arXiv subjects

Xuan Li

Publications and source records attributed to Xuan Li.

At least 145 records · Page 8Linked to original sources

arXiv2020

An End-to-End Dialogue State Tracking System with Machine Reading Comprehension and Wide & Deep Classification

This paper describes our approach in DSTC 8 Track 4: Schema-Guided Dialogue State Tracking. The goal of this task is to predict the intents and slots in each user turn to complete the dialogue state tracking (DST) based on the information provided by the task's schema. Different from traditional stage-wise DST, we propose an end-to-end DST system to avoid error accumulation between the dialogue turns. The DST system consists of a machine reading comprehension (MRC) model for non-categorical slots and a Wide & Deep model for categorical slots. As far as we know, this is the first time that MRC and Wide & Deep model are applied to DST problem in a fully end-to-end way. Experimental results show that our framework achieves an excellent performance on the test dataset including 50% zero-shot services with a joint goal accuracy of 0.8652 and a slot tagging F1-Score of 0.9835.

cs.CL↗

arXiv2019

Static and Dynamic Fusion for Multi-modal Cross-ethnicity Face Anti-spoofing

Regardless of the usage of deep learning and handcrafted methods, the dynamic information from videos and the effect of cross-ethnicity are rarely considered in face anti-spoofing. In this work, we propose a static-dynamic fusion mechanism for multi-modal face anti-spoofing. Inspired by motion divergences between real and fake faces, we incorporate the dynamic image calculated by rank pooling with static information into a conventional neural network (CNN) for each modality (i.e., RGB, Depth and infrared (IR)). Then, we develop a partially shared fusion method to learn complementary information from multiple modalities. Furthermore, in order to study the generalization capability of the proposal in terms of cross-ethnicity attacks and unknown spoofs, we introduce the largest public cross-ethnicity Face Anti-spoofing (CASIA-CeFA) dataset, covering 3 ethnicities, 3 modalities, 1607 subjects, and 2D plus 3D attack types. Experiments demonstrate that the proposed method achieves state-of-the-art results on CASIA-CeFA, CASIA-SURF, OULU-NPU and SiW.

cs.CV↗

arXiv2019

Boosting Image Recognition with Non-differentiable Constraints

In this paper, we study the problem of image recognition with non-differentiable constraints. A lot of real-life recognition applications require a rich output structure with deterministic constraints that are discrete or modeled by a non-differentiable function. A prime example is recognizing digit sequences, which are restricted by such rules (e.g., \textit{container code detection}, \textit{social insurance number recognition}, etc.). We investigate the usefulness of adding non-differentiable constraints in learning for the task of digit sequence recognition. Toward this goal, we synthesize six different datasets from MNIST and Cropped SVHN, with three discrete rules inspired by real-life protocols. To deal with the non-differentiability of these rules, we propose a reinforcement learning approach based on the policy gradient method. We find that incorporating this rule-based reinforcement can effectively increase the accuracy for all datasets and provide a good inductive bias which improves the model even with limited data. On one of the datasets, MNIST\_Rule2, models trained with rule-based reinforcement increase the accuracy by 4.7\% for 2000 samples and 23.6\% for 500 samples. We further test our model against synthesized adversarial examples, e.g., blocking out digits, and observe that adding our rule-based reinforcement increases the model robustness with a relatively smaller performance drop.

cs.CV↗

arXiv2019

An End-to-end Video Text Detector with Online Tracking

Video text detection is considered as one of the most difficult tasks in document analysis due to the following two challenges: 1) the difficulties caused by video scenes, i.e., motion blur, illumination changes, and occlusion; 2) the properties of text including variants of fonts, languages, orientations, and shapes. Most existing methods attempt to enhance the performance of video text detection by cooperating with video text tracking, but treat these two tasks separately. In this work, we propose an end-to-end video text detection model with online tracking to address these two challenges. Specifically, in the detection branch, we adopt ConvLSTM to capture spatial structure information and motion memory. In the tracking branch, we convert the tracking problem to text instance association, and an appearance-geometry descriptor with memory mechanism is proposed to generate robust representation of text instances. By integrating these two branches into one trainable framework, they can promote each other and the computational cost is significantly reduced. Experiments on existing video text benchmarks including ICDAR2013 Video, Minetto and YVT demonstrate that the proposed method significantly outperforms state-of-the-art methods. Our method improves F-score by about 2 on all datasets and it can run realtime with 24.36 fps on TITAN Xp.

cs.CV↗

arXiv2019

Write-Optimized and Consistent RDMA-based NVM Systems

In order to deliver high performance in cloud computing, we generally exploit and leverage RDMA (Remote Direct Memory Access) in networking and NVM (Non-Volatile Memory) in end systems. Due to no involvement of CPU, one-sided RDMA becomes efficient to access the remote memory, and NVM technologies have the strengths of non-volatility, byte-addressability and DRAM-like latency. In order to achieve end-to-end high performance, many efforts aim to synergize one-sided RDMA and NVM. Due to the need to guarantee Remote Data Atomicity (RDA), we have to consume extra network round-trips, remote CPU participation and double NVM writes. In order to address these problems, we propose a zero-copy log-structured memory design for Efficient Remote Data Atomicity, called Erda. In Erda, clients directly transfer data to the destination address at servers via one-sided RDMA writes without redundant copy and remote CPU consumption. To detect the incompleteness of fetched data, we verify a checksum without client-server coordination. We further ensure metadata consistency by leveraging an 8-byte atomic update in the hash table, which also contains the address information for the stale data. When a failure occurs, the server properly restores to a consistent version. Experimental results show that compared with Redo Logging (a CPU involvement scheme) and Read After Write (a network dominant scheme), Erda reduces NVM writes approximately by 50%, as well as significantly improves throughput and decreases latency.

