as their weaknesses in relationship to it.\nFinally, I emphasise the close affinity of the Epicurean views and the Bellman\nequation.\n",
"title": "Is Epicurus the father of Reinforcement Learning?"
}
| null | null | null | null | true | null |
1413
| null |
Default
| null | null |
null |
{
"abstract": " The use of low-precision fixed-point arithmetic along with stochastic\nrounding has been proposed as a promising alternative to the commonly used\n32-bit floating point arithmetic to enhance training neural networks training\nin terms of performance and energy efficiency. In the first part of this paper,\nthe behaviour of the 12-bit fixed-point arithmetic when training a\nconvolutional neural network with the CIFAR-10 dataset is analysed, showing\nthat such arithmetic is not the most appropriate for the training phase. After\nthat, the paper presents and evaluates, under the same conditions, alternative\nlow-precision arithmetics, starting with the 12-bit floating-point arithmetic.\nThese two representations are then leveraged using local scaling in order to\nincrease accuracy and get closer to the baseline 32-bit floating-point\narithmetic. Finally, the paper introduces a simplified model in which both the\noutputs and the gradients of the neural networks are constrained to\npower-of-two values, just using 7 bits for their representation. The evaluation\ndemonstrates a minimal loss in accuracy for the proposed Power-of-Two neural\nnetwork, avoiding the use of multiplications and divisions and thereby,\nsignificantly reducing the training time as well as the energy consumption and\nmemory requirements during the training and inference phases.\n",
"title": "Low-Precision Floating-Point Schemes for Neural Network Training"
}
| null | null | null | null | true | null |
1414
| null |
Default
| null | null |
null |
{
"abstract": " Person re-identification task has been greatly boosted by deep convolutional\nneural networks (CNNs) in recent years. The core of which is to enlarge the\ninter-class distinction as well as reduce the intra-class variance. However, to\nachieve this, existing deep models prefer to adopt image pairs or triplets to\nform verification loss, which is inefficient and unstable since the number of\ntraining pairs or triplets grows rapidly as the number of training data grows.\nMoreover, their performance is limited since they ignore the fact that\ndifferent dimension of embedding may play different importance. In this paper,\nwe propose to employ identification loss with center loss to train a deep model\nfor person re-identification. The training process is efficient since it does\nnot require image pairs or triplets for training while the inter-class\ndistinction and intra-class variance are well handled. To boost the\nperformance, a new feature reweighting (FRW) layer is designed to explicitly\nemphasize the importance of each embedding dimension, thus leading to an\nimproved embedding. Experiments on several benchmark datasets have shown the\nsuperiority of our method over the state-of-the-art alternatives on both\naccuracy and speed.\n",
"title": "Deep Person Re-Identification with Improved Embedding and Efficient Training"
}
| null | null | null | null | true | null |
1415
| null |
Default
| null | null |
null |
{
"abstract": " We consider the task of unsupervised extraction of meaningful latent\nrepresentations of speech by applying autoencoding neural networks to speech\nwaveforms. The goal is to learn a representation able to capture high level\nsemantic content from the signal, e.g. phoneme identities, while being\ninvariant to confounding low level details in the signal such as the underlying\npitch contour or background noise. The behavior of autoencoder models depends\non the kind of constraint that is applied to the latent representation. We\ncompare three variants: a simple dimensionality reduction bottleneck, a\nGaussian Variational Autoencoder (VAE), and a discrete Vector Quantized VAE\n(VQ-VAE). We analyze the quality of learned representations in terms of speaker\nindependence, the ability to predict phonetic content, and the ability to\naccurately reconstruct individual spectrogram frames. Moreover, for discrete\nencodings extracted using the VQ-VAE, we measure the ease of mapping them to\nphonemes. We introduce a regularization scheme that forces the representations\nto focus on the phonetic content of the utterance and report performance\ncomparable with the top entries in the ZeroSpeech 2017 unsupervised acoustic\nunit discovery task.\n",
"title": "Unsupervised speech representation learning using WaveNet autoencoders"
}
| null | null | null | null | true | null |
1416
| null |
Default
| null | null |
null |
{
"abstract": " The many-body localization (MBL) is commonly related to a strong spatial\ndisorder. We show that MBL may alternatively be generated by adding a temporal\ndisorder to periodically driven many-body systems. We reach this conclusion by\nmapping the evolution of such systems on the dynamics of the time-independent,\ndisordered, Hubbard-like models. Our result opens the way to experimental\nstudies of MBL in systems that reveal crystalline structures in the time\ndomain. In particular, we discuss two relevant setups which can be implemented\nin experiments on ultra-cold atomic gases.\n",
"title": "Many-body localization caused by temporal disorder"
}
| null | null | null | null | true | null |
1417
| null |
Default
| null | null |
null |
{
"abstract": " Verifying that a statistically significant result is scientifically\nmeaningful is not only good scientific practice, it is a natural way to control\nthe Type I error rate. Here we introduce a novel extension of the p-value - a\nsecond-generation p-value - that formally accounts for scientific relevance and\nleverages this natural Type I Error control. The approach relies on a\npre-specified interval null hypothesis that represents the collection of effect\nsizes that are scientifically uninteresting or are practically null. The\nsecond-generation p-value is the proportion of data-supported hypotheses that\nare also null hypotheses. As such, second-generation p-values indicate when the\ndata are compatible with null hypotheses, or with alternative hypotheses, or\nwhen the data are inconclusive. Moreover, second-generation p-values provide a\nproper scientific adjustment for multiple comparisons and reduce false\ndiscovery rates. This is an advance for environments rich in data, where\ntraditional p-value adjustments are needlessly punitive. Second-generation\np-values promote transparency, rigor and reproducibility of scientific results\nby a priori specifying which candidate hypotheses are practically meaningful\nand by providing a more reliable statistical summary of when the data are\ncompatible with alternative or null hypotheses.\n",
"title": "Second-generation p-values: improved rigor, reproducibility, & transparency in statistical analyses"
}
| null | null | null | null | true | null |
1418
| null |
Default
| null | null |
null |
{
"abstract": " In this paper, we adopt a new noisy wireless network model introduced very\nrecently by Censor-Hillel et al. in [ACM PODC 2017, CHHZ17]. More specifically,\nfor a given noise parameter $p\\in [0,1],$ any sender has a probability of $p$\nof transmitting noise or any receiver of a single transmission in its\nneighborhood has a probability $p$ of receiving noise.\nIn this paper, we first propose a new asymptotically latency-optimal\napproximation algorithm (under faultless model) that can complete\nsingle-message broadcasting task in $D+O(\\log^2 n)$ time units/rounds in any\nWMN of size $n,$ and diameter $D$. We then show this diameter-linear\nbroadcasting algorithm remains robust under the noisy wireless network model\nand also improves the currently best known result in CHHZ17 by a\n$\\Theta(\\log\\log n)$ factor.\nIn this paper, we also further extend our robust single-message broadcasting\nalgorithm to $k$ multi-message broadcasting scenario and show it can broadcast\n$k$ messages in $O(D+k\\log n+\\log^2 n)$ time rounds. This new robust\nmulti-message broadcasting scheme is not only asymptotically optimal but also\nanswers affirmatively the problem left open in CHHZ17 on the existence of an\nalgorithm that is robust to sender and receiver faults and can broadcast $k$\nmessages in $O(D+k\\log n + polylog(n))$ time rounds.\n",
"title": "Latency Optimal Broadcasting in Noisy Wireless Mesh Networks"
}
| null | null |
[
"Computer Science"
] | null | true | null |
1419
| null |
Validated
| null | null |
null |
{
"abstract": " We study the problem of constructing synthetic graphs that resemble\nreal-world directed graphs in terms of their degree correlations. We define the\nproblem of directed 2K construction (D2K) that takes as input the directed\ndegree sequence (DDS) and a joint degree and attribute matrix (JDAM) so as to\ncapture degree correlation specifically in directed graphs. We provide\nnecessary and sufficient conditions to decide whether a target D2K is\nrealizable, and we design an efficient algorithm that creates realizations with\nthat target D2K. We evaluate our algorithm in creating synthetic graphs that\ntarget real-world directed graphs (such as Twitter) and we show that it brings\nsignificant benefits compared to state-of-the-art approaches.\n",
"title": "Construction of Directed 2K Graphs"
}
| null | null | null | null | true | null |
1420
| null |
Default
| null | null |
null |
{
"abstract": " Recognition of Handwritten Mathematical Expressions (HMEs) is a challenging\nproblem because of the ambiguity and complexity of two-dimensional handwriting.\nMoreover, the lack of large training data is a serious issue, especially for\nacademic recognition systems. In this paper, we propose pattern generation\nstrategies that generate shape and structural variations to improve the\nperformance of recognition systems based on a small training set. For data\ngeneration, we employ the public databases: CROHME 2014 and 2016 of online\nHMEs. The first strategy employs local and global distortions to generate shape\nvariations. The second strategy decomposes an online HME into sub-online HMEs\nto get more structural variations. The hybrid strategy combines both these\nstrategies to maximize shape and structural variations. The generated online\nHMEs are converted to images for offline HME recognition. We tested our\nstrategies in an end-to-end recognition system constructed from a recent deep\nlearning model: Convolutional Neural Network and attention-based\nencoder-decoder. The results of experiments on the CROHME 2014 and 2016\ndatabases demonstrate the superiority and effectiveness of our strategies: our\nhybrid strategy achieved classification rates of 48.78% and 45.60%,\nrespectively, on these databases. These results are competitive compared to\nothers reported in recent literature. Our generated datasets are openly\navailable for research community and constitute a useful resource for the HME\nrecognition research in future.\n",
"title": "Pattern Generation Strategies for Improving Recognition of Handwritten Mathematical Expressions"
}
| null | null | null | null | true | null |
1421
| null |
Default
| null | null |
null |
{
"abstract": " This is an expository article on properties of actions on Lie groups by\nsubgroups of their automorphism groups. After recalling various results on the\nstructure of the automorphism groups, we discuss actions with dense orbits,\ninvariant and quasi-invariant measures, the induced actions on the spaces of\nprobability measures on the groups, and results concerning various issues in\nergodic theory, topological dynamics, smooth dynamical systems, and probability\ntheory on Lie groups.\n",
"title": "Actions of automorphism groups of Lie groups"
}
| null | null | null | null | true | null |
1422
| null |
Default
| null | null |
null |
{
