arXiv subjects

Hao Zhang

Publications and source records attributed to Hao Zhang.

At least 1,117 records · Page 62Linked to original sources

arXiv2020

Inverse Compton Signatures of Gamma-Ray Burst Afterglows

The afterglow emission from gamma-ray bursts (GRBs) is believed to originate from a relativistic blast wave driven into the circumburst medium. Although the afterglow emission from radio up to X-ray frequencies is thought to originate from synchrotron radiation emitted by relativistic, non-thermal electrons accelerated by the blast wave, the origin of the emission at high energies (HE; $\gtrsim$~GeV) remains uncertain. The recent detection of sub-TeV emission from GRB~190114C by MAGIC raises further debate on what powers the very high-energy (VHE; $\gtrsim 300$GeV) emission. Here, we explore the inverse Compton scenario as a candidate for the HE and VHE emissions, considering two sources of seed photons for scattering: synchrotron photons from the blast wave (synchrotron self-Compton or SSC) and isotropic photon fields external to the blast wave (external Compton). For each case, we compute the multi-wavelength afterglow spectra and light curves. We find that SSC will dominate particle cooling and the GeV emission, unless a dense ambient infrared photon field, typical of star-forming regions, is present. Additionally, considering the extragalactic background light attenuation, we discuss the detectability of VHE afterglows by existing and future gamma-ray instruments for a wide range of model parameters. Studying GRB~190114C, we find that its afterglow emission in the \fermi-LAT band is synchrotron-dominated.The late-time \fermi-LAT measurement (i.e., $t\sim 10^4$~s), and the MAGIC observation also set an upper limit on the energy density of a putative external infrared photon field (i.e. $\lesssim 3\times 10^{-9}\,{\rm erg\,cm^{-3}}$), making the inverse Compton dominant in the sub-TeV energies.

astro-ph.HE↗

arXiv2020

Deep Autoencoding Topic Model with Scalable Hybrid Bayesian Inference

To build a flexible and interpretable model for document analysis, we develop deep autoencoding topic model (DATM) that uses a hierarchy of gamma distributions to construct its multi-stochastic-layer generative network. In order to provide scalable posterior inference for the parameters of the generative network, we develop topic-layer-adaptive stochastic gradient Riemannian MCMC that jointly learns simplex-constrained global parameters across all layers and topics, with topic and layer specific learning rates. Given a posterior sample of the global parameters, in order to efficiently infer the local latent representations of a document under DATM across all stochastic layers, we propose a Weibull upward-downward variational encoder that deterministically propagates information upward via a deep neural network, followed by a Weibull distribution based stochastic downward generative model. To jointly model documents and their associated labels, we further propose supervised DATM that enhances the discriminative power of its latent representations. The efficacy and scalability of our models are demonstrated on both unsupervised and supervised learning tasks on big corpora.

cs.LG↗

arXiv2020

Span-based Localizing Network for Natural Language Video Localization

Given an untrimmed video and a text query, natural language video localization (NLVL) is to locate a matching span from the video that semantically corresponds to the query. Existing solutions formulate NLVL either as a ranking task and apply multimodal matching architecture, or as a regression task to directly regress the target video span. In this work, we address NLVL task with a span-based QA approach by treating the input video as text passage. We propose a video span localizing network (VSLNet), on top of the standard span-based QA framework, to address NLVL. The proposed VSLNet tackles the differences between NLVL and span-based QA through a simple yet effective query-guided highlighting (QGH) strategy. The QGH guides VSLNet to search for matching video span within a highlighted region. Through extensive experiments on three benchmark datasets, we show that the proposed VSLNet outperforms the state-of-the-art methods; and adopting span-based QA framework is a promising direction to solve NLVL.

cs.CL↗

arXiv2020

An Unsupervised Semantic Sentence Ranking Scheme for Text Documents

This paper presents Semantic SentenceRank (SSR), an unsupervised scheme for automatically ranking sentences in a single document according to their relative importance. In particular, SSR extracts essential words and phrases from a text document, and uses semantic measures to construct, respectively, a semantic phrase graph over phrases and words, and a semantic sentence graph over sentences. It applies two variants of article-structure-biased PageRank to score phrases and words on the first graph and sentences on the second graph. It then combines these scores to generate the final score for each sentence. Finally, SSR solves a multi-objective optimization problem for ranking sentences based on their final scores and topic diversity through semantic subtopic clustering. An implementation of SSR that runs in quadratic time is presented, and it outperforms, on the SummBank benchmarks, each individual judge's ranking and compares favorably with the combined ranking of all judges.

