Top arXiv papers

sign in to customize
  • PDF
    Person re-identification (Re-ID) aims at recognizing the same person from images taken across different cameras. To address this task, one typically requires a large amount labeled data for training an effective Re-ID model, which might not be practical for real-world applications. To alleviate this limitation, we choose to exploit a sufficient amount of pre-existing labeled data from a different (auxiliary) dataset. By jointly considering such an auxiliary dataset and the dataset of interest (but without label information), our proposed adaptation and re-identification network (ARN) performs unsupervised domain adaptation, which leverages information across datasets and derives domain-invariant features for Re-ID purposes. In our experiments, we verify that our network performs favorably against state-of-the-art unsupervised Re-ID approaches, and even outperforms a number of baseline Re-ID methods which require fully supervised data for training.
  • PDF
    We generalize the notion of fixed point homogeneous isometric group actions to the context of singular Riemannian foliations. We find that in some cases, positively curved manifolds admitting these so-called point leaf maximal SRF's are diffeo/homeomorphic to compact rank one symmetric spaces. In all cases, manifolds admitting such foliations are cohomology CROSSes or finite quotients of them. Among non-simply connected manifolds, we find examples of such foliations which are non-homogeneous
  • PDF
    In this paper we provide a broad benchmarking of recent genetic programming approaches to symbolic regression in the context of state of the art machine learning approaches. We use a set of nearly 100 regression benchmark problems culled from open source repositories across the web. We conduct a rigorous benchmarking of four recent symbolic regression approaches as well as nine machine learning approaches from scikit-learn. The results suggest that symbolic regression performs strongly compared to state-of-the-art gradient boosting algorithms, although in terms of running times is among the slowest of the available methodologies. We discuss the results in detail and point to future research directions that may allow symbolic regression to gain wider adoption in the machine learning community.
  • PDF
    To process a large volume of data, modern data management systems use a collection of machines connected through a network. This paper looks into the feasibility of scaling up such a shared-nothing system while processing a compute- and communication-intensive workload---processing distributed joins. By exploiting multiple processing cores within the individual machines, we implement a system to process database joins that parallelizes computation within each node, pipelines the computation with communication, parallelizes the communication by allowing multiple simultaneous data transfers (send/receive), and removes synchronization barriers (a scalability bottleneck in a distributed data processing system). Our experimental results show that using only four threads per node the framework achieves a 3.5x gains in intra-node performance while compared with a single-threaded counterpart. Moreover, with the join processing workload the cluster-wide performance (and speedup) is observed to be dictated by the intra-node computational loads; this property brings a near-linear speedup with increasing nodes in the system, a feature much desired in modern large-scale data processing system.
  • PDF
    Pre-screening of ship proposals is now employed by top ship detectors to avoid exhaustive search across image. In very high resolution (VHR) optical image, ships appeared as a cluster of abnormal bright pixels in open sea clutter (noise-like background). Anomaly-based detector utilizing Panchromatic (PAN) data has been widely used in many researches to detect ships, however, still facing two main drawbacks: 1) detection rate tend to be low particularly when a ship is low contrast and 2) these models require a high manual configuration to select a threshold value best separate ships from sea surface background. This paper aims at further investigation of anomaly-based model to solve those issues. First, pan-sharpened Multi Spectral (MS) data is incorporated together with PAN to enhance ship discrimination. Second, we propose an improved anomaly-based model combining both global intensity anomaly and local texture anomaly map. Regarding noise appeared due to the present of sea clutter and because of pan-sharpen process, texture abnormality suppression term based on quantization theory is introduced. Experimental results on VNREDSat-1 VHR optical satellite images suggest that the pan-sharpened near-infrared (P-NIR) band can improve discrimination of ships from surrounding waters. Compared to state-of-the-art anomaly-based detectors, our proposed anomaly-based model on the combination of PAN and P-NIR data cannot only achieved highest ship detection's recall rate (91.14% and 45.9% on high-contrast and low-contrast dataset respectively) but also robust to different automatic threshold selection techniques.
  • PDF
    Every lawsuit document contains the information about the party's claim, court's analysis, decision and others, and all of this information are helpful to understand the case better and predict the judge's decision on similar case in the future. However, the extraction of these information from the document is difficult because the language is too complicated and sentences varied at length. We treat this problem as a task of sequence labeling, and this paper presents the first research to extract relevant information from the civil lawsuit document in China with the hierarchical RNN framework.
  • PDF
    Deep learning searches for nonlinear factors for predicting asset returns. Predictability is achieved via multiple layers of composite factors as opposed to additive ones. Viewed in this way, asset pricing studies can be revisited using multi-layer deep learners, such as rectified linear units (ReLU) or long-short-term-memory (LSTM) for time-series effects. State-of-the-art algorithms including stochastic gradient descent (SGD), TensorFlow and dropout design provide imple- mentation and efficient factor exploration. To illustrate our methodology, we revisit the equity market risk premium dataset of Welch and Goyal (2008). We find the existence of nonlinear factors which explain predictability of returns, in particular at the extremes of the characteristic space. Finally, we conclude with directions for future research.
  • PDF