"abstract": " In this paper, we theoretically study x-ray multiphoton ionization dynamics\nof heavy atoms taking into account relativistic and resonance effects. When an\natom is exposed to an intense x-ray pulse generated by an x-ray free-electron\nlaser (XFEL), it is ionized to a highly charged ion via a sequence of\nsingle-photon ionization and accompanying relaxation processes, and its final\ncharge state is limited by the last ionic state that can be ionized by a\nsingle-photon ionization. If x-ray multiphoton ionization involves deep\ninner-shell electrons in heavy atoms, energy shifts by relativistic effects\nplay an important role in ionization dynamics, as pointed out in [Phys.\\ Rev.\\\nLett.\\ \\textbf{110}, 173005 (2013)]. On the other hand, if the x-ray beam has a\nbroad energy bandwidth, the high-intensity x-ray pulse can drive resonant\nphoto-excitations for a broad range of ionic states and ionize even beyond the\ndirect one-photon ionization limit, as first proposed in [Nature\\ Photon.\\\n\\textbf{6}, 858 (2012)]. To investigate both relativistic and resonance\neffects, we extend the \\textsc{xatom} toolkit to incorporate relativistic\nenergy corrections and resonant excitations in x-ray multiphoton ionization\ndynamics calculations. Charge-state distributions are calculated for Xe atoms\ninteracting with intense XFEL pulses at a photon energy of 1.5~keV and 5.5~keV,\nrespectively. For both photon energies, we demonstrate that the role of\nresonant excitations in ionization dynamics is altered due to significant\nshifts of orbital energy levels by relativistic effects. Therefore it is\nnecessary to take into account both effects to accurately simulate multiphoton\nmultiple ionization dynamics at high x-ray intensity.\n",
"title": "Interplay between relativistic energy corrections and resonant excitations in x-ray multiphoton ionization dynamics of Xe atoms"
}
| null | null | null | null | true | null |
1423
| null |
Default
| null | null |
null |
{
"abstract": " Magnetic materials hosting correlated electrons play an important role for\ninformation technology and signal processing. The currently used ferro-, ferri-\nand antiferromagnetic materials provide microscopic moments (spins) that are\nmainly collinear. Recently more complex spin structures such as spin helices\nand cycloids have regained a lot of interest. The interest has been initiated\nby the discovery of the skyrmion lattice phase in non-centrosymmetric helical\nmagnets. In this review we address how spin helices and skyrmion lattices\nenrich the microwave characteristics of magnetic materials. When discussing\nperspectives for microwave electronics and magnonics we focus particularly on\ninsulating materials as they avoid eddy current losses, offer low spin-wave\ndamping, and might allow for electric field control of collective spin\nexcitations. Thereby, they further fuel the vision of magnonics operated at low\nenergy consumption.\n",
"title": "Collective spin excitations of helices and magnetic skyrmions: review and perspectives of magnonics in non-centrosymmetric magnets"
}
| null | null | null | null | true | null |
1424
| null |
Default
| null | null |
null |
{
"abstract": " Large amount of image denoising literature focuses on single channel images\nand often experimentally validates the proposed methods on tens of images at\nmost. In this paper, we investigate the interaction between denoising and\nclassification on large scale dataset. Inspired by classification models, we\npropose a novel deep learning architecture for color (multichannel) image\ndenoising and report on thousands of images from ImageNet dataset as well as\ncommonly used imagery. We study the importance of (sufficient) training data,\nhow semantic class information can be traded for improved denoising results. As\na result, our method greatly improves PSNR performance by 0.34 - 0.51 dB on\naverage over state-of-the art methods on large scale dataset. We conclude that\nit is beneficial to incorporate in classification models. On the other hand, we\nalso study how noise affect classification performance. In the end, we come to\na number of interesting conclusions, some being counter-intuitive.\n",
"title": "On the Relation between Color Image Denoising and Classification"
}
| null | null | null | null | true | null |
1425
| null |
Default
| null | null |
null |
{
"abstract": " In this paper, we analyze in depth a simplicial decomposition like\nalgorithmic framework for large scale convex quadratic programming. In\nparticular, we first propose two tailored strategies for handling the master\nproblem. Then, we describe a few techniques for speeding up the solution of the\npricing problem. We report extensive numerical experiments on both real\nportfolio optimization and general quadratic programming problems, showing the\nefficiency and robustness of the method when compared to Cplex.\n",
"title": "A simplicial decomposition framework for large scale convex quadratic programming"
}
| null | null | null | null | true | null |
1426
| null |
Default
| null | null |
null |
{
"abstract": " Blockchains are distributed data structures that are used to achieve\nconsensus in systems for cryptocurrencies (like Bitcoin) or smart contracts\n(like Ethereum). Although blockchains gained a lot of popularity recently,\nthere is no logic-based model for blockchains available. We introduce BCL, a\ndynamic logic to reason about blockchain updates, and show that BCL is sound\nand complete with respect to a simple blockchain model.\n",
"title": "A Logic of Blockchain Updates"
}
| null | null |
[
"Computer Science"
] | null | true | null |
1427
| null |
Validated
| null | null |
null |
{
"abstract": " We prove in a mathematically rigorous way the asymptotic formula of Flaherty\nand Keller on the effective property of densely packed periodic elastic\ncomposites with hard inclusions. The proof is based on the primal-dual\nvariational principle, where the upper bound is derived by using the\nKeller-type test functions and the lower bound by singular functions made of\nnuclei of strain. Singular functions are solutions of the Lamé system and\ncapture precisely singular behavior of the stress in the narrow region between\ntwo adjacent hard inclusions.\n",
"title": "A proof of the Flaherty-Keller formula on the effective property of densely packed elastic composites"
}
| null | null | null | null | true | null |
1428
| null |
Default
| null | null |
null |
{
"abstract": " We give a simple optimistic algorithm for which it is easy to derive regret\nbounds of $\\tilde{O}(\\sqrt{t_{\\rm mix} SAT})$ after $T$ steps in uniformly\nergodic Markov decision processes with $S$ states, $A$ actions, and mixing time\nparameter $t_{\\rm mix}$. These bounds are the first regret bounds in the\ngeneral, non-episodic setting with an optimal dependence on all given\nparameters. They could only be improved by using an alternative mixing time\nparameter.\n",
"title": "Regret Bounds for Reinforcement Learning via Markov Chain Concentration"
}
| null | null |
[
"Statistics"
] | null | true | null |
1429
| null |
Validated
| null | null |
null |
{
"abstract": " We study the superradiant evolution of a set of $N$ two-level systems\nspontaneously radiating under the effect of phase-breaking mechanisms. We\ninvestigate the dynamics generated by non-radiative losses and pure dephasing,\nand their interplay with spontaneous emission. Our results show that in the\nparameter region relevant to many solid-state cavity quantum electrodynamics\nexperiments, even with a dephasing rate much faster than the radiative lifetime\nof a single two-level system, a sub-optimal collective superfluorescent burst\nis still observable. We also apply our theory to the dilute excitation regime,\noften used to describe optical excitations in solid-state systems. In this\nregime, excitations can be described in terms of bright and dark bosonic\nquasiparticles. We show how the effect of dephasing and losses in this regime\ntranslates into inter-mode scattering rates and quasiparticle lifetimes.\n",
"title": "Superradiance with local phase-breaking effects"
}
| null | null |
[
"Physics"
] | null | true | null |
1430
| null |
Validated
| null | null |
null |
{
"abstract": " We introduce a self-consistent multi-species kinetic theory based on the\nstructure of the narrow voltage-gated potassium channel. Transition rates\ndepend on a complete energy spectrum with contributions including the\ndehydration amongst species, interaction with the dipolar charge of the filter\nand, bulk solution properties. It displays high selectivity between species\ncoexisting with fast conductivity, and Coulomb blockade phenomena, and it fits\nwell to data.\n",
"title": "Kinetic model of selectivity and conductivity of the KcsA filter"
}
| null | null | null | null | true | null |
1431
| null |
Default
| null | null |
null |
{
"abstract": " In a recent paper, it was claimed that any homogeneous Finsler space of odd\ndimension admits a homogeneous geodesic through any point. For the proof, the\nalgebraic method dealing with the reductive decomposition of the Lie algebra of\nthe isometry group was used. However, the proof contains a serious gap. In the\npresent paper, homogeneous geodesics in Finsler homogeneous spaces are studied\nusing the affine method, which was developed in earlier papers by the author.\nThe mentioned statement is proved correctly and it is further proved that any\nhomogeneous Berwald space or homogeneous reversible Finsler space admits a\nhomogeneous geodesic through any point.\n",
"title": "The affine approach to homogeneous geodesics in homogeneous Finsler spaces"
}
| null | null | null | null | true | null |
1432
| null |
Default
| null | null |
null |
{
"abstract": " Observational and theoretical arguments support the idea that violent events\nconnected with $AGN$ activity and/or intense star forming episodes have played\na significant role in the early phases of galaxy formation at high red shifts.\nBeing old stellar systems, globular clusters seem adequate candidates to search\nfor the eventual signatures that might have been left by those energetic\nphenomena. The analysis of the colour distributions of several thousands of\nglobular clusters in the Virgo and Fornax galaxy clusters reveals the existence\nof some interesting and previously undetected features. A simple pattern\nrecognition technique, indicates the presence of \"colour modulations\",\ndistinctive for each galaxy cluster. The results suggest that the globular\ncluster formation process has not been completely stochastic but, rather,\nincluded a significant fraction of globulars that formed in a synchronized way\nand over supra-galactic spatial scales.\n",
"title": "About Synchronized Globular Cluster Formation over Supra-galactic Scales"
}
| null | null |
[
"Physics"
] | null | true | null |
1433
| null |
Validated
| null | null |
null |
{
"abstract": " We provide $L^p$-versus $L^\\infty$-bounds for eigenfunctions on a real\nspherical space $Z$ of wavefront type. It is shown that these bounds imply a\nnon-trivial error term estimate for lattice counting on $Z$. The paper also\nserves as an introduction to geometric counting on spaces of the mentioned\ntype. Section 7 on higher rank is new and extends the result from v1 to higher\nrank. Final version. To appear in Acta Math. Sinica.\n",
"title": "Geometric counting on wavefront real spherical spaces"
}
| null | null | null | null | true | null |
1434
| null |
Default
| null | null |
null |
{
"abstract": " We consider a strongly interacting quantum dot connected to two leads held at\nquite different temperatures. Our aim is to study the behavior of the Kondo\neffect in the presence of large thermal biases. We use three different\napproaches, namely, a perturbation formalism based on the Kondo Hamiltonian, a\nslave-boson mean-field theory for the Anderson model at large charging energies\nand a truncated equation-of-motion approach beyond the Hartree-Fock\napproximation. The two former formalisms yield a suppression of the Kondo peak\nfor thermal gradients above the Kondo temperature, showing a remarkably good\nagreement despite their different ranges of validity. The third technique\nallows us to analyze the full density of states within a wide range of\nenergies. Additionally, we have investigated the quantum transport properties\n(electric current and thermocurrent) beyond linear response. In the\nvoltage-driven case, we reproduce the split differential conductance due to the\npresence of different electrochemical potentials. In the temperature-driven\ncase, we observe a strongly nonlinear thermocurrent as a function of the\napplied thermal gradient. Depending on the parameters, we can find nontrivial\nzeros in the electric current for finite values of the temperature bias.\nImportantly, these thermocurrent zeros yield direct access to the system's\ncharacteristic energy scales (Kondo temperature and charging energy).\n",
"title": "Fate of the spin-\\frac{1}{2} Kondo effect in the presence of temperature gradients"
}