cs.DC↗

arXiv2019

Optimal $k$-Coverage Charging Problem

Wireless rechargeable sensor networks, consisting of sensor nodes with rechargeable batteries and mobile chargers to replenish their batteries, have gradually become a promising solution to the bottleneck of energy limitation that hinders the wide deployment of wireless sensor networks (WSN). In this paper, we focus on the mobile charger scheduling and path optimization scenario in which the $k$-coverage ability of a network system needs to be maintained. We formulate the optimal $k$-coverage charging problem of finding a feasible path for a mobile charger to charge a set of sensor nodes within their estimated charging deadlines under the constraint of maintaining the $k$-coverage ability of the network system, with an objective of minimizing the energy consumption on traveling per tour. We show the hardness of the problem that even finding a feasible path for the trivial case of the problem is an NP-complete one. We model the problem and apply dynamic programming to design an algorithm that finds an exact solution to the optimal $k$-coverage charging problem. However, the computational complexity is still prohibitive for large size networks. We then introduce Deep Q-learning, a reinforcement learning algorithm to tackle the problem.

cs.NI↗

arXiv2019

Non-volatile silicon photonic memory with more than 4-bit per cell capability

We present the first demonstration of an integrated photonic phase-change memory using GeSbTe-225 on silicon-on-insulator and demonstrate reliable multilevel operation with a single programming pulse. We also compare our results on silicon with previous demonstrations on silicon nitride. Crucially, achieving this on silicon enables tighter integration of traditional electronics with photonic memories in future, making phase-change photonic memory a viable and integrable technology.

physics.optics↗

arXiv2018

Plasmonic nanogap enhanced phase change devices with dual electrical-optical functionality

Modern-day computers use electrical signaling for processing and storing data which is bandwidth limited and power-hungry. These limitations are bypassed in the field of communications, where optical signaling is the norm. To exploit optical signaling in computing, however, new on-chip devices that work seamlessly in both electrical and optical domains are needed. Phase change devices can in principle provide such functionality, but doing so in a single device has proved elusive due to conflicting requirements of size-limited electrical switching and diffraction-limited photonic devices. Here, we combine plasmonics, photonics and electronics to deliver a novel integrated phase-change memory and computing cell that can be electrically or optically switched between binary or multilevel states, and read-out in either mode, thus merging computing and communications technologies.

physics.optics↗

arXiv2018

A Variable Reduction Method for Large-Scale Security Constrained Unit Commitment

Efficient methods for large-scale security constrained unit commitment (SCUC) problems have long been an important research topic and a challenge especially in market clearing computation. For large-scale SCUC, the Lagrangian relaxation methods (LR) and the mixed integer programming methods (MIP) are most widely adopted. However, LR usually suffers from slow convergence; and the computational burden of MIP is heavy when the binary variable number is large. In this paper, a variable reduction method is proposed: First, the time-coupled constraints in the original SCUC problem are relaxed and many single-period SCUC problems (s-UC) are obtained. Second, LR is used to solve the s-UCs. Different from traditional LR with iterative subgradient method, it is found that the optimal multipliers and the approximate UC solutions of s-UCs can be obtained by solving linear programs. Third, a criterion for choosing and fixing the UC variables in the SCUC problem is established, hence the number of binary variables is reduced. Last, the SCUC with reduced binary variables is solved by MIP solver to obtain the final UC solution. The proposed method is tested on the IEEE 118-bus system and a 6484-bus system. The results show the method is very efficient and effective.

math.OC↗

arXiv2018

Multi-Stage Robust Transmission Constrained Unit Commitment: A Decomposition Framework with Implicit Decision Rules

With the integration of large-scale renewable energy sources to power systems, many optimization methods have been applied to solve the stochastic/uncertain transmission-constrained unit commitment (TCUC) problem. Among all methods, two-stage and multi-stage robust optimization-based methods are the most widely adopted ones. In the two-stage methods, nonanticipativity of economic dispatch (ED) decisions are not considered. While in multi-stage methods, explicit decision rules (for example, affine decision rules) are usually adopted to guarantee nonanticipativity of ED decisions. With explicit decision rules, the computational burden can be heavy and the optimality of the solution is affected. In this paper, a multi-stage robust TCUC formulation with implicit decision rules is proposed, as well as a decomposition framework to solve it. The solutions are proved to be multi-stage robust and nonanticipativity of ED decisions is guaranteed. Meanwhile, a computationally efficient time-decoupled solution method for the feasibility check subproblems is also proposed such that the method is suitable for large-scale TCUC problems with uncertain loads/renewable injections. Numerical tests are conducted on the IEEE 118-bus system and Polish 2383-bus system. Performances of several state-of-the-art methods are compared.