    The morphology of planetary nebulae emerging from the common envelope phase of binary star evolution is investigated. Using initial conditions based on the numerical results of hydrodynamical simulations of the common envelope phase it is found that the shapes and sizes of the resulting nebula are very sensitive to the effective temperature of the remnant core, the mass-loss rate at the onset of the common envelope phase, and the mass ratio of the binary system. These parameters are related to the efficiency of the mass ejection after the spiral-in phase, the stellar evolutionary phase (i.e., RG, AGB or TP-AGB), and the degree of departure from spherical symmetry in the stellar wind mass loss process itself respectively. It is found that the shapes are mostly bipolar in the early phase of evolution, but can quickly transition to elliptical and barrel-type shapes. Solutions for nested lobes are found where the outer lobes are usually bipolar and the inner lobes are elliptical, bipolar or barrel-type, a result due to the flow of the photo-evaporated gas from the equatorial region. It is found that the lobes can be produced without the need for two distinct mass ejection events. In all the computations, the bulk of the mass is concentrated in the orbital or equatorial plane, in the form of a large toroid, which can be either neutral (early phases) or photoionized (late phases), depending of the evolutionary state of the system.
  • PDF
    To efficiently evaluate system reliability based on Monte Carlo simulation, importance sampling is used widely. The optimal importance sampling density was derived in 1950s for the deterministic simulation model, which maps an input to an output deterministically, and is approximated in practice using various methods. For the stochastic simulation model whose output is random given an input, the optimal importance sampling density was derived only recently. In the existing literature, metamodel-based approaches have been used to approximate this optimal density. However, building a satisfactory metamodel is often difficult or time-consuming in practice. This paper proposes a cross-entropy based method, which is automatic and does not require specific domain knowledge. The proposed method uses an expectation--maximization algorithm to guide the choice of a mixture distribution model for approximating the optimal density. The method iteratively updates the approximated density to minimize its estimated discrepancy, measured by estimated cross-entropy, from the optimal density. The mixture model's complexity is controlled using the cross-entropy information criterion. The method is empirically validated using a numerical study and applied to a case study of evaluating the reliability of wind turbine using a stochastic simulation model.
  • PDF
    We present an empirical study of gender bias in coreference resolution systems. We first introduce a novel, Winograd schema-style set of minimal pair sentences that differ only by pronoun gender. With these "Winogender schemas," we evaluate and confirm systematic gender bias in three publicly-available coreference resolution systems, and correlate this bias with real-world and textual gender statistics.
  • PDF
    Neural Sequence-to-Sequence models have proven to be accurate and robust for many sequence prediction tasks, and have become the standard approach for automatic translation of text. The models work in a five stage blackbox process that involves encoding a source sequence to a vector space and then decoding out to a new target sequence. This process is now standard, but like many deep learning methods remains quite difficult to understand or debug. In this work, we present a visual analysis tool that allows interaction with a trained sequence-to-sequence model through each stage of the translation process. The aim is to identify which patterns have been learned and to detect model errors. We demonstrate the utility of our tool through several real-world large-scale sequence-to-sequence use cases.
  • PDF
    In this paper, we summarize recent progresses made in deep learning based acoustic models and the motivation and insights behind the surveyed techniques. We first discuss acoustic models that can effectively exploit variable-length contextual information, such as recurrent neural networks (RNNs), convolutional neural networks (CNNs), and their various combination with other models. We then describe acoustic models that are optimized end-to-end with emphasis on feature representations learned jointly with rest of the system, the connectionist temporal classification (CTC) criterion, and the attention-based sequence-to-sequence model. We further illustrate robustness issues in speech recognition systems, and discuss acoustic model adaptation, speech enhancement and separation, and robust training strategies. We also cover modeling techniques that lead to more efficient decoding and discuss possible future directions in acoustic model research.
  • PDF
    Audio content analysis in terms of sound events is an important research problem for a variety of applications. Recently, the development of weak labeling approaches for audio or sound event detection (AED) and availability of large scale weakly labeled dataset have finally opened up the possibility of large scale AED. However, a deeper understanding of how weak labels affect the learning for sound events is still missing from literature. In this work, we first describe a CNN based approach for weakly supervised training of audio events. The approach follows some basic design principle desirable in a learning method relying on weakly labeled audio. We then describe important characteristics, which naturally arise in weakly supervised learning of sound events. We show how these aspects of weak labels affect the generalization of models. More specifically, we study how characteristics such as label density and corruption of labels affects weakly supervised training for audio events. We also study the feasibility of directly obtaining weak labeled data from the web without any manual label and compare it with a dataset which has been manually labeled. The analysis and understanding of these factors should be taken into picture in the development of future weak label learning methods. Audioset, a large scale weakly labeled dataset for sound events is used in our experiments.
  • PDF