| null | null | null | null | true | null |
1435
| null |
Default
| null | null |
null |
{
"abstract": " The Cherenkov Telescope Array (CTA) is the next generation ground-based\n$\\gamma$-ray observatory. It will provide an order of magnitude better\nsensitivity and an extended energy coverage, 20 GeV - 300 TeV, relative to\ncurrent Imaging Atmospheric Cherenkov Telescopes (IACTs). IACTs, despite\nfeaturing an excellent sensitivity, are characterized by a limited field of\nview that makes the blind search of new sources very time inefficient.\nFortunately, the $\\textit{Fermi}$-LAT collaboration recently released a new\ncatalog of 1,556 sources detected in the 10 GeV - 2 TeV range by the Large Area\nTelescope (LAT) in the first 7 years of its operation (the 3FHL catalog). This\ncatalog is currently the most appropriate description of the sky that will be\nenergetically accessible to CTA. Here, we discuss a detailed analysis of the\nextragalactic source population (mostly blazars) that will be studied in the\nnear future by CTA. This analysis is based on simulations built from the\nexpected array configurations and information reported in the 3FHL catalog.\nThese results show the improvements that CTA will provide on the extragalactic\nTeV source population studies, which will be carried out by Key Science\nProjects as well as dedicated proposals.\n",
"title": "Extragalactic source population studies at very high energies in the Cherenkov Telescope Array era"
}
| null | null | null | null | true | null |
1436
| null |
Default
| null | null |
null |
{
"abstract": " The increase in customer expectation in terms of cost and services rendered,\ncoupled with competitive business environment and uncertainty in cost of raw\nmaterials have posed challenges on effective supply chain engineering making it\nessential to do cost-benefit analysis before making final decisions on\nproduction distribution logistics. This paper provides a conceptual model that\nprovide guidance in supply chain decision making for business expansion. It\npresents a mathematical model for production-distribution of an integrated\nsupply chain derived from current operations of SBC Tanzania Ltd which is a\nmajor supply chain that manages products' distribution in whole of Tanzania. In\naddition to finding the optimal cost, we also carried out a sensitivity\nanalysis on the model so as to find ways in which the company can expand at\noptimal cost, while meeting customers' demands. Genetic algorithms is used to\nrun the simulation for their efficient in solving combinatorial problems.\n",
"title": "Modeling the SBC Tanzania Production-Distribution Logistics Network"
}
| null | null | null | null | true | null |
1437
| null |
Default
| null | null |
null |
{
"abstract": " Although the cusp-core controversy for dwarf galaxies is seen as a problem, I\nargue that the cored central profiles can be explained by flattened cusps\nbecause they suffer from conflicting measurements and poor statistics and\nbecause there is a large number of conventional processes that could have\nflattened them since their creation, none of which requires new physics. Other\nproblems, such as \"too big to fail\", are not discussed.\n",
"title": "Dark matter in dwarf galaxies"
}
| null | null | null | null | true | null |
1438
| null |
Default
| null | null |
null |
{
"abstract": " A* is a best-first search algorithm for finding optimal-cost paths in graphs.\nA* benefits significantly from parallelism because in many applications, A* is\nlimited by memory usage, so distributed memory implementations of A* that use\nall of the aggregate memory on the cluster enable problems that can not be\nsolved by serial, single-machine implementations to be solved. We survey\napproaches to parallel A*, focusing on decentralized approaches to A* which\npartition the state space among processors. We also survey approaches to\nparallel, limited-memory variants of A* such as parallel IDA*.\n",
"title": "A Survey of Parallel A*"
}
| null | null | null | null | true | null |
1439
| null |
Default
| null | null |
null |
{
"abstract": " Integrated waveguides exhibiting efficient second-order nonlinearities are\ncrucial to obtain compact and low power optical signal processing devices.\nSilicon nitride (SiN) has shown second harmonic generation (SHG) capabilities\nin resonant structures and single-pass devices leveraging intermodal phase\nmatching, which is defined by waveguide design. Lithium niobate allows\ncompensating for the phase mismatch using periodically poled waveguides,\nhowever the latter are not reconfigurable and remain difficult to integrate\nwith SiN or silicon (Si) circuits. Here we show the all-optical enhancement of\nSHG in SiN waveguides by more than 30 dB. We demonstrate that a Watt-level\nlaser causes a periodic modification of the waveguide second-order\nsusceptibility. The resulting second order nonlinear grating has a periodicity\nallowing for quasi phase matching (QPM) between the pump and SH mode. Moreover,\nchanging the pump wavelength or polarization updates the period, relaxing phase\nmatching constraints imposed by the waveguide geometry. We show that the\ngrating is long term inscribed in the waveguides, and we estimate a second\norder nonlinearity of the order of 0.3 pm/V, while a maximum conversion\nefficiency (CE) of 1.8x10-6 W-1 cm-2 is reached.\n",
"title": "Large second harmonic generation enhancement in SiN waveguides by all-optically induced quasi phase matching"
}
| null | null | null | null | true | null |
1440
| null |
Default
| null | null |
null |
{
"abstract": " In this paper, we revisit the large-scale constrained linear regression\nproblem and propose faster methods based on some recent developments in\nsketching and optimization. Our algorithms combine (accelerated) mini-batch SGD\nwith a new method called two-step preconditioning to achieve an approximate\nsolution with a time complexity lower than that of the state-of-the-art\ntechniques for the low precision case. Our idea can also be extended to the\nhigh precision case, which gives an alternative implementation to the Iterative\nHessian Sketch (IHS) method with significantly improved time complexity.\nExperiments on benchmark and synthetic datasets suggest that our methods indeed\noutperform existing ones considerably in both the low and high precision cases.\n",
"title": "Large Scale Constrained Linear Regression Revisited: Faster Algorithms via Preconditioning"
}
| null | null |
[
"Statistics"
] | null | true | null |
1441
| null |
Validated
| null | null |
null |
{
"abstract": " We study the key domain wall properties in segmented nanowires loop-based\nstructures used in domain wall based sensors. The two reasons for device\nfailure, namely the distribution of domain wall propagation field (depinning)\nand the nucleation field are determined with Magneto-Optical Kerr Effect (MOKE)\nand Giant Magnetoresistance (GMR) measurements for thousands of elements to\nobtain significant statistics. Single layers of Ni$_{81}$Fe$_{19}$, a complete\nGMR stack with Co$_{90}$Fe$_{10}$/Ni$_{81}$Fe$_{19}$ as a free layer and a\nsingle layer of Co$_{90}$Fe$_{10}$ are deposited and industrially patterned to\ndetermine the influence of the shape anisotropy, the magnetocrystalline\nanisotropy and the fabrication processes. We show that the propagation field is\nlittle influenced by the geometry but significantly by material parameters. The\ndomain wall nucleation fields can be described by a typical Stoner-Wohlfarth\nmodel related to the measured geometrical parameters of the wires and fitted by\nconsidering the process parameters. The GMR effect is subsequently measured in\na substantial number of devices (3000), in order to accurately gauge the\nvariation between devices. This reveals a corrected upper limit to the\nnucleation fields of the sensors that can be exploited for fast\ncharacterization of working elements.\n",
"title": "Geometrical dependence of domain wall propagation and nucleation fields in magnetic domain wall sensor devices"
}
| null | null | null | null | true | null |
1442
| null |
Default
| null | null |
null |
{
"abstract": " This article improves the existing proven rates of regret decay in optimal\npolicy estimation. We give a margin-free result showing that the regret decay\nfor estimating a within-class optimal policy is second-order for empirical risk\nminimizers over Donsker classes, with regret decaying at a faster rate than the\nstandard error of an efficient estimator of the value of an optimal policy. We\nalso give a result from the classification literature that shows that faster\nregret decay is possible via plug-in estimation provided a margin condition\nholds. Four examples are considered. In these examples, the regret is expressed\nin terms of either the mean value or the median value; the number of possible\nactions is either two or finitely many; and the sampling scheme is either\nindependent and identically distributed or sequential, where the latter\nrepresents a contextual bandit sampling scheme.\n",
"title": "Faster Rates for Policy Learning"
}
| null | null |
[
"Mathematics",
"Statistics"
] | null | true | null |
1443
| null |
Validated
| null | null |
null |
{
"abstract": " We investigate the magnetic properties of the multiferroic quantum-spin\nsystem LiCu$_2$O$_2$ by electron spin resonance (ESR) measurements at $X$- and\n$Q$-band frequencies in a wide temperature range $(T_{\\rm N1} \\leq T \\leq\n300$\\,K). The observed anisotropies of the $g$ tensor and the ESR linewidth in\nuntwinned single crystals result from the crystal-electric field and from local\nexchange geometries acting on the magnetic Cu$^{2+}$ ions in the zigzag-ladder\nlike structure of LiCu$_2$O$_2$. Supported by a microscopic analysis of the\nexchange paths involved, we show that both the symmetric anisotropic exchange\ninteraction and the antisymmetric Dzyaloshinskii-Moriya interaction provide the\ndominant spin-spin relaxation channels in this material.\n",
"title": "Anisotropic Exchange in ${\\bf LiCu_2O_2}$"
}
| null | null |
[
"Physics"
] | null | true | null |
1444
| null |
Validated
| null | null |
null |
{
"abstract": " The friendship paradox states that in a social network, egos tend to have\nlower degree than their alters, or, \"your friends have more friends than you\ndo\". Most research has focused on the friendship paradox and its implications\nfor information transmission, but treating the network as static and\nunweighted. Yet, people can dedicate only a finite fraction of their attention\nbudget to each social interaction: a high-degree individual may have less time\nto dedicate to individual social links, forcing them to modulate the quantities\nof contact made to their different social ties. Here we study the friendship\nparadox in the context of differing contact volumes between egos and alters,\nfinding a connection between contact volume and the strength of the friendship\nparadox. The most frequently contacted alters exhibit a less pronounced\nfriendship paradox compared with the ego, whereas less-frequently contacted\nalters are more likely to be high degree and give rise to the paradox. We argue\ntherefore for a more nuanced version of the friendship paradox: \"your closest\nfriends have slightly more friends than you do\", and in certain networks even:\n\"your best friend has no more friends than you do\". We demonstrate that this\nrelationship is robust, holding in both a social media and a mobile phone\ndataset. These results have implications for information transfer and influence\nin social networks, which we explore using a simple dynamical model.\n",
"title": "Which friends are more popular than you? Contact strength and the friendship paradox in social networks"
}
| null | null | null | null | true | null |
1445
| null |
Default
| null | null |
null |
{
"abstract": " Many stochastic optimization algorithms work by estimating the gradient of\nthe cost function on the fly by sampling datapoints uniformly at random from a\ntraining set. However, the estimator might have a large variance, which\ninadvertently slows down the convergence rate of the algorithms. One way to\nreduce this variance is to sample the datapoints from a carefully selected\nnon-uniform distribution. In this work, we propose a novel non-uniform sampling\napproach that uses the multi-armed bandit framework. Theoretically, we show\nthat our algorithm asymptotically approximates the optimal variance within a\nfactor of 3. Empirically, we show that using this datapoint-selection technique\nresults in a significant reduction in the convergence time and variance of\nseveral stochastic optimization algorithms such as SGD, SVRG and SAGA. This\napproach for sampling datapoints is general, and can be used in conjunction\nwith any algorithm that uses an unbiased gradient estimation -- we expect it to\nhave broad applicability beyond the specific examples explored in this work.\n",