math.OC↗

arXiv2018

Robust Transmission Constrained Unit Commitment:A Column Merging Method

With rapid integration of power sources with uncertainty, robustness must be carefully considered in the transmission constrained unit commitment (TCUC) problem. The overall computational complexity of the robust TCUC methods is closely related to the vertex number of the uncertainty set. The vertex number is further associated with 1) the period number in the scheduling horizon as well as 2) the number of nodes with uncertain injections. In this paper, a column merging method (CMM) is proposed to reduce the computation burden by merging the uncertain nodes, while still guar-anteeing the robustness of the solution. By the CMM, the transmission constraints are modified, with the parameters obtained based on an analytical solution of a uniform approximation problem, so that the computational time is negligi-ble. The CMM is applied under a greedy-algorithm based framework, where the number of merged nodes and the ap-proximation error can be well balanced. The CMM is designed as a preprocessing tool to improve the solution efficiency for robust TCUC problems and is compatible with many solution methods (like two-stage and multi-stage robust optimi-zation methods). Numerical tests show the method is effective.

math.OC↗

arXiv2018

Energy and system size dependent heavy flavor measurements at PHENIX at RHIC

Heavy flavor production is an ideal tool to study the properties of the Quark Gluon Plasma (QGP). The heavy flavor production at the Relativistic Heavy Ion Collider (RHIC) has its unique kinematic coverage and different production mechanisms from the Larger Hadron Collider (LHC) measurements. Heavy flavor products created in heavy ion collisions experience the whole evolution of nuclear medium. It's critical to measure both open and closed heavy flavor products in different collision systems to isolate cold/hot nuclear medium effects and initial/final state interactions. We report recent heavy flavor measurements at PHENIX in 200 GeV $p$+$p$, $p$+Al, $p$+Au, $^{3}$He+Au and Au+Au collisions that include: the correlated di-muon analysis in 200 GeV $p$+$p$ and $p$+Au collisions; the rapidity and $N_{part}$ dependent $J/ψ$ $R_{AB}$ measured in asymmetric small systems; open heavy flavor $v_{2}$ measured in 200 GeV d+Au and Au+Au collisions. These measurements provide further information about the heavy flavor production mechanism, initial and final state nuclear modification and flavor dependent energy loss in QGP.

hep-ex↗

arXiv2018

Robust Edge-Preserved Surface Mesh Polycube Deformation

The problem of polycube construction or deformation is an essential problem in computer graphics. In this paper, we present a robust, simple, efficient and automatic algorithm to deform the meshes of arbitrary shapes into their polycube ones. We derive a clear relationship between a mesh and its corresponding polycube shape. Our algorithm is edge-preserved, and works on surface meshes with or without boundaries. Our algorithm outperforms previous ones in speed, robustness, efficiency. Our method is simple to implement. To demonstrate the robustness and effectiveness of our method, we apply it to hundreds of models of varying complexity and topology. We demonstrat that our method compares favorably to other state-of-the-art polycube deformation methods.

cs.GR↗

arXiv2018

Conformal Mesh Parameterization Using Discrete Calabi Flow

In this paper, we introduce discrete Calabi flow to the graphics research community and present a novel conformal mesh parameterization algorithm. Calabi energy has a succinct and explicit format. Its corresponding flow is conformal and convergent under certain conditions. Our method is based on the Calabi energy and Calabi flow with solid theoretical and mathematical base. We demonstrate our approach on dozens of models and compare it with other related flow based methods, such as the well-known Ricci flow and CETM. Our experiments show that the performance of our algorithm is comparably the same with other methods. The discrete Calabi flow in our method provides another perspective on conformal flow and conformal parameterization.

cs.GR↗

arXiv2017

The ParallelEye Dataset: Constructing Large-Scale Artificial Scenes for Traffic Vision Research

Video image datasets are playing an essential role in design and evaluation of traffic vision algorithms. Nevertheless, a longstanding inconvenience concerning image datasets is that manually collecting and annotating large-scale diversified datasets from real scenes is time-consuming and prone to error. For that virtual datasets have begun to function as a proxy of real datasets. In this paper, we propose to construct large-scale artificial scenes for traffic vision research and generate a new virtual dataset called "ParallelEye". First of all, the street map data is used to build 3D scene model of Zhongguancun Area, Beijing. Then, the computer graphics, virtual reality, and rule modeling technologies are utilized to synthesize large-scale, realistic virtual urban traffic scenes, in which the fidelity and geography match the real world well. Furthermore, the Unity3D platform is used to render the artificial scenes and generate accurate ground-truth labels, e.g., semantic/instance segmentation, object bounding box, object tracking, optical flow, and depth. The environmental conditions in artificial scenes can be controlled completely. As a result, we present a viable implementation pipeline for constructing large-scale artificial scenes for traffic vision research. The experimental results demonstrate that this pipeline is able to generate photorealistic virtual datasets with low