    We present a simulation code which can solve broad ranges of partial differential equations in a full sphere. The code expands tensorial variables in a spectral series of spin-weighted spherical harmonics in the angular directions and a scaled Jacobi polynomial basis in the radial direction, as described in Part-I. Nonlinear terms are calculated by transforming from the coefficients in the spectral series to the value of each quantity on the physical grid, where it is easy to calculate products and perform other local operations. The expansion makes it straightforward to solve equations in tensor form (i.e., without decomposition into scalars). We propose and study several unit tests which demonstrate the code can accurately solve linear problems, implement boundary conditions, and transform between spectral and physical space. We then run a series of benchmark problems proposed in Marti et al (2014), implementing the hydrodynamic and magnetohydrodynamic equations. We are able to calculate more accurate solutions than reported in Marti et al 2014 by running at higher spatial resolution and using a higher-order timestepping scheme. We find the rotating convection and convective dynamo benchmark problems depend sensitively on details of timestepping and data analysis. We also demonstrate that in low resolution simulations of the dynamo problem, small changes in a numerical scheme can lead to large changes in the solution. To aid future comparison to these benchmarks, we include the source code used to generate the data, as well as the data and analysis scripts used to generate the figures.
  • PDF
    This paper presents segmentation-free strategies for the recognition of handwritten numeral strings of unknown length. A synthetic dataset of touching numeral strings of sizes 2-, 3- and 4-digits was created to train end-to-end solutions based on Convolutional Neural Networks. A robust experimental protocol is used to show that the proposed segmentation-free methods may reach the state-of-the-art performance without suffering the heavy burden of over-segmentation based methods. In addition, they confirmed the importance of introducing contextual information in the design of end-to-end solutions, such as the proposed length classifier when recognizing numeral strings.
  • PDF
    In the absence of global positioning information, place recognition is a key capability for enabling localization, mapping and navigation in any environment. Most place recognition methods rely on images, point clouds, or a combination of both. In this work we leverage a segment extraction and matching approach to achieve place recognition in Light Detection and Ranging (LiDAR) based 3D point cloud maps. One challenge related to this approach is the recognition of segments despite changes in point of view or occlusion. We propose using a learning based method in order to reach a higher recall accuracy then previously proposed methods. Using Convolutional Neural Networks (CNNs), which are state-of-the-art classifiers, we propose a new approach to segment recognition based on learned descriptors. In this paper we compare the effectiveness of three different structures and training methods for CNNs. We demonstrate through several experiments on real-world data collected in an urban driving scenario that the proposed learning based methods outperform hand-crafted descriptors.
  • PDF
    Reduced Order Modeling (ROM) for engineering applications has been a major research focus in the past few decades due to the unprecedented physical insight into turbulence offered by high-fidelity CFD. The primary goal of a ROM is to model the key physics/features of a flow-field without computing the full Navier-Stokes (NS) equations. This is accomplished by projecting the high-dimensional dynamics to a low-dimensional subspace, typically utilizing dimensionality reduction techniques like Proper Orthogonal Decomposition (POD), coupled with Galerkin projection. In this work, we demonstrate a deep learning based approach to build a ROM using the POD basis of canonical DNS datasets, for turbulent flow control applications. We find that a type of Recurrent Neural Network, the Long Short Term Memory (LSTM) which has been primarily utilized for problems like speech modeling and language translation, shows attractive potential in modeling temporal dynamics of turbulence. Additionally, we introduce the Hurst Exponent as a tool to study LSTM behavior for non-stationary data, and uncover useful characteristics that may aid ROM development for a variety of applications.
  • PDF
    The proliferation of advanced information technologies (IT), especially the wide spread of Internet of Things (IoTs) makes wireless spectrum a precious resource. Cognitive radio network (CRN) has been recognized as the key to achieve efficient utility of communication bands. Because of the great difficulty, high complexity and regulations in dynamic spectrum access (DSA), it is very challenging to protect CRNs from malicious attackers or selfish abusers. Primary user emulation (PUE) attacks is one type of easy-to-launch but hard-to-detect attacks in CRNs that malicious entities mimic PU signals in order to either occupy spectrum resource selfishly or conduct Denial of Service (DoS) attacks. Inspired by the physical features widely used as the fingerprint of variant electronic devices, an adaptive and realistic PUE attack detection technique is proposed in this paper. It leverages the PU transmission features that attackers are not able to mimic. In this work, the transmission power is selected as one of the hard-to-mimic features due to the intrinsic discrepancy between PUs and attackers, while considering constraints in real implementations. Our experimental results verified the effectiveness and correctness of the proposed mechanism.
  • PDF
    Commonsense knowledge bases such as ConceptNet represent knowledge in the form of relational triples. Inspired by the recent work by Li et al., we analyse if knowledge base completion models can be used to mine commonsense knowledge from raw text. We propose novelty of predicted triples with respect to the training set as an important factor in interpreting results. We critically analyse the difficulty of mining novel commonsense knowledge, and show that a simple baseline method outperforms the previous state of the art on predicting more novel.
  • PDF
    We propose a novel approach for loss reserving based on deep neural networks. The approach allows for jointly modeling of paid losses and claims outstanding, and incorporation of heterogenous inputs. We validate the models on loss reserving data across lines of business, and show that they attain or exceed the predictive accuracy of existing stochastic methods. The models require minimal feature engineering and expert input, and can be automated to produce forecasts at a high frequency.
  • PDF
    We propose a technique for declaratively specifying strategies for semi-supervised learning (SSL). The proposed method can be used to specify ensembles of semi-supervised learning, as well as agreement constraints and entropic regularization constraints between these learners, and can be used to model both well-known heuristics such as co-training and novel domain-specific heuristics. In addition to representing individual SSL heuristics, we show that multiple heuristics can also be automatically combined using Bayesian optimization methods. We show consistent improvements on a suite of well-studied SSL benchmarks, including a new state-of-the-art result on a difficult relation extraction task.