"title": "Stochastic Optimization with Bandit Sampling"
}
| null | null | null | null | true | null |
1446
| null |
Default
| null | null |
null |
{
"abstract": " Robust reinforcement learning aims to produce policies that have strong\nguarantees even in the face of environments/transition models whose parameters\nhave strong uncertainty. Existing work uses value-based methods and the usual\nprimitive action setting. In this paper, we propose robust methods for learning\ntemporally abstract actions, in the framework of options. We present a Robust\nOptions Policy Iteration (ROPI) algorithm with convergence guarantees, which\nlearns options that are robust to model uncertainty. We utilize ROPI to learn\nrobust options with the Robust Options Deep Q Network (RO-DQN) that solves\nmultiple tasks and mitigates model misspecification due to model uncertainty.\nWe present experimental results which suggest that policy iteration with linear\nfeatures may have an inherent form of robustness when using coarse feature\nrepresentations. In addition, we present experimental results which demonstrate\nthat robustness helps policy iteration implemented on top of deep neural\nnetworks to generalize over a much broader range of dynamics than non-robust\npolicy iteration.\n",
"title": "Learning Robust Options"
}
| null | null | null | null | true | null |
1447
| null |
Default
| null | null |
null |
{
"abstract": " The central theme of this work is that a stable levitation of a denser\nnon-magnetizable liquid droplet, against gravity, inside a relatively lighter\nferrofluid -- a system barely considered in ferrohydrodynamics -- is possible,\nand exhibits unique interfacial features; the stability of the levitation\ntrajectory, however, is subject to an appropriate magnetic field modulation. We\nexplore the shapes and the temporal dynamics of a plane non-magnetizable\ndroplet levitating inside ferrofluid against gravity due to a spatially\ncomplex, but systematically generated, magnetic field in two dimensions. The\neffect of the viscosity ratio, the stability of the levitation path and the\npossibility of existence of multiple-stable equilibrium states is investigated.\nWe find, for certain conditions on the viscosity ratio, that there can be\ndevelopments of cusps and singularities at the droplet surface; this phenomenon\nwe also observe experimentally and compared with the simulations. Our\nsimulations closely replicate the singular projection on the surface of the\nlevitating droplet. Finally, we present an dynamical model for the vertical\ntrajectory of the droplet. This model reveals a condition for the onset of\nlevitation and the relation for the equilibrium levitation height. The\nlinearization of the model around the steady state captures that the nature of\nthe equilibrium point goes under a transition from being a spiral to a node\ndepending upon the control parameters, which essentially means that the\ntemporal route to the equilibrium can be either monotonic or undulating. The\nanalytical model for the droplet trajectory is in close agreement with the\ndetailed simulations. (See draft for full abstract).\n",
"title": "Levitation of non-magnetizable droplet inside ferrofluid"
}
| null | null | null | null | true | null |
1448
| null |
Default
| null | null |
null |
{
"abstract": " We present NMR spectra of remote-magnetized deuterated water, detected in an\nunshielded environment by means of a differential atomic magnetometer. The\nmeasurements are performed in a $\\mu$T field, while pulsed techniques are\napplied -following the sample displacement- in a 100~$\\mu$T field, to tip both\nD and H nuclei by controllable amounts. The broadband nature of the detection\nsystem enables simultaneous detection of the two signals and accurate\nevaluation of their decay times. The outcomes of the experiment demonstrate the\npotential of ultra-low-field NMR spectroscopy in important applications where\nthe correlation between proton and deuteron spin-spin relaxation rates as a\nfunction of external parameters contains significant information.\n",
"title": "Simultaneous Detection of H and D NMR Signals in a micro-Tesla Field"
}
| null | null | null | null | true | null |
1449
| null |
Default
| null | null |
null |
{
"abstract": " Large datasets often have unreliable labels-such as those obtained from\nAmazon's Mechanical Turk or social media platforms-and classifiers trained on\nmislabeled datasets often exhibit poor performance. We present a simple,\neffective technique for accounting for label noise when training deep neural\nnetworks. We augment a standard deep network with a softmax layer that models\nthe label noise statistics. Then, we train the deep network and noise model\njointly via end-to-end stochastic gradient descent on the (perhaps mislabeled)\ndataset. The augmented model is overdetermined, so in order to encourage the\nlearning of a non-trivial noise model, we apply dropout regularization to the\nweights of the noise model during training. Numerical experiments on noisy\nversions of the CIFAR-10 and MNIST datasets show that the proposed dropout\ntechnique outperforms state-of-the-art methods.\n",
"title": "Learning Deep Networks from Noisy Labels with Dropout Regularization"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
1450
| null |
Validated
| null | null |
null |
{
"abstract": " Modern networks are of huge sizes as well as high dynamics, which challenges\nthe efficiency of community detection algorithms. In this paper, we study the\nproblem of overlapping community detection on distributed and dynamic graphs.\nGiven a distributed, undirected and unweighted graph, the goal is to detect\noverlapping communities incrementally as the graph is dynamically changing. We\npropose an efficient algorithm, called \\textit{randomized Speaker-Listener\nLabel Propagation Algorithm} (rSLPA), based on the \\textit{Speaker-Listener\nLabel Propagation Algorithm} (SLPA) by relaxing the probability distribution of\nlabel propagation. Besides detecting high-quality communities, rSLPA can\nincrementally update the detected communities after a batch of edge insertion\nand deletion operations. To the best of our knowledge, rSLPA is the first\nalgorithm that can incrementally capture the same communities as those obtained\nby applying the detection algorithm from the scratch on the updated graph.\nExtensive experiments are conducted on both synthetic and real-world datasets,\nand the results show that our algorithm can achieve high accuracy and\nefficiency at the same time.\n",
"title": "On Efficiently Detecting Overlapping Communities over Distributed Dynamic Graphs"
}
| null | null | null | null | true | null |
1451
| null |
Default
| null | null |
null |
{
"abstract": " Continuous latent time series models are prevalent in Bayesian modeling;\nexamples include the Kalman filter, dynamic collaborative filtering, or dynamic\ntopic models. These models often benefit from structured, non mean field\nvariational approximations that capture correlations between time steps. Black\nbox variational inference with reparameterization gradients (BBVI) allows us to\nexplore a rich new class of Bayesian non-conjugate latent time series models;\nhowever, a naive application of BBVI to such structured variational models\nwould scale quadratically in the number of time steps. We describe a BBVI\nalgorithm analogous to the forward-backward algorithm which instead scales\nlinearly in time. It allows us to efficiently sample from the variational\ndistribution and estimate the gradients of the ELBO. Finally, we show results\non the recently proposed dynamic word embedding model, which was trained using\nour method.\n",
"title": "Structured Black Box Variational Inference for Latent Time Series Models"
}
| null | null | null | null | true | null |
1452
| null |
Default
| null | null |
null |
{
"abstract": " We prove upper bounds on the $L^p$ norms of eigenfunctions of the discrete\nLaplacian on regular graphs. We then apply these ideas to study the $L^p$ norms\nof joint eigenfunctions of the Laplacian and an averaging operator over a\nfinite collection of algebraic rotations of the $2$-sphere. Under mild\nconditions, such joint eigenfunctions are shown to satisfy for large $p$ the\nsame bounds as those known for Laplace eigenfunctions on a surface of\nnon-positive curvature.\n",
"title": "$L^p$ Norms of Eigenfunctions on Regular Graphs and on the Sphere"
}
| null | null |
[
"Mathematics"
] | null | true | null |
1453
| null |
Validated
| null | null |
null |
{
"abstract": " A key resource for distributed quantum-enhanced protocols is entanglement\nbetween spatially separated modes. Yet, the robust generation and detection of\nnonlocal entanglement between spatially separated regions of an ultracold\natomic system remains a challenge. Here, we use spin mixing in a tightly\nconfined Bose-Einstein condensate to generate an entangled state of\nindistinguishable particles in a single spatial mode. We show experimentally\nthat this local entanglement can be spatially distributed by self-similar\nexpansion of the atomic cloud. Spatially resolved spin read-out is used to\nreveal a particularly strong form of quantum correlations known as\nEinstein-Podolsky-Rosen steering between distinct parts of the expanded cloud.\nBased on the strength of Einstein-Podolsky-Rosen steering we construct a\nwitness, which testifies up to genuine five-partite entanglement.\n",
"title": "Spatially distributed multipartite entanglement enables Einstein-Podolsky-Rosen steering of atomic clouds"
}
| null | null | null | null | true | null |
1454
| null |
Default
| null | null |
null |
{
"abstract": " Most end devices are now equipped with multiple network interfaces.\nApplications can exploit all available interfaces and benefit from multipath\ntransmission. Recently Multipath TCP (MPTCP) was proposed to implement\nmultipath transmission at the transport layer and has attracted lots of\nattention from academia and industry. However, MPTCP only supports TCP-based\napplications and its multipath routing flexibility is limited. In this paper,\nwe investigate the possibility of orchestrating multipath transmission from the\nnetwork layer of end devices, and develop a Multipath IP (MPIP) design\nconsisting of signaling, session and path management, multipath routing, and\nNAT traversal. We implement MPIP in Linux and Android kernels. Through\ncontrolled lab experiments and Internet experiments, we demonstrate that MPIP\ncan effectively achieve multipath gains at the network layer. It not only\nsupports the legacy TCP and UDP protocols, but also works seamlessly with\nMPTCP. By facilitating user-defined customized routing, MPIP can route traffic\nfrom competing applications in a coordinated fashion to maximize the aggregate\nuser Quality-of-Experience.\n",
"title": "Multipath IP Routing on End Devices: Motivation, Design, and Performance"
}
| null | null | null | null | true | null |
1455
| null |
Default
| null | null |
null |
{
"abstract": " In this paper we show how the defense relation among abstract arguments can\nbe used to encode the reasons for accepting arguments. After introducing a\nnovel notion of defenses and defense graphs, we propose a defense semantics\ntogether with a new notion of defense equivalence of argument graphs, and\ncompare defense equivalence with standard equivalence and strong equivalence,\nrespectively. Then, based on defense semantics, we define two kinds of reasons\nfor accepting arguments, i.e., direct reasons and root reasons, and a notion of\nroot equivalence of argument graphs. Finally, we show how the notion of root\nequivalence can be used in argumentation summarization.\n",
"title": "Defense semantics of argumentation: encoding reasons for accepting arguments"
}
| null | null |
[
"Computer Science"
] | null | true | null |
1456
| null |
Validated
| null | null |
null |
{