cs.IR↗

arXiv2020

Data-Driven Construction of Data Center Graph of Things for Anomaly Detection

Data center (DC) contains both IT devices and facility equipment, and the operation of a DC requires a high-quality monitoring (anomaly detection) system. There are lots of sensors in computer rooms for the DC monitoring system, and they are inherently related. This work proposes a data-driven pipeline (ts2graph) to build a DC graph of things (sensor graph) from the time series measurements of sensors. The sensor graph is an undirected weighted property graph, where sensors are the nodes, sensor features are the node properties, and sensor connections are the edges. The sensor node property is defined by features that characterize the sensor events (behaviors), instead of the original time series. The sensor connection (edge weight) is defined by the probability of concurrent events between two sensors. A graph of things prototype is constructed from the sensor time series of a real data center, and it successfully reveals meaningful relationships between the sensors. To demonstrate the use of the DC sensor graph for anomaly detection, we compare the performance of graph neural network (GNN) and existing standard methods on synthetic anomaly data. GNN outperforms existing algorithms by a factor of 2 to 3 (in terms of precision and F1 score), because it takes into account the topology relationship between DC sensors. We expect that the DC sensor graph can serve as the infrastructure for the DC monitoring system since it represents the sensor relationships.

cs.LG↗

arXiv2020

Graph2Plan: Learning Floorplan Generation from Layout Graphs

We introduce a learning framework for automated floorplan generation which combines generative modeling using deep neural networks and user-in-the-loop designs to enable human users to provide sparse design constraints. Such constraints are represented by a layout graph. The core component of our learning framework is a deep neural network, Graph2Plan, which converts a layout graph, along with a building boundary, into a floorplan that fulfills both the layout and boundary constraints. Given an input building boundary, we allow a user to specify room counts and other layout constraints, which are used to retrieve a set of floorplans, with their associated layout graphs, from a database. For each retrieved layout graph, along with the input boundary, Graph2Plan first generates a corresponding raster floorplan image, and then a refined set of boxes representing the rooms. Graph2Plan is trained on RPLAN, a large-scale dataset consisting of 80K annotated floorplans. The network is mainly based on convolutional processing over both the layout graph, via a graph neural network (GNN), and the input building boundary, as well as the raster floorplan images, via conventional image convolution.

cs.CV↗

arXiv2020

WHAI: Weibull Hybrid Autoencoding Inference for Deep Topic Modeling

To train an inference network jointly with a deep generative topic model, making it both scalable to big corpora and fast in out-of-sample prediction, we develop Weibull hybrid autoencoding inference (WHAI) for deep latent Dirichlet allocation, which infers posterior samples via a hybrid of stochastic-gradient MCMC and autoencoding variational Bayes. The generative network of WHAI has a hierarchy of gamma distributions, while the inference network of WHAI is a Weibull upward-downward variational autoencoder, which integrates a deterministic-upward deep neural network, and a stochastic-downward deep generative model based on a hierarchy of Weibull distributions. The Weibull distribution can be used to well approximate a gamma distribution with an analytic Kullback-Leibler divergence, and has a simple reparameterization via the uniform noise, which help efficiently compute the gradients of the evidence lower bound with respect to the parameters of the inference network. The effectiveness and efficiency of WHAI are illustrated with experiments on big corpora.

stat.ML↗

arXiv2020

PQ-NET: A Generative Part Seq2Seq Network for 3D Shapes

We introduce PQ-NET, a deep neural network which represents and generates 3D shapes via sequential part assembly. The input to our network is a 3D shape segmented into parts, where each part is first encoded into a feature representation using a part autoencoder. The core component of PQ-NET is a sequence-to-sequence or Seq2Seq autoencoder which encodes a sequence of part features into a latent vector of fixed size, and the decoder reconstructs the 3D shape, one part at a time, resulting in a sequential assembly. The latent space formed by the Seq2Seq encoder encodes both part structure and fine part geometry. The decoder can be adapted to perform several generative tasks including shape autoencoding, interpolation, novel shape generation, and single-view 3D reconstruction, where the generated shapes are all composed of meaningful parts.

cs.CV↗

arXiv2020

Sub-picosecond photo-induced displacive phase transition in two-dimensional MoTe$_2$

Photo-induced phase transitions (PIPTs) provide an ultrafast, energy-efficient way for precisely manipulating the topological properties of transition-metal ditellurides, and can be used to stabilize a topological phase in an otherwise semiconducting material. Using first-principles calculations, we demonstrate that the PIPT in monolayer MoTe$_2$ from the semiconducting 2H phase to the topological 1T$'$ phase can be triggered purely by electronic excitations that soften multiple lattice vibrational modes. These softenings, driven by a Peierls-like mechanism within the conduction bands, lead to structural symmetry breaking within sub-picosecond timescales, which is shorter than the timescale of a thermally driven phase transition. The transition is predicted to be triggered by photons with energies over $1.96$\,eV, with an associated excited carrier density of $3.4\times10^{14}$\,cm$^{-2}$, which enables a controllable phase transformation by varying the laser wavelength. Our results provide insight into the underlying physics of the phase transition in 2D transition-metal ditellurides, and show an ultrafast phase transition mechanism for manipulation of the topological properties of 2D systems.