  • PDF
    We study a two-level impurity coupled locally to a quantum gas on an optical lattice. For state-dependent interactions between the impurity and the gas, we show that its evolution encodes information on the local excitation spectrum of gas at the coupling site. Based on this, we design a nondestructive method to probe the system's excitations in a broad range of energies by measuring the state of the probe using standard atom optics methods. We illustrate our findings with numerical simulations for quantum lattice systems, including realistic dephasing noise on the quantum probe, and discuss how a controllable dephasing rate on the quantum probe may enable distinguishing regular and chaotic spectra.
  • PDF
    Imbalanced data widely exists in many high-impact applications. An example is in air traffic control, where we aim to identify the leading indicators for each type of accident cause from historical records. Among all three types of accident causes, historical records with 'personnel issues' are much more than the other two types ('aircraft issues' and 'environmental issues') combined. Thus, the resulting dataset is highly imbalanced, and can be naturally modeled as a network. Up until now, most existing work on imbalanced data analysis focused on the classification setting, and very little is devoted to learning the node representation from imbalanced networks. To address this problem, in this paper, we propose Vertex-Diminished Random Walk (VDRW) for imbalanced network analysis. The key idea is to encourage the random particle to walk within the same class by adjusting the transition probabilities each step. It resembles the existing Vertex Reinforced Random Walk in terms of the dynamic nature of the transition probabilities, as well as some convergence properties. However, it is more suitable for analyzing imbalanced networks as it leads to more separable node representations in the embedding space. Then, based on VDRW, we propose a semi-supervised network representation learning framework named ImVerde for imbalanced networks, in which context sampling uses VDRW and the label information to create node-context pairs, and balanced-batch sampling adopts a simple under-sampling method to balance these pairs in different classes. Experimental results demonstrate that ImVerde based on VDRW outperforms state-of-the-art algorithms for learning network representation from imbalanced data.
  • PDF
    Most existing algorithms for dictionary learning assume that all entries of the (high-dimensional) input data are fully observed. However, in several practical applications (such as hyper-spectral imaging or blood glucose monitoring), only an incomplete fraction of the data entries may be available. For incomplete settings, no provably correct and polynomial-time algorithm has been reported in the dictionary learning literature. In this paper, we provide provable approaches for learning - from incomplete samples - a family of dictionaries whose atoms have sufficiently "spread-out" mass. First, we propose a descent-style iterative algorithm that linearly converges to the true dictionary when provided a sufficiently coarse initial estimate. Second, we propose an initialization algorithm that utilizes a small number of extra fully observed samples to produce such a coarse initial estimate. Finally, we theoretically analyze their performance and provide asymptotic statistical and computational guarantees.
  • PDF
    This paper addresses the problem of detecting anomalous activity in traffic networks where the network is not directly observed. Given knowledge of what the node-to-node traffic in a network should be, any activity that differs significantly from this baseline would be considered anomalous. We propose a Bayesian hierarchical model for estimating the traffic rates and detecting anomalous changes in the network. The probabilistic nature of the model allows us to perform statistical goodness-of-fit tests to detect significant deviations from a baseline network. We show that due to the more defined structure of the hierarchical Bayesian model, such tests perform well even when the empirical models estimated by the EM algorithm are misspecified. We apply our model to both simulated and real datasets to demonstrate its superior performance over existing alternatives.
  • PDF
    The analysis of animal cross section images, such as cross sections of laboratory mice, is critical in assessing the effect of experimental drugs such as the biodistribution of candidate compounds in preclinical drug development stage. Tissue distribution of radiolabeled candidate therapeutic compounds can be quantified using techniques like Quantitative Whole-Body Autoradiography (QWBA).QWBA relies, among other aspects, on the accurate segmentation or identification of key organs of interest in the animal cross section image such as the brain, spine, heart, liver and others. We present a deep learning based organ segmentation solution to this problem, using which we can achieve automated organ segmentation with high precision (dice coefficient in the 0.83-0.95 range depending on organ) for the key organs of interest.
  • PDF
    A patch-based convolutional neural network (CNN) model presented in this paper for vocal melody extraction in polyphonic music is inspired from object detection in image processing. The input of the model is a novel time-frequency representation which enhances the pitch contours and suppresses the harmonic components of a signal. This succinct data representation and the patch-based CNN model enable an efficient training process with limited labeled data. Experiments on various datasets show excellent speed and competitive accuracy comparing to other deep learning approaches.
  • PDF
    Asymmetric segregation of key proteins at cell division -- be it a beneficial or deleterious protein -- is ubiquitous in unicellular organisms and often considered as an evolved trait to increase fitness in a stressed environment. Here, we provide a general framework to describe the evolutionary origin of this asymmetric segregation. We compute the population fitness as a function of the protein segregation asymmetry $a$, and show that the value of $a$ which optimizes the population growth manifests a phase transition between symmetric and asymmetric partitioning phases. Surprisingly, the nature of phase transition is different for the case of beneficial proteins as opposed to proteins which decrease the single-cell growth rate. Our study elucidates the optimization problem faced by evolution in the context of protein segregation, and motivates further investigation of asymmetric protein segregation in biological systems.
  • PDF