"abstract": " While optimizing convex objective (loss) functions has been a powerhouse for\nmachine learning for at least two decades, non-convex loss functions have\nattracted fast growing interests recently, due to many desirable properties\nsuch as superior robustness and classification accuracy, compared with their\nconvex counterparts. The main obstacle for non-convex estimators is that it is\nin general intractable to find the optimal solution. In this paper, we study\nthe computational issues for some non-convex M-estimators. In particular, we\nshow that the stochastic variance reduction methods converge to the global\noptimal with linear rate, by exploiting the statistical property of the\npopulation loss. En route, we improve the convergence analysis for the batch\ngradient method in \\cite{mei2016landscape}.\n",
"title": "Fast Global Convergence via Landscape of Empirical Loss"
}
| null | null |
[
"Statistics"
] | null | true | null |
1457
| null |
Validated
| null | null |
null |
{
"abstract": " A photodetector may be characterized by various figures of merit such as\nresponse time, bandwidth, dark count rate, efficiency, wavelength resolution,\nand photon-number resolution. On the other hand, quantum theory says that any\nmeasurement device is fully described by its POVM, which stands for\nPositive-Operator-Valued Measure, and which generalizes the textbook notion of\nthe eigenstates of the appropriate hermitian operator (the \"observable\") as\nmeasurement outcomes. Here we show how to define a multitude of photodetector\nfigures of merit in terms of a given POVM. We distinguish classical and quantum\nfigures of merit and issue a conjecture regarding trade-off relations between\nthem. We discuss the relationship between POVM elements and photodetector\nclicks, and how models of photodetectors may be tested by measuring either POVM\nelements or figures of merit. Finally, the POVM is advertised as a\nplatform-independent way of comparing different types of photodetectors, since\nany such POVM refers to the Hilbert space of the incoming light, and not to any\nHilbert space internal to the detector.\n",
"title": "Photodetector figures of merit in terms of POVMs"
}
| null | null | null | null | true | null |
1458
| null |
Default
| null | null |
null |
{
"abstract": " All living systems can function only far away from equilibrium, and for this\nreason chemical kinetic methods are critically important for uncovering the\nmechanisms of biological processes. Here we present a new theoretical method of\ninvestigating dynamics of protein-DNA interactions, which govern all major\nbiological processes. It is based on a first-passage analysis of biochemical\nand biophysical transitions, and it provides a fully analytic description of\nthe processes. Our approach is explained for the case of a single protein\nsearching for a specific binding site on DNA. In addition, the application of\nthe method to investigations of the effect of DNA sequence heterogeneity, and\nthe role multiple targets and traps in the protein search dynamics are\ndiscussed.\n",
"title": "Kinetics of Protein-DNA Interactions: First-Passage Analysis"
}
| null | null | null | null | true | null |
1459
| null |
Default
| null | null |
null |
{
"abstract": " We numerically study jamming transitions in pedestrian flow interacting with\nan attraction, mostly based on the social force model for pedestrians who can\njoin the attraction. We formulate the joining probability as a function of\nsocial influence from others, reflecting that individual choice behavior is\nlikely influenced by others. By controlling pedestrian influx and the social\ninfluence parameter, we identify various pedestrian flow patterns. For the\nbidirectional flow scenario, we observe a transition from the free flow phase\nto the freezing phase, in which oppositely walking pedestrians reach a complete\nstop and block each other. On the other hand, a different transition behavior\nappears in the unidirectional flow scenario, i.e., from the free flow phase to\nthe localized jam phase and then to the extended jam phase. It is also observed\nthat the extended jam phase can end up in freezing phenomena with a certain\nprobability when pedestrian flux is high with strong social influence. This\nstudy highlights that attractive interactions between pedestrians and an\nattraction can trigger jamming transitions by increasing the number of\nconflicts among pedestrians near the attraction. In order to avoid excessive\npedestrian jams, we suggest suppressing the number of conflicts under a certain\nlevel by moderating pedestrian influx especially when the social influence is\nstrong.\n",
"title": "Jamming transitions induced by an attraction in pedestrian flow"
}
| null | null | null | null | true | null |
1460
| null |
Default
| null | null |
null |
{
"abstract": " Survival analysis has been developed and applied in the number of areas\nincluding manufacturing, finance, economics and healthcare. In healthcare\ndomain, usually clinical data are high-dimensional, sparse and complex and\nsometimes there exists few amount of time-to-event (labeled) instances.\nTherefore building an accurate survival model from electronic health records is\nchallenging. With this motivation, we address this issue and provide a new\nsurvival analysis framework using deep learning and active learning with a\nnovel sampling strategy. First, our approach provides better representation\nwith lower dimensions from clinical features using labeled (time-to-event) and\nunlabeled (censored) instances and then actively trains the survival model by\nlabeling the censored data using an oracle. As a clinical assistive tool, we\nintroduce a simple effective treatment recommendation approach based on our\nsurvival model. In the experimental study, we apply our approach on\nSEER-Medicare data related to prostate cancer among African-Americans and white\npatients. The results indicate that our approach outperforms significantly than\nbaseline models.\n",
"title": "A Deep Active Survival Analysis Approach for Precision Treatment Recommendations: Application of Prostate Cancer"
}
| null | null | null | null | true | null |
1461
| null |
Default
| null | null |
null |
{
"abstract": " The study of time-varying (dynamic) networks (graphs) is of fundamental\nimportance for computer network analytics. Several methods have been proposed\nto detect the effect of significant structural changes in a time series of\ngraphs. The main contribution of this work is a detailed analysis of a dynamic\ncommunity graph model. This model is formed by adding new vertices, and\nrandomly attaching them to the existing nodes. It is a dynamic extension of the\nwell-known stochastic blockmodel. The goal of the work is to detect the time at\nwhich the graph dynamics switches from a normal evolution -- where balanced\ncommunities grow at the same rate -- to an abnormal behavior -- where\ncommunities start merging. In order to circumvent the problem of decomposing\neach graph into communities, we use a metric to quantify changes in the graph\ntopology as a function of time. The detection of anomalies becomes one of\ntesting the hypothesis that the graph is undergoing a significant structural\nchange. In addition the the theoretical analysis of the test statistic, we\nperform Monte Carlo simulations of our dynamic graph model to demonstrate that\nour test can detect changes in graph topology.\n",
"title": "Detecting Topological Changes in Dynamic Community Networks"
}
| null | null | null | null | true | null |
1462
| null |
Default
| null | null |
null |
{
"abstract": " We consider the multi-label ranking approach to multi-label learning.\nBoosting is a natural method for multi-label ranking as it aggregates weak\npredictions through majority votes, which can be directly used as scores to\nproduce a ranking of the labels. We design online boosting algorithms with\nprovable loss bounds for multi-label ranking. We show that our first algorithm\nis optimal in terms of the number of learners required to attain a desired\naccuracy, but it requires knowledge of the edge of the weak learners. We also\ndesign an adaptive algorithm that does not require this knowledge and is hence\nmore practical. Experimental results on real data sets demonstrate that our\nalgorithms are at least as good as existing batch boosting algorithms.\n",
"title": "Online Boosting Algorithms for Multi-label Ranking"
}
| null | null | null | null | true | null |
1463
| null |
Default
| null | null |
null |
{
"abstract": " Wave-particle duality in quantum mechanics allows for a halo bound state\nwhose spatial extension far exceeds a range of the interaction potential. What\nis even more striking is that such quantum halos can be arbitrarily large on\nspecial occasions. The two examples known so far are the Efimov effect and the\nsuper Efimov effect, which predict that spatial extensions of higher excited\nstates grow exponentially and double exponentially, respectively. Here, we\nestablish yet another new class of arbitrarily large quantum halos formed by\nspinless bosons with short-range interactions in two dimensions. When the\ntwo-body interaction is absent but the three-body interaction is resonant, four\nbosons exhibit an infinite tower of bound states whose spatial extensions scale\nas $R_n\\sim e^{(\\pi n)^2/27}$ for a large $n$. The emergent scaling law is\nuniversal and is termed a semisuper Efimov effect, which together with the\nEfimov and super Efimov effects constitutes a trio of few-body universality\nclasses allowing for arbitrarily large quantum halos.\n",
"title": "Semisuper Efimov effect of two-dimensional bosons at a three-body resonance"
}
| null | null | null | null | true | null |
1464
| null |
Default
| null | null |
null |
{
"abstract": " We prove a quantitative Fourth Moment Theorem for Wigner integrals of any\norder with symmetric kernels, generalizing an earlier result from Kemp et al.\n(2012). The proof relies on free stochastic analysis and uses a new biproduct\nformula for bi-integrals. A consequence of our main result is a\nNualart-Ortiz-Latorre type characterization of convergence in law to the\nsemicircular distribution for Wigner integrals. As an application, we provide\nBerry-Esseen type bounds in the context of the free Breuer-Major theorem for\nthe free fractional Brownian motion.\n",
"title": "Free quantitative fourth moment theorems on Wigner space"
}
| null | null | null | null | true | null |
1465
| null |
Default
| null | null |
null |
{
"abstract": " Generative Adversarial Networks (GANs) have been shown to be able to sample\nimpressively realistic images. GAN training consists of a saddle point\noptimization problem that can be thought of as an adversarial game between a\ngenerator which produces the images, and a discriminator, which judges if the\nimages are real. Both the generator and the discriminator are commonly\nparametrized as deep convolutional neural networks. The goal of this paper is\nto disentangle the contribution of the optimization procedure and the network\nparametrization to the success of GANs. To this end we introduce and study\nGenerative Latent Optimization (GLO), a framework to train a generator without\nthe need to learn a discriminator, thus avoiding challenging adversarial\noptimization problems. We show experimentally that GLO enjoys many of the\ndesirable properties of GANs: learning from large data, synthesizing\nvisually-appealing samples, interpolating meaningfully between samples, and\nperforming linear arithmetic with noise vectors.\n",
"title": "Optimizing the Latent Space of Generative Networks"
}
| null | null | null | null | true | null |
1466
| null |
Default
| null | null |
null |
{
"abstract": " We show that the l-adic realization functor is conservative when restricted\nto the Chow motives of abelian type over a finite field. A weak version of this\nconservativity result extends to mixed motives of abelian type.\n",
"title": "Conservativity of realizations on motives of abelian type over finite fields"
}
| null | null | null | null | true | null |
1467
| null |
Default
| null | null |
null |
{