cond-mat.mtrl-sci↗

arXiv2020

Developing a Recommendation Benchmark for MLPerf Training and Inference

Deep learning-based recommendation models are used pervasively and broadly, for example, to recommend movies, products, or other information most relevant to users, in order to enhance the user experience. Among various application domains which have received significant industry and academia research attention, such as image classification, object detection, language and speech translation, the performance of deep learning-based recommendation models is less well explored, even though recommendation tasks unarguably represent significant AI inference cycles at large-scale datacenter fleets. To advance the state of understanding and enable machine learning system development and optimization for the commerce domain, we aim to define an industry-relevant recommendation benchmark for the MLPerf Training andInference Suites. The paper synthesizes the desirable modeling strategies for personalized recommendation systems. We lay out desirable characteristics of recommendation model architectures and data sets. We then summarize the discussions and advice from the MLPerf Recommendation Advisory Board.

cs.LG↗

arXiv2020

Hybridized quadrupolar excitations in the spin-anisotropic frustrated magnet FeI2

Magnetic order is usually associated with well-defined magnon excitations. Exotic magnetic fluctuations with fractional, topological or multipolar character, have been proposed for radically different forms of magnetic matter such as spin-liquids. As a result, considerable efforts have searched for, and uncovered, low-spin materials with suppressed dipolar order at low temperatures. Here, we report neutron-scattering experiments and quantitative theoretical modeling of an exceptional spin-1 system -- the uniaxial triangular magnet FeI2 -- where a bright and dispersive band of mixed dipolar-quadrupolar fluctuations emerges just above a dipolar ordered ground-state. This excitation arises from anisotropic exchange interactions that hybridize overlapping modes carrying fundamentally different quantum numbers. Remarkably, a generalization of spin-wave theory to local SU(3) degrees of freedom accounts for all details of the low-energy dynamical response of FeI2 without going beyond quadratic order. Our work highlights that quantum excitations without classical counterparts can be realized even in presence of fully developed magnetic order.

cond-mat.str-el↗

arXiv2020

Representing Multi-Robot Structure through Multimodal Graph Embedding for the Selection of Robot Teams

Multi-robot systems of increasing size and complexity are used to solve large-scale problems, such as area exploration and search and rescue. A key decision in human-robot teaming is dividing a multi-robot system into teams to address separate issues or to accomplish a task over a large area. In order to address the problem of selecting teams in a multi-robot system, we propose a new multimodal graph embedding method to construct a unified representation that fuses multiple information modalities to describe and divide a multi-robot system. The relationship modalities are encoded as directed graphs that can encode asymmetrical relationships, which are embedded into a unified representation for each robot. Then, the constructed multimodal representation is used to determine teams based upon unsupervised learning. We perform experiments to evaluate our approach on expert-defined team formations, large-scale simulated multi-robot systems, and a system of physical robots. Experimental results show that our method successfully decides correct teams based on the multifaceted internal structures describing multi-robot systems, and outperforms baseline methods based upon only one mode of information, as well as other graph embedding-based division methods.

cs.RO↗

arXiv2020

DR-KFS: A Differentiable Visual Similarity Metric for 3D Shape Reconstruction

We introduce a differential visual similarity metric to train deep neural networks for 3D reconstruction, aimed at improving reconstruction quality. The metric compares two 3D shapes by measuring distances between multi-view images differentiably rendered from the shapes. Importantly, the image-space distance is also differentiable and measures visual similarity, rather than pixel-wise distortion. Specifically, the similarity is defined by mean-squared errors over HardNet features computed from probabilistic keypoint maps of the compared images. Our differential visual shape similarity metric can be easily plugged into various 3D reconstruction networks, replacing their distortion-based losses, such as Chamfer or Earth Mover distances, so as to optimize the network weights to produce reconstructions with better structural fidelity and visual quality. We demonstrate this both objectively, using well-known shape metrics for retrieval and classification tasks that are independent from our new metric, and subjectively through a perceptual study.

cs.GR↗

arXiv2020

Simultaneous Learning from Human Pose and Object Cues for Real-Time Activity Recognition