    We present MaskFusion, a real-time, object-aware, semantic and dynamic RGB-D SLAM system that goes beyond traditional systems that output a geometry-only map -- MaskFusion recognizes, segments and assigns semantic class labels to different objects in the scene, while tracking and reconstructing them even when they move independently from the camera. As an RGB-D camera scans a cluttered scene, image-based instance-level semantic segmentation creates semantic object masks that enable real-time object recognition and the creation of an object-level representation for the world map. Unlike previous recognition-based SLAM systems, MaskFusion does not require prior knowledge or known models of the objects it can recognize and can deal with multiple independent motions. Unlike recent semantics enabled SLAM systems that perform voxel-level semantic segmentation MaskFusion takes full advantage of using instance-level semantic segmentation to enable semantic labels to be fused into an object-aware map. We show augmented-reality applications, that demonstrate the unique features of the map output by MaskFusion: instance-aware, semantic and dynamic.
  • PDF
    [Abridged] We study the abundance and clustering properties of HI at redshifts $z\leqslant5$ using TNG100, a large state-of-the-art magneto-hydrodynamic simulation of a 75 Mpc/h box size. We show that most of the HI lies within dark matter halos and quantify the average HI mass hosted by halos of mass M at redshift z. We find that only halos with circular velocities larger than $\simeq$ 30 km/s contain HI. While the density profiles of HI exhibit a large halo-to-halo scatter, the mean profiles are universal across mass and redshift. The HI in low-mass halos is mostly located in the central galaxy, while in massive halos is concentrated in the satellites. We show that the HI and matter density probability distribution functions differ significantly. Our results point out that for small halos the HI bulk velocity goes in the same direction and has the same magnitude as the halo peculiar velocity, while in large halos differences show up. We find that halo HI velocity dispersion follows a power-law with halo mass. We find a complicated HI bias, with HI becoming non-linear already at $k=0.3$ h/Mpc at $z\gtrsim3$. Our simulation reproduces the DLAs bias value from observations. We find that the clustering of HI can be accurately reproduced by perturbative methods. We identify a new secondary bias, by showing that the clustering of halos depends not only on mass but also on HI content. We compute the amplitude of the HI shot-noise and find that it is small at all redshifts. We study the clustering of HI in redshift-space, and show that linear theory can explain the ratio between the monopoles in redshift- and real-space down to small scales at high redshift. We find that the amplitude of the Fingers-of-God effect is larger for HI than for matter. We point out that accurate 21 cm maps can be created from N-body or approximate simulations rather than full hydrodynamic simulations.
  • PDF
    Increasing numbers of physicists engage in research activities that address biological questions from physics perspectives or strive to develop physics insights from active biological processes. The on-going development and success of such activities morph our ways of thinking about what it is to 'do biophysics' and add to our understanding of the physics of life. Many scientists in this research and teaching landscape are homed in physics departments. A challenge for a hosting department is how to group, name and structure such biophysicists to best add value to their emerging research and teaching but also to the portfolio of the whole department. Here we discuss these issues and speculate on strategies.
  • PDF
    Measuring the distance between concepts is an important field of study of Natural Language Processing, as it can be used to improve tasks related to the interpretation of those same concepts. WordNet, which includes a wide variety of concepts associated with words (i.e., synsets), is often used as a source for computing those distances. In this paper, we explore a distance for WordNet synsets based on visual features, instead of lexical ones. For this purpose, we extract the graphic features generated within a deep convolutional neural networks trained with ImageNet and use those features to generate a representative of each synset. Based on those representatives, we define a distance measure of synsets, which complements the traditional lexical distances. Finally, we propose some experiments to evaluate its performance and compare it with the current state-of-the-art.
  • PDF
    Outdoor vision-based systems suffer from atmospheric turbulences, and rain is one of the worst factors for vision degradation. Current rain removal methods show limitations either for complex dynamic scenes, or under torrential rain with opaque occlusions. We propose a novel derain framework which applies superpixel (SP) segmentation to decompose the scene into depth consistent units. Alignment of scene contents are done at the SP level, which proves to be robust against rain occlusion interferences and fast camera motion. Two alignment output tensors, i.e., optimal temporal match tensor and sorted spatial-temporal match tensor, provide informative clues for the location of rain streaks and the occluded background contents. Different classical and novel methods such as Robust Principle Component Analysis and Convolutional Neural Networks are applied and compared for their respective advantages in efficiently exploiting the rich spatial-temporal features provided by the two tensors. Extensive evaluations show that advantage of up to 5dB is achieved on the scene restoration PSNR over state-of-the-art methods, and the advantage is especially obvious with highly complex and dynamic scenes. Visual evaluations show that the proposed framework is not only able to suppress heavy and opaque occluding rain streaks, but also large semi-transparent regional fluctuations and distortions.
  • PDF
    During a first St. Petersburg period Leonhard Euler, in his early twenties, became interested in the Basel problem: summing the series of inverse squares (posed by Pietro Mengoli in mid 17th century). In the words of Andre Weil (1989) "as with most questions that ever attracted his attention, he never abandoned it". Euler introduced on the way the alternating "phi-series", the better converging companion of the zeta function, the first example of a polylogarithm at a root of unity. He realized - empirically! - that odd zeta values appear to be new (transcendental?) numbers. It is amazing to see how, a quarter of a millennium later, the numbers Euler played with, "however repugnant" this game might have seemed to his contemporary lovers of the "higher kind of calculus", reappeared in the analytic calculation of the anomalous magnetic moment of the electron, the most precisely calculated and measured physical quantity. Mathematicians, inspired by ideas of Grothendieck, are reviving the dream of Galois of uncovering a group structure in the ring of periods (that includes the multiple zeta values) - applied to the study of Feynman amplitudes.