"abstract": " Software startups face with multiple technical and business challenges, which\ncould make the startup journey longer, or even become a failure. Little is\nknown about entrepreneurial decision making as a direct force to startup\ndevelopment outcome. In this study, we attempted to apply a behaviour theory of\nentrepreneurial firms to understand the root-cause of some software startup s\nchallenges. Six common challenges related to prototyping and product\ndevelopment in twenty software startups were identified. We found the behaviour\ntheory as a useful theoretical lens to explain the technical challenges.\nSoftware startups search for local optimal solutions, emphasise on short-run\nfeedback rather than long-run strategies, which results in vague prototype\nplanning, paradox of demonstration and evolving throw-away prototypes. The\nfinding implies that effectual entrepreneurial processes might require a more\nsuitable product development approach than the current state-of-practice.\n",
"title": "Towards understanding startup product development as effectual entrepreneurial behaviors"
}
| null | null | null | null | true | null |
1468
| null |
Default
| null | null |
null |
{
"abstract": " We show that a generalized Dirac structure survives beyond the linear regime\nof the low-energy dispersion relations of graphene. A generalized uncertainty\nprinciple of the kind compatible with specific quantum gravity scenarios with a\nfundamental minimal length (here graphene lattice spacing) and Lorentz\nviolation (here the particle/hole asymmetry, the trigonal warping, etc.) is\nnaturally obtained. We then show that the corresponding emergent field theory\nis a table-top realization of such scenarios, by explicitly computing the third\norder Hamiltonian, and giving the general recipe for any order. Remarkably, our\nresults imply that going beyond the low-energy approximation does not spoil the\nwell-known correspondence with analogue massless quantum electrodynamics\nphenomena (as usually believed), but rather it is a way to obtain experimental\nsignatures of quantum-gravity-like corrections to such phenomena.\n",
"title": "Generalized Dirac structure beyond the linear regime in graphene"
}
| null | null | null | null | true | null |
1469
| null |
Default
| null | null |
null |
{
"abstract": " A generative model based on training deep architectures is proposed. The\nmodel consists of K networks that are trained together to learn the underlying\ndistribution of a given data set. The process starts with dividing the input\ndata into K clusters and feeding each of them into a separate network. After\nfew iterations of training networks separately, we use an EM-like algorithm to\ntrain the networks together and update the clusters of the data. We call this\nmodel Mixture of Networks. The provided model is a platform that can be used\nfor any deep structure and be trained by any conventional objective function\nfor distribution modeling. As the components of the model are neural networks,\nit has high capability in characterizing complicated data distributions as well\nas clustering data. We apply the algorithm on MNIST hand-written digits and\nYale face datasets. We also demonstrate the clustering ability of the model\nusing some real-world and toy examples.\n",
"title": "Generative Mixture of Networks"
}
| null | null |
[
"Computer Science",
"Statistics"
] | null | true | null |
1470
| null |
Validated
| null | null |
null |
{
"abstract": " Single magnetic skyrmions are localized whirls in the magnetization with an\ninteger winding number. They have been observed on nano-meter scales up to room\ntemperature in multilayer structures. Due to their small size, topological\nwinding number, and their ability to be manipulated by extremely tiny forces,\nthey are often called interesting candidates for future memory devices. The\ntwo-lane racetrack has to exhibit two lanes that are separated by an energy\nbarrier. The information is then encoded in the position of a skyrmion which is\nlocated in one of these close-by lanes. The artificial barrier between the\nlanes can be created by an additional nanostrip on top of the track. Here we\nstudy the dependence of the potential barrier on the shape of the additional\nnanostrip, calculating the potentials for a rectangular, triangular, and\nparabolic cross section, as well as interpolations between the first two. We\nfind that a narrow barrier is always repulsive and that the height of the\npotential strongly depends on the shape of the nanostrip, whereas the shape of\nthe potential is more universal. We finally show that the shape-dependence is\nredundant for possible applications.\n",
"title": "Shape-dependence of the barrier for skyrmions on a two-lane racetrack"
}
| null | null | null | null | true | null |
1471
| null |
Default
| null | null |
null |
{
"abstract": " We complement the theory developed in Preinerstorfer and Pötscher (2016)\nwith further finite sample results on size and power of heteroskedasticity and\nautocorrelation robust tests. These allows us, in particular, to show that the\nsufficient conditions for the existence of size-controlling critical values\nrecently obtained in Pötscher and Preinerstorfer (2016) are often also\nnecessary. We furthermore apply the results obtained to tests for hypotheses on\ndeterministic trends in stationary time series regressions, and find that many\ntests currently used are strongly size-distorted.\n",
"title": "Further Results on Size and Power of Heteroskedasticity and Autocorrelation Robust Tests, with an Application to Trend Testing"
}
| null | null | null | null | true | null |
1472
| null |
Default
| null | null |
null |
{
"abstract": " Effect modification means the magnitude or stability of a treatment effect\nvaries as a function of an observed covariate. Generally, larger and more\nstable treatment effects are insensitive to larger biases from unmeasured\ncovariates, so a causal conclusion may be considerably firmer if this pattern\nis noted if it occurs. We propose a new strategy, called the submax-method,\nthat combines exploratory and confirmatory efforts to determine whether there\nis stronger evidence of causality - that is, greater insensitivity to\nunmeasured confounding - in some subgroups of individuals. It uses the joint\ndistribution of test statistics that split the data in various ways based on\ncertain observed covariates. For $L$ binary covariates, the method splits the\npopulation $L$ times into two subpopulations, perhaps first men and women,\nperhaps then smokers and nonsmokers, computing a test statistic from each\nsubpopulation, and appends the test statistic for the whole population, making\n$2L+1$ test statistics in total. Although $L$ binary covariates define $2^{L}$\ninteraction groups, only $2L+1$ tests are performed, and at least $L+1$ of\nthese tests use at least half of the data. The submax-method achieves the\nhighest design sensitivity and the highest Bahadur efficiency of its component\ntests. Moreover, the form of the test is sufficiently tractable that its large\nsample power may be studied analytically. The simulation suggests that the\nsubmax method exhibits superior performance, in comparison with an approach\nusing CART, when there is effect modification of moderate size. Using data from\nthe NHANES I Epidemiologic Follow-Up Survey, an observational study of the\neffects of physical activity on survival is used to illustrate the method. The\nmethod is implemented in the $\\texttt{R}$ package $\\texttt{submax}$ which\ncontains the NHANES example.\n",
"title": "A powerful approach to the study of moderate effect modification in observational studies"
}
| null | null | null | null | true | null |
1473
| null |
Default
| null | null |
null |
{
"abstract": " Many internet ventures rely on advertising for their revenue. However, users\nfeel discontent by the presence of ads on the websites they visit, as the\ndata-size of ads is often comparable to that of the actual content. This has an\nimpact not only on the loading time of webpages, but also on the internet bill\nof the user in some cases. In absence of a mutually-agreed procedure for opting\nout of advertisements, many users resort to ad-blocking browser-extensions. In\nthis work, we study the performance of popular ad-blockers on a large set of\nnews websites. Moreover, we investigate the benefits of ad-blockers on user\nprivacy as well as the mechanisms used by websites to counter them. Finally, we\nexplore the traffic overhead due to the ad-blockers themselves.\n",
"title": "Ad-blocking: A Study on Performance, Privacy and Counter-measures"
}
| null | null |
[
"Computer Science"
] | null | true | null |
1474
| null |
Validated
| null | null |
null |
{
"abstract": " Calculating the value of $C^{k\\in\\{1,\\infty\\}}$ class of smoothness\nreal-valued function's derivative in point of $\\mathbb{R}^+$ in radius of\nconvergence of its Taylor polynomial (or series), applying an analog of\nNewton's binomial theorem and $q$-difference operator. $(P,q)$-power difference\nintroduced in section 5. Additionally, by means of Newton's interpolation\nformula, the discrete analog of Taylor series, interpolation using\n$q$-difference and $p,q$-power difference is shown.\n",
"title": "On the quantum differentiation of smooth real-valued functions"
}
| null | null |
[
"Mathematics"
] | null | true | null |
1475
| null |
Validated
| null | null |
null |
{
"abstract": " Let $P$ and $Q$ be two convex polytopes both contained in the interior of an\nEuclidean ball $r\\textbf{B}^{d}$. We prove that $P=Q$ provided that their sight\ncones from any point on the sphere $rS^{d-1}$ are congruent. We also prove an\nanalogous result for spherical projections.\n",
"title": "On recognizing shapes of polytopes from their shadows"
}
| null | null |
[
"Mathematics"
] | null | true | null |
1476
| null |
Validated
| null | null |
null |
{
"abstract": " Existence of steady states in elastic media at small strains with diffusion\nof a solvent or fluid due to Fick's or Darcy's laws is proved by combining\nusage of variational methods inspired from static situations with Schauder's\nfixed-point arguments. In the plain variant, the problem consists in the force\nequilibrium coupled with the continuity equation, and the underlying operator\nis non-potential and non-pseudomonotone so that conventional methods are not\napplicable. In advanced variants, electrically-charged multi-component flows\nthrough an electrically charged elastic solid are treated, employing critical\npoints of the saddle-point type. Eventually, anisothermal variants involving\nheat-transfer equation are treated, too.\n",
"title": "Variational methods for steady-state Darcy/Fick flow in swollen and poroelastic solids"
}
| null | null | null | null | true | null |
1477
| null |
Default
| null | null |
null |
{
"abstract": " The field of plasma-based particle accelerators has seen tremendous progress\nover the past decade and experienced significant growth in the number of\nactivities. During this process, the involved scientific community has expanded\nfrom traditional university-based research and is now encompassing many large\nresearch laboratories worldwide, such as BNL, CERN, DESY, KEK, LBNL and SLAC.\nAs a consequence, there is a strong demand for a consolidated effort in\neducation at the intersection of accelerator, laser and plasma physics. The\nCERN Accelerator School on Plasma Wake Acceleration has been organized as a\nresult of this development. In this paper, we describe the interactive\ncomponent of this one-week school, which consisted of three case studies to be\nsolved in 11 working groups by the participants of the CERN Accelerator School.\n",
"title": "Case Studies on Plasma Wakefield Accelerator Design"
}
| null | null | null | null | true | null |
1478
| null |
Default
| null | null |
null |
{
"abstract": " In this paper, we propose a novel application of Generative Adversarial\nNetworks (GAN) to the synthesis of cells imaged by fluorescence microscopy.\nCompared to natural images, cells tend to have a simpler and more geometric\nglobal structure that facilitates image generation. However, the correlation\nbetween the spatial pattern of different fluorescent proteins reflects\nimportant biological functions, and synthesized images have to capture these\nrelationships to be relevant for biological applications. We adapt GANs to the\ntask at hand and propose new models with casual dependencies between image\nchannels that can generate multi-channel images, which would be impossible to\nobtain experimentally. We evaluate our approach using two independent\ntechniques and compare it against sensible baselines. Finally, we demonstrate\nthat by interpolating across the latent space we can mimic the known changes in\nprotein localization that occur through time during the cell cycle, allowing us\nto predict temporal evolution from static images.\n",