Real-time human activity recognition plays an essential role in real-world human-centered robotics applications, such as assisted living and human-robot collaboration. Although previous methods based on skeletal data to encode human poses showed promising results on real-time activity recognition, they lacked the capability to consider the context provided by objects within the scene and in use by the humans, which can provide a further discriminant between human activity categories. In this paper, we propose a novel approach to real-time human activity recognition, through simultaneously learning from observations of both human poses and objects involved in the human activity. We formulate human activity recognition as a joint optimization problem under a unified mathematical framework, which uses a regression-like loss function to integrate human pose and object cues and defines structured sparsity-inducing norms to identify discriminative body joints and object attributes. To evaluate our method, we perform extensive experiments on two benchmark datasets and a physical robot in a home assistance setting. Experimental results have shown that our method outperforms previous methods and obtains real-time performance for human activity recognition with a processing speed of 10^4 Hz.

cs.CV↗

arXiv2020

Dual-discriminator GAN: A GAN way of profile face recognition

A wealth of angle problems occur when facial recognition is performed: At present, the feature extraction network presents eigenvectors with large differences between the frontal face and profile face recognition of the same person in many cases. For this reason, the state-of-the-art facial recognition network will use multiple samples for the same target to ensure that eigenvector differences caused by angles are ignored during training. However, there is another solution available, which is to generate frontal face images with profile face images before recognition. In this paper, we proposed a method of generating frontal faces with image-to-image profile faces based on Generative Adversarial Network (GAN).

cs.CV↗

arXiv2020

Quantum Sensing by Using STIRAP with Dressed States Driving

Exploring quantum technology to precisely measure physical quantities is a meaningful task for practical scientific researches. Here, we propose a novel quantum sensing model based on dressed states driving (DSD) in stimulated Raman adiabatic passage. The model is universal for sensing different physical quantities, such as magnetic field, mass, rotation and etc. For different sensors, the used systems can range from macroscopic scale, e.g. optomechanical systems, to microscopic nanoscale, e.g. nitrogen-vacancy color centres in diamond. By investigating the dynamics of color detuning of DSD passage, the results show the sensitivity of sensors can be enhanced by tuning system with more adiabatic and accelerated processes in non-degenerate and degenerate color detuning regime, respectively. To show application examples, we apply our approach to build optomechanical mass sensor and solid spin magnetometer with practical parameters.

quant-ph↗

arXiv2020

Exceptional point enhanced optical gyroscope in mechanical PT-symmetric system

As an important device for detecting rotation, high sensitivity gyroscope is required for practical applications. In recent years, exceptional point (EP) shows its potential in enhancing the sensitivity of sensing in optical cavity. Here we propose an EP enhanced optical gyroscope based on mechanical PT-symmetric system in microcavity. By pumping the two optical modes with different colors, i.e. blue and red detuning, an effective mechanical PT-symmetric system can be obtained and the system can be prepared in EP with appropriate parameters. Compared with the situation of diabolic point, EP can enhance the sensitivity of gyroscope with more than one order of magnitude in the weak perturbation regime. The results show the gyroscope can be enhanced effectively by monitoring mechanical modes rather than optical modes. Our work provides a promising approach to design gyroscope with higher sensitivity in optical microcavity and has potential values in some fields including fundamental physic and precision measurement.

physics.optics↗

arXiv2020

Study of the Modified Gaussian Model on olivine diagnostic spectral features and its applications in space weathering experiments

The absorption features of olivine in visible and near-infrared (VNIR) reflectance spectra are the key spectral parameters in its mineralogical studies. Generally, these spectral parameters can be obtained by exploiting the Modified Gaussian Model (MGM) with a proper continuum removal. However, different continua may change the deconvolution results of these parameters. This paper investigates the diagnostic spectral features of olivine with diverse chemical compositions. Four different continuum removal methods with MGM for getting the deconvolution results are presented and the regression equations for predicting the Mg-number (Fo#) are introduced. The results show that different continua superimposed on the mineral absorption features will make the absorption center shift, as well as the obvious alterations in shape, width, and strength of the absorption band. Additionally, it is also found that the logarithm of a second-order polynomial continuum can match the overall shape of the spectrum in logarithmic space, and the improved regression equations applied to estimate the chemical composition of olivine-dominated spectra also have a better performance. As an application example, the improved approach is applied to pulse laser irradiated olivine grains to simulate and study the space weathering effects on olivine diagnostic spectral features. The experiments confirm that space weathering can make the absorption band center shift toward longer wavelength. Therefore, the Fo# estimated from remote sensing spectra may be less than its actual chemical composition. These results may provide valuable information for revealing the difference between the spectra of olivine grains and olivine-dominated asteroids.

astro-ph.IM↗