  • PDF
    Prosody is usually defined in terms of the three distinct but interacting domains of pitch, intensity and duration patterning, or, more generally, as phonological and phonetic properties of 'suprasegmentals', speech segments which are larger than consonants and vowels. Rather than taking this approach, the concept of multiple time domains for prosody processing is taken up, and methods of time domain analysis are discussed: annotation mining with timing dispersion measures, time tree induction, oscillator models in phonology and phonetics, and finally the use of the Amplitude Envelope Modulation Spectrum (AEMS). While frequency demodulation (in the form of pitch tracking) is a central issue in prosodic analysis, in the present context, it is amplitude envelope demodulation long time domain spectra which are focused. Using this method, multiple rhythms are described as multiple frequency zones in the AEMS, a new Frequency Zone Hypothesis of rhythm, and pointers to research fields beyond the time domains of foot, syllable and mora are outlined.
  • PDF
    Current end-to-end machine reading and question answering (Q\&A) models are primarily based on recurrent neural networks (RNNs) with attention. Despite their success, these models are often slow for both training and inference due to the sequential nature of RNNs. We propose a new Q\&A architecture called QANet, which does not require recurrent networks: Its encoder consists exclusively of convolution and self-attention, where convolution models local interactions and self-attention models global interactions. On the SQuAD dataset, our model is 3x to 13x faster in training and 4x to 9x faster in inference, while achieving equivalent accuracy to recurrent models. The speed-up gain allows us to train the model with much more data. We hence combine our model with data generated by backtranslation from a neural machine translation model. On the SQuAD dataset, our single model, trained with augmented data, achieves 84.6 F1 score on the test set, which is significantly better than the best published F1 score of 81.8.
  • PDF
    Many natural language processing tasks require dealing with Named Entities (NEs) in the texts themselves and sometimes also in external knowledge sources. While this is often easy for humans, recent neural methods that rely on learned word embeddings for NLP tasks have difficulty with it, especially with out of vocabulary or rare NEs. In this paper, we propose a new neural method for this problem, and present empirical evaluations on a structured Question-Answering task, three related Goal-Oriented dialog tasks and a reading-comprehension-based task. They show that our proposed method can be effective in dealing with both in-vocabulary and out of vocabulary (OOV) NEs. We create extended versions of dialog bAbI tasks 1,2 and 4 and Out-of-vocabulary (OOV) versions of the CBT test set which will be made publicly available online.
  • PDF
    Motivated by the study of kernels of bilinear pseudodifferential operators with symbols in a Hörmander class of critical order, we investigate boundedness properties of strongly singular Calderón--Zygmund operators in the bilinear setting. For such operators, whose kernels satisfy integral-type conditions, we establish boundedness properties in the setting of Lebesgue spaces as well as endpoint mappings involving the space of functions of bounded mean oscillations and the Hardy space. Assuming pointwise-type conditions on the kernels, we show that strongly singular bilinear Calderón--Zygmund operators satisfy pointwise estimates in terms of maximal operators, which imply their boundedness in weighted Lebesgue spaces.
  • PDF
    Rauzy-type dynamics are group actions on a collection of combinatorial objects. The first and best known example (the Rauzy dynamics) concerns an action on permutations, associated to interval exchange transformations (IET) for the Poincaré map on compact orientable translation surfaces. The equivalence classes on the objects induced by the group action have been classified by Kontsevich and Zorich in [KZ03] and correspond bijectively to the connected components of the strata of the moduli space of abelian differentials. In a paper [Boi14] Boissy proposed a Rauzy-type dynamics that acts on a subset of the permutations (the standard permutations) and conjectured that the Rauzy classes of this dynamics are exactly the Rauzy classes of the Rauzy dynamics restricted to standard permutations. In this paper, we apply the labelling method introduced in [D18] to classify this dynamics thus proving Boissy's conjecture. Finally, this paper conclude our serie of three papers on the study of the Rauzy dynamics by presenting two new results on the Rauzy classes: An quadratic algorithm for outputting a path between two connected permutations and a tight $\Theta(n)$ bound on the diameter of the Rauzy classes for the alternating distance.
  • PDF
    We investigate a wide class of two-dimensional hyperbolic systems with singularities, and prove the almost sure invariance principle (ASIP) for the random process generated by sequences of dynamically Hölder observables. The observables could be unbounded, and the process may be non-stationary and need not have linearly growing variances. Our results apply to Anosov diffeomorphisms, Sinai dispersing billiards and their perturbations. The random processes under consideration are related to the fluctuation of Lyapunov exponents, the shrinking target problem, etc.
  • PDF
    A \textitk-arc in the projective space ${\rm PG}(n,q)$ is a set of $k$ projective points such that no subcollection of $n+1$ points is contained in a hyperplane. Given a set $\mathcal{P}$ of $k$ projective points of ${\rm PG}(n,\mathbb{C})$, we introduce computational methods in ${\rm GAP}$ that are effective for verifying both the existence of $k$-arcs related to $\mathcal{P}$ in ${\rm PG}(n, q)$ for infinitely many prime powers $q$ generally and for determining all possible instances when there may not be a $k$-arc in ${\rm PG}(n,q)$. Using these methods, we prove, for infinitely many primes $p$, that there exists a transitive ${\rm PSL}(2,11)$-invariant $110$-arc in ${\rm PG}(4,p)$ for infinitely many primes $p$ and in ${\rm PG}(4,p^2)$ for infinitely many primes $p$, where ${\rm PSL}(2,11)$ is given in its representation as a subgroup of ${\rm PGL}(5,p)$ or ${\rm PGL}(5,p^2)$, respectively. Similarly, we show that there exists a transitive ${\rm PSL}(2,11)$-invariant $60$-arcs in ${\rm PG}(4,p)$ for infinitely many primes $p$.
  • PDF