"title": "GANs for Biological Image Synthesis"
}
| null | null | null | null | true | null |
1479
| null |
Default
| null | null |
null |
{
"abstract": " A clustering algorithm is applied to Cassini Imaging Science Subsystem\ncontinuum and methane band images of Saturns northern hemisphere to objectively\nclassify regional albedo features and aid in their dynamical interpretation.\nThe procedure is based on a technique applied previously to visible-infrared\nimages of Earth. It provides a new perspective on giant planet cloud morphology\nand its relationship to the dynamics and a meteorological context for the\nanalysis of other types of simultaneous Saturn observations. The method\nidentifies six clusters that exhibit distinct morphology, vertical structure,\nand preferred latitudes of occurrence. These correspond to areas dominated by\ndeep convective cells; low contrast areas, some including thinner and thicker\nclouds possibly associated with baroclinic instability; regions with possible\nisolated thin cirrus clouds; darker areas due to thinner low level clouds or\nclearer skies due to downwelling, or due to absorbing particles; and fields of\nrelatively shallow cumulus clouds. The spatial associations among these cloud\ntypes suggest that dynamically, there are three distinct types of latitude\nbands on Saturn: deep convectively disturbed latitudes in cyclonic shear\nregions poleward of the eastward jets; convectively suppressed regions near and\nsurrounding the westward jets; and baroclinically unstable latitudes near\neastward jet cores and in the anti-cyclonic regions equatorward of them. These\nare roughly analogous to some of the features of Earths tropics, subtropics,\nand midlatitudes, respectively. Temporal variations of feature contrast and\ncluster occurrence suggest that the upper tropospheric haze in the northern\nhemisphere may have thickened by 2014.\n",
"title": "An objective classification of Saturn cloud features from Cassini ISS images"
}
| null | null | null | null | true | null |
1480
| null |
Default
| null | null |
null |
{
"abstract": " We re-examine the notion of stress in peridynamics. Based on the idea of\ntraction we define two new peridynamic stress tensors $\\mathbf{P}^{\\mathbf{y}}$\nand $\\mathbf{P}$ which stand, respectively, for analogues of the Cauchy and 1st\nPiola-Kirchhoff stress tensors from classical elasticity. We show that the\ntensor $\\mathbf{P}$ differs from the earlier defined peridynamic stress tensor\n$\\nu$; though their divergence is equal. We address the question of symmetry of\nthe tensor $\\mathbf{P}^{\\mathbf{y}}$ which proves to be symmetric in case of\nbond-based peridynamics; as opposed to the inverse Piola transform of $\\nu$\n(corresponding to the analogue of Cauchy stress tensor) which fails to be\nsymmetric in general. We also derive a general formula of the force-flux in\nperidynamics and compute the limit of $\\mathbf{P}$ for vanishing non-locality,\ndenoted by $\\mathbf{P}_0$. We show that this tensor $\\mathbf{P}_0$ surprisingly\ncoincides with the collapsed tensor $\\nu_0$, a limit of the original tensor\n$\\nu$. At the end, using this flux-formula, we suggest an explanation why the\ncollapsed tensor $\\mathbf{P}_0$ (and hence $\\nu_0$) can be indeed identified\nwith the 1st Piola-Kirchhoff stress tensor.\n",
"title": "The Peridynamic Stress Tensors and the Non-local to Local Passage"
}
| null | null | null | null | true | null |
1481
| null |
Default
| null | null |
null |
{
"abstract": " Successful human-robot cooperation hinges on each agent's ability to process\nand exchange information about the shared environment and the task at hand.\nHuman communication is primarily based on symbolic abstractions of object\nproperties, rather than precise quantitative measures. A comprehensive robotic\nframework thus requires an integrated communication module which is able to\nestablish a link and convert between perceptual and abstract information.\nThe ability to interpret composite symbolic descriptions enables an\nautonomous agent to a) operate in unstructured and cluttered environments, in\ntasks which involve unmodeled or never seen before objects; and b) exploit the\naggregation of multiple symbolic properties as an instance of ensemble\nlearning, to improve identification performance even when the individual\npredicates encode generic information or are imprecisely grounded.\nWe propose a discriminative probabilistic model which interprets symbolic\ndescriptions to identify the referent object contextually w.r.t.\\ the structure\nof the environment and other objects. The model is trained using a collected\ndataset of identifications, and its performance is evaluated by quantitative\nmeasures and a live demo developed on the PR2 robot platform, which integrates\nelements of perception, object extraction, object identification and grasping.\n",
"title": "Identification of Unmodeled Objects from Symbolic Descriptions"
}
| null | null | null | null | true | null |
1482
| null |
Default
| null | null |
null |
{
"abstract": " We present a prototype for a news search engine that presents balanced\nviewpoints across liberal and conservative articles with the goal of\nde-polarizing content and allowing users to escape their filter bubble. The\nbalancing is done according to flexible user-defined constraints, and leverages\nrecent advances in constrained bandit optimization. We showcase our balanced\nnews feed by displaying it side-by-side with the news feed produced by a\ntraditional (polarized) feed.\n",
"title": "Balanced News Using Constrained Bandit-based Personalization"
}
| null | null | null | null | true | null |
1483
| null |
Default
| null | null |
null |
{
"abstract": " Models of complex systems are widely used in the physical and social\nsciences, and the concept of layering, typically building upon graph-theoretic\nstructure, is a common feature. We describe an intuitionistic substructural\nlogic called ILGL that gives an account of layering. The logic is a bunched\nsystem, combining the usual intuitionistic connectives, together with a\nnon-commutative, non-associative conjunction (used to capture layering) and its\nassociated implications. We give soundness and completeness theorems for a\nlabelled tableaux system with respect to a Kripke semantics on graphs. We then\ngive an equivalent relational semantics, itself proven equivalent to an\nalgebraic semantics via a representation theorem. We utilise this result in two\nways. First, we prove decidability of the logic by showing the finite\nembeddability property holds for the algebraic semantics. Second, we prove a\nStone-type duality theorem for the logic. By introducing the notions of ILGL\nhyperdoctrine and indexed layered frame we are able to extend this result to a\npredicate version of the logic and prove soundness and completeness theorems\nfor an extension of the layered graph semantics . We indicate the utility of\npredicate ILGL with a resource-labelled bigraph model.\n",
"title": "Intuitionistic Layered Graph Logic: Semantics and Proof Theory"
}
| null | null | null | null | true | null |
1484
| null |
Default
| null | null |
null |
{
"abstract": " Color names based image representation is successfully used in person\nre-identification, due to the advantages of being compact, intuitively\nunderstandable as well as being robust to photometric variance. However, there\nexists the diversity between underlying distribution of color names' RGB values\nand that of image pixels' RGB values, which may lead to inaccuracy when\ndirectly comparing them in Euclidean space. In this paper, we propose a new\nmethod named soft Gaussian mapping (SGM) to address this problem. We model the\ndiscrepancies between color names and pixels using a Gaussian and utilize the\ninverse of covariance matrix to bridge the gap between them. Based on SGM, an\nimage could be converted to several soft Gaussian maps. In each soft Gaussian\nmap, we further seek to establish stable and robust descriptors within a local\nregion through a max pooling operation. Then, a robust image representation\nbased on color names is obtained by concatenating the statistical descriptors\nin each stripe. When labeled data are available, one discriminative subspace\nprojection matrix is learned to build efficient representations of an image via\ncross-view coupling learning. Experiments on the public datasets - VIPeR,\nPRID450S and CUHK03, demonstrate the effectiveness of our method.\n",
"title": "Learning Efficient Image Representation for Person Re-Identification"
}
| null | null | null | null | true | null |
1485
| null |
Default
| null | null |
null |
{
"abstract": " This PhD thesis is devoted to the low-energy structure of the nucleon (proton\nand neutron) as seen through electromagnetic probes, e.g., electron and Compton\nscattering. The research presented here is based primarily on dispersion theory\nand chiral effective-field theory. The main motivation is the recent proton\nradius puzzle, which is the discrepancy between the classic proton charge\nradius determinations (based on electron-proton scattering and normal hydrogen\nspectroscopy) and the highly precise extraction based on first muonic-hydrogen\nexperiments by the CREMA Collaboration. The precision of muonic-hydrogen\nexperiments is presently limited by the knowledge of proton structure effects\nbeyond the charge radius. A major part of this thesis is devoted to calculating\nthese effects using everything we know about the nucleon electromagnetic\nstructure from both theory and experiment.\nThe thesis consists of eight chapters. The first and last are, respectively,\nthe introduction and conclusion. The remainder of this thesis can roughly be\ndivided into the following three topics: finite-size effects in hydrogen-like\natoms, real and virtual Compton scattering, and two-photon-exchange effects.\n",
"title": "Exciting Nucleons in Compton Scattering and Hydrogen-Like Atoms"
}
| null | null | null | null | true | null |
1486
| null |
Default
| null | null |
null |
{
"abstract": " Phase transitions in isotropic quantum antiferromagnets are associated with\nthe condensation of bosonic triplet excitations. In three dimensional quantum\nantiferromagnets, such as TlCuCl$_3$, condensation can be either pressure or\nmagnetic field induced. The corresponding magnetic order obeys universal\nscaling with thermal critical exponent $\\phi$. Employing a relativistic quantum\nfield theory, the present work predicts the emergence of multiple (three)\nuniversalities under combined pressure and field tuning. Changes of\nuniversality are signalled by changes of the critical exponent $\\phi$.\nExplicitly, we predict the existence of two new exponents $\\phi=1$ and $1/2$ as\nwell as recovering the known exponent $\\phi=3/2$. We also predict logarithmic\ncorrections to the power law scaling.\n",
"title": "Multiple universalities in order-disorder magnetic phase transitions"
}
| null | null | null | null | true | null |
1487
| null |
Default
| null | null |
null |
{
"abstract": " From philosophers of ancient times to modern economists, biologists and other\nresearchers are engaged in revealing causal relations. The most challenging\nproblem is inferring the type of the causal relationship: whether it is uni- or\nbi-directional or only apparent - implied by a hidden common cause only. Modern\ntechnology provides us tools to record data from complex systems such as the\necosystem of our planet or the human brain, but understanding their functioning\nneeds detection and distinction of causal relationships of the system\ncomponents without interventions. Here we present a new method, which\ndistinguishes and assigns probabilities to the presence of all the possible\ncausal relations between two or more time series from dynamical systems. The\nnew method is validated on synthetic datasets and applied to EEG\n(electroencephalographic) data recorded in epileptic patients. Given the\nuniversality of our method, it may find application in many fields of science.\n",
"title": "Exact Inference of Causal Relations in Dynamical Systems"
}
| null | null | null | null | true | null |
1488
| null |
Default
| null | null |
null |
{