    We introduce an efficient dynamical tree method that enables us, for the first time, to explicitly demonstrate thermo-remanent magnetization memory effect in a hierarchical energy landscape. Our simulation nicely reproduces the nontrivial waiting-time and waiting-temperature dependences in this non-equilibrium phenomenon. We further investigate the condensation effect, in which a small set of micro-states dominates the thermodynamic behavior, in the multi-layer trap model. Importantly, a structural phase transition of the tree is shown to coincide with the onset of condensation phenomenon. Our results underscore the importance of hierarchical structure and demonstrate the intimate relation between glassy behavior and structure of barrier trees.
  • PDF
    We consider systems of strict multivariate polynomial inequalities over the reals. All polynomial coefficients are parameters ranging over the reals, where for each coefficient we prescribe its sign. We are interested in the existence of positive real solutions of our system for all choices of coefficients subject to our sign conditions. We give a decision procedure for the existence of such solutions. In the positive case our procedure yields a parametric positive solution as a rational function in the coefficients. Our framework allows to reformulate heuristic subtropical approaches for non-parametric systems of polynomial inequalities that have been recently used in qualitative biological network analysis and, independently, in satisfiability modulo theory solving. We apply our results to characterize the incompleteness of those methods.
  • PDF
    We study sums of Fourier coefficients of Hecke-Maass cusp forms for the group $\mathrm{SL}(n,\mathbb Z)$ with general $n\geq 3$ over certain short intervals under the assumption of the generalised Lindelöf hypothesis. In particular, we evaluate the second moment of the sums in question asymptotically.
  • PDF
    In this paper, we discuss automorphism related parameters of a graph associated to a finite vector space. The fixing neighborhood of a pair $(u,v)$ of vertices of a graph $G$ is the set of all those vertices $w$ of $G$, such that the orbits of $u$ and $v$ under the action of stabilizer of $w$ are not equal. The fixed number of a graph is the minimum number $k$ such that every subset of vertices of $G$ of cardinality $k$ is a fixing set of $G$. We study some properties of automorphisms of a graph associated to finite vector space and find the fixing neighborhood of pair of vertices of the graph. We also find the fixed number of the graph. It is shown that, for every positive integer $N$, there exists a graph $G$ with $fxd(G)-fix(G)\geq N$, where $fxd(G)$ is the fixed number and $fix(G)$ is the fixing number of $G$.
  • PDF
    Penitentes are spikes formed on the surface of the snow, which are present typically at high altitude in the Andes and Himalayas. They are a consequence of a thermodynamic instability, as a result of the surface sublimation at a given point due to the incidence of light scattered by surrounding features. Here, based on existing literature, we model the time evolution of penitente formation as a purely radiation-driven phenomenon. The physical system is governed by a 1D diffusion equation with a nonlocal source term, which represents the light coming in from all the line of sight accessible from that point of the curve. For small perturbations on the initial profile, the surface undergoes an instability which triggers the formation of spiky structures. For solar radiation coming in the surface at a given angle, our numerical simulations account for a feature observed in the real system: penitentes get tilted in the direction of the sunlight.
  • PDF
    The Araki-Lieb inequality is commonly used to calculate the entropy of subsystems when they are initially in pure states as this forces the entropy of the two subsystems to be equal after the complete system evolves. Then, it is easy to calculate the entropy of a large subsystem by finding the entropy of the small one. To the best of our knowledge, there does not exist a way of calculating the entropy when one of the subsystems is initially in a mixed state. We show here that it is possible to use the Araki-Lieb inequality in this case and find the von Neumann entropy for the large (infinite) system. We show this in the two-level atom-field interaction.
  • PDF
    We study the differential equation $ - (p(x) y')' + q(x) y' = \lambda y,$ where $p(x)$ is a polynomial of degree at most 2 and $q(x)$ is a polynomial of degree at most 1. This includes the classical Jacobi polynomials, Hermite polynomials, Legendre polynomials, Chebychev polynomials and Laguerre polynomials. We provide a general electrostatic interpretation of roots of such polynomials: the set of real numbers $\left\{x_1, \dots, x_n\right\}$ satisfies $$ p(x_i) \sum_k = 1 \atop k ≠i^n\frac2x_k - x_i = q(x_i) - p'(x_i) \qquad \mboxfor all~ 1≤i ≤n$$ if and only if they are roots of a polynomial solving the differential equation. We also derive a system of ODEs depending on $p(x),q(x)$ whose solutions converge to the roots of the orthogonal polynomial at an exponential rate.
  • PDF
    In his recent work arXiv:1708.06713, X. Yang proved a conjecture raised by Yau in 1982, which states that any compact Kähler manifold with positive holomorphic sectional curvature must be projective. This gives a metric criterion of the projectivity. In this note, we prove a generalization to this statement by showing that any compact Kähler manifold with positive 2nd scalar curvature (which is the average of holomorphic sectional curvature over $2$-dimensional subspaces of the tangent space) must be projective. In view of generic 2-tori being non-Abelian, this condition is sharp in some sense. Vanishing theorems are also proved for the Hodge numbers when the condition is replaced by the positivity of the weaker interpolating $k$-scalar curvature.
  • PDF
    We observed twelve Plutinos over two separated years with the 4.3m Lowell's Discovery Channel Telescope. Here, we present the first lightcurve data for those objects. Three of them (2014JL$_{80}$, 2014JO$_{80}$, 2014JQ$_{80}$) display a large lightcurve amplitude explainable by a single elongated object, but are most likely caused by a contact binary system due to their lightcurves morphology. These potential contact binaries have rotational periods from 6.3h to 34.9h and peak-to-peak lightcurve variability between 0.6 and 0.8mag. We present partial lightcurves allowing us to constrain the lightcurve amplitude and the rotational period of another nine Plutinos. By merging our data with the literature, we estimate that up to $\sim$40$\%$ of the Plutinos could be contact binaries. Interestingly, we found that all the suspected contact binaries in the 3:2 resonance are small with absolute magnitude H$>$6mag. Based on our sample and the literature, up to $\sim$50$\%$ of the small Plutinos are potential contact binaries.