"abstract": " Deep neural networks are increasingly being used in a variety of machine\nlearning applications applied to rich user data on the cloud. However, this\napproach introduces a number of privacy and efficiency challenges, as the cloud\noperator can perform secondary inferences on the available data. Recently,\nadvances in edge processing have paved the way for more efficient, and private,\ndata processing at the source for simple tasks and lighter models, though they\nremain a challenge for larger, and more complicated models. In this paper, we\npresent a hybrid approach for breaking down large, complex deep models for\ncooperative, privacy-preserving analytics. We do this by breaking down the\npopular deep architectures and fine-tune them in a particular way. We then\nevaluate the privacy benefits of this approach based on the information exposed\nto the cloud service. We also asses the local inference cost of different\nlayers on a modern handset for mobile applications. Our evaluations show that\nby using certain kind of fine-tuning and embedding techniques and at a small\nprocessing costs, we can greatly reduce the level of information available to\nunintended tasks applied to the data feature on the cloud, and hence achieving\nthe desired tradeoff between privacy and performance.\n",
"title": "Privacy-Preserving Deep Inference for Rich User Data on The Cloud"
}
| null | null |
[
"Computer Science"
] | null | true | null |
1489
| null |
Validated
| null | null |
null |
{
"abstract": " In this paper, we develop new first-order method for composite non-convex\nminimization problems with simple constraints and inexact oracle. The objective\nfunction is given as a sum of \"`hard\"', possibly non-convex part, and\n\"`simple\"' convex part. Informally speaking, oracle inexactness means that, for\nthe \"`hard\"' part, at any point we can approximately calculate the value of the\nfunction and construct a quadratic function, which approximately bounds this\nfunction from above. We give several examples of such inexactness: smooth\nnon-convex functions with inexact Hölder-continuous gradient, functions given\nby auxiliary uniformly concave maximization problem, which can be solved only\napproximately. For the introduced class of problems, we propose a gradient-type\nmethod, which allows to use different proximal setup to adapt to geometry of\nthe feasible set, adaptively chooses controlled oracle error, allows for\ninexact proximal mapping. We provide convergence rate for our method in terms\nof the norm of generalized gradient mapping and show that, in the case of\ninexact Hölder-continuous gradient, our method is universal with respect to\nHölder parameters of the problem. Finally, in a particular case, we show that\nsmall value of the norm of generalized gradient mapping at a point means that a\nnecessary condition of local minimum approximately holds at that point.\n",
"title": "Gradient Method With Inexact Oracle for Composite Non-Convex Optimization"
}
| null | null | null | null | true | null |
1490
| null |
Default
| null | null |
null |
{
"abstract": " Recent progress in variational inference has paid much attention to the\nflexibility of variational posteriors. One promising direction is to use\nimplicit distributions, i.e., distributions without tractable densities as the\nvariational posterior. However, existing methods on implicit posteriors still\nface challenges of noisy estimation and computational infeasibility when\napplied to models with high-dimensional latent variables. In this paper, we\npresent a new approach named Kernel Implicit Variational Inference that\naddresses these challenges. As far as we know, for the first time implicit\nvariational inference is successfully applied to Bayesian neural networks,\nwhich shows promising results on both regression and classification tasks.\n",
"title": "Kernel Implicit Variational Inference"
}
| null | null | null | null | true | null |
1491
| null |
Default
| null | null |
null |
{
"abstract": " The ADR algebra $R_A$ of a finite-dimensional algebra $A$ is a\nquasihereditary algebra. In this paper we study the Ringel dual\n$\\mathcal{R}(R_A)$ of $R_A$. We prove that $\\mathcal{R}(R_A)$ can be identified\nwith $(R_{A^{op}})^{op}$, under certain 'minimal' regularity conditions for\n$A$. We also give necessary and sufficient conditions for the ADR algebra to be\nRingel selfdual.\n",
"title": "The Ringel dual of the Auslander-Dlab-Ringel algebra"
}
| null | null | null | null | true | null |
1492
| null |
Default
| null | null |
null |
{
"abstract": " We study the structure of the $(\\mathfrak{g},K)$-modules of the principal\nseries representations of $SL(3,\\mathbb{R})$ and $Sp(2,\\mathbb{R})$ induced\nfrom minimal parabolic subgroups, in the case when the infinitesimal character\nis nonsingular. The composition factors of these modules are known by\nKazhdan-Lusztig-Vogan conjecture. In this paper, we give complete descriptions\nof the socle filtrations of these modules.\n",
"title": "The socle filtrations of principal series representations of $SL(3,\\mathbb{R})$ and $Sp(2,\\mathbb{R})$"
}
| null | null |
[
"Mathematics"
] | null | true | null |
1493
| null |
Validated
| null | null |
null |
{
"abstract": " We study theoretically and experimentally the influence of temporally shaping\nthe light pulses in an atom interferometer, with a focus on the phase response\nof the interferometer. We show that smooth light pulse shapes allow rejecting\nhigh frequency phase fluctuations (above the Rabi frequency) and thus relax the\nrequirements on the phase noise or frequency noise of the interrogation lasers\ndriving the interferometer. The light pulse shape is also shown to modify the\nscale factor of the interferometer, which has to be taken into account in the\nevaluation of its accuracy budget. We discuss the trade-offs to operate when\nchoosing a particular pulse shape, by taking into account phase noise\nrejection, velocity selectivity, and applicability to large momentum transfer\natom interferometry.\n",
"title": "Improving the phase response of an atom interferometer by means of temporal pulse shaping"
}
| null | null | null | null | true | null |
1494
| null |
Default
| null | null |
null |
{
"abstract": " It is shown that the non-relativistic ground state energy of helium-like and\nlithium-like ions with static nuclei can be interpolated in full physics range\nof nuclear charges $Z$ with accuracy of not less than 6 decimal digits (d.d.)\nor 7-8 significant digits (s.d.) using a meromorphic function in appropriate\nvariable with a few free parameters. It is demonstrated that finite nuclear\nmass effects do not change 4-5 s.d. for $Z \\in [1,50]$ for 2-,3-electron\nsystems and the leading relativistic and QED corrections leave unchanged 3-4\ns.d. for $Z \\in [1,12]$ in the ground state energy for 2-electron system, thus,\nthe interpolation reproduces definitely those figures. A meaning of proposed\ninterpolation is in a construction of unified, {\\it two-point} Pade approximant\n(for both small and large $Z$ expansions) with fitting some parameters at\nintermediate $Z$.\n",
"title": "Helium-like and Lithium-like ions: Ground state energy"
}
| null | null | null | null | true | null |
1495
| null |
Default
| null | null |
null |
{
"abstract": " Traditional data cleaning identifies dirty data by classifying original data\nsequences, which is a class$-$imbalanced problem since the proportion of\nincorrect data is much less than the proportion of correct ones for most\ndiagnostic systems in Magnetic Confinement Fusion (MCF) devices. When using\nmachine learning algorithms to classify diagnostic data based on\nclass$-$imbalanced training set, most classifiers are biased towards the major\nclass and show very poor classification rates on the minor class. By\ntransforming the direct classification problem about original data sequences\ninto a classification problem about the physical similarity between data\nsequences, the class$-$balanced effect of Time$-$Domain Global Similarity\n(TDGS) method on training set structure is investigated in this paper.\nMeanwhile, the impact of improved training set structure on data cleaning\nperformance of TDGS method is demonstrated with an application example in EAST\nPOlarimetry$-$INTerferometry (POINT) system.\n",
"title": "Improvement of training set structure in fusion data cleaning using Time-Domain Global Similarity method"
}
| null | null | null | null | true | null |
1496
| null |
Default
| null | null |
null |
{
"abstract": " Three complementary methods have been implemented in the code Denovo that\naccelerate neutral particle transport calculations with methods that use\nleadership-class computers fully and effectively: a multigroup block (MG)\nKrylov solver, a Rayleigh Quotient Iteration (RQI) eigenvalue solver, and a\nmultigrid in energy (MGE) preconditioner. The MG Krylov solver converges more\nquickly than Gauss Seidel and enables energy decomposition such that Denovo can\nscale to hundreds of thousands of cores. RQI should converge in fewer\niterations than power iteration (PI) for large and challenging problems. RQI\ncreates shifted systems that would not be tractable without the MG Krylov\nsolver. It also creates ill-conditioned matrices. The MGE preconditioner\nreduces iteration count significantly when used with RQI and takes advantage of\nthe new energy decomposition such that it can scale efficiently. Each\nindividual method has been described before, but this is the first time they\nhave been demonstrated to work together effectively.\nThe combination of solvers enables the RQI eigenvalue solver to work better\nthan the other available solvers for large reactors problems on leadership\nclass machines. Using these methods together, RQI converged in fewer iterations\nand in less time than PI for a full pressurized water reactor core. These\nsolvers also performed better than an Arnoldi eigenvalue solver for a reactor\nbenchmark problem when energy decomposition is needed. The MG Krylov, MGE\npreconditioner, and RQI solver combination also scales well in energy. This\nsolver set is a strong choice for very large and challenging problems.\n",
"title": "Eigenvalue Solvers for Modeling Nuclear Reactors on Leadership Class Machines"
}
| null | null | null | null | true | null |
1497
| null |
Default
| null | null |
null |
{
"abstract": " The thermoregulation system in animals removes body heat in hot temperatures\nand retains body heat in cold temperatures. The better the animal removes heat,\nthe worse the animal retains heat and visa versa. It is the balance between\nthese two conflicting goals that determines the mammal's size, heart rate and\namount of hair. The rat's loss of tail hair and human's loss of its body hair\nare responses to these conflicting thermoregulation needs as these animals\nevolved to larger size over time.\n",
"title": "Thermoregulation in mice, rats and humans: An insight into the evolution of human hairlessness"
}
| null | null | null | null | true | null |
1498
| null |
Default
| null | null |
null |
{
"abstract": " We introduce a notion of Koszul A-infinity algebra that generalizes Priddy's\nnotion of a Koszul algebra and we use it to construct small A-infinity algebra\nmodels for Hochschild cochains. As an application, this yields new techniques\nfor computing free loop space homology algebras of manifolds that are either\nformal or coformal (over a field or over the integers). We illustrate these\ntechniques in two examples.\n",
"title": "Koszul A-infinity algebras and free loop space homology"
}
| null | null | null | null | true | null |
1499
| null |
Default
| null | null |
null |
{
"abstract": " By exploiting the property that the RBM log-likelihood function is the\ndifference of convex functions, we formulate a stochastic variant of the\ndifference of convex functions (DC) programming to minimize the negative\nlog-likelihood. Interestingly, the traditional contrastive divergence algorithm\nis a special case of the above formulation and the hyperparameters of the two\nalgorithms can be chosen such that the amount of computation per mini-batch is\nidentical. We show that for a given computational budget the proposed algorithm\nalmost always reaches a higher log-likelihood more rapidly, compared to the\nstandard contrastive divergence algorithm. Further, we modify this algorithm to\nuse the centered gradients and show that it is more efficient and effective\ncompared to the standard centered gradient algorithm on benchmark datasets.\n",
"title": "Learning RBM with a DC programming Approach"
}
| null | null | null | null | true | null |
1500
| null |
Default
| null | null |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.