Recent comments

Max Lu Apr 25 2018 22:08 UTC

"This is a very inspiring paper! The new framework (ZR = All Reality) it provided allows us to understand all kinds of different reality technologies (VR, AR, MR, XR etc) that are currently loosely connected to each other and has been confusing to many people. Instead of treating our perceived sens

...(continued)
Stefano Pirandola Apr 23 2018 12:23 UTC

The most important reading here is Sam Braunstein's foundational paper: https://authors.library.caltech.edu/3827/1/BRAprl98.pdf published in January 98, already containing the key results for the strong convergence of the CV protocol. This is a must-read for those interested in CV quantum informatio

...(continued)
Mark M. Wilde Apr 23 2018 12:09 UTC

One should also consult my paper "Strong and uniform convergence in the teleportation simulation of bosonic Gaussian channels" https://arxiv.org/abs/1712.00145v4 posted in January 2018, in this context.

Stefano Pirandola Apr 23 2018 11:46 UTC

Some quick clarifications on the Braunstein-Kimble (BK) protocol for CV teleportation
and the associated teleportation simulation of bosonic channels.
(Disclaimer: the following is rather technical and CVs might not be so popular on this blog...so I guess this post will get a lot of dislikes :)

1)

...(continued)
NJBouman Apr 22 2018 18:26 UTC

[Fredrik Johansson][1] has pointed out to me (the author) the following about the multiplication benchmark w.r.t. GMP. This will be taken into account in the upcoming revision.

Fredrik Johansson wrote:
> You shouldn't be comparing your code to `mpn_mul`, because this function is not actually th

...(continued)
Joel Wallman Apr 18 2018 13:34 UTC

A very nice approach! Could you clarify the conclusion a little bit though? The aspirational goal for a quantum benchmark is to test how well we approximate a *specific* representation of a group (up to similarity transforms), whereas what your approach demonstrates is that without additional knowle

...(continued)
serfati philippe Mar 29 2018 14:07 UTC

see my 2 papers on direction of vorticity (nov1996 + feb1999) = https://www.researchgate.net/profile/Philippe_Serfati (published author, see also mendeley, academia.edu, orcid etc)

serfati philippe Mar 29 2018 13:34 UTC

see my 4 papers, 1998-1999, on contact and superposed vortex patches, cusps (and eg splashs), corners, generalized ones on lR^n and (ir/)regular ones =. http://www.researchgate.net/profile/Philippe_Serfati/ (published author).

Luis Cruz Mar 16 2018 15:34 UTC

Related Work:

- [Performance-Based Guidelines for Energy Efficient Mobile Applications](http://ieeexplore.ieee.org/document/7972717/)
- [Leafactor: Improving Energy Efficiency of Android Apps via Automatic Refactoring](http://ieeexplore.ieee.org/document/7972807/)

Dan Elton Mar 16 2018 04:36 UTC

Comments are appreciated. Message me here or on twitter @moreisdifferent

Code is open source and available at :
[https://github.com/delton137/PIMD-F90][1]

[1]: https://github.com/delton137/PIMD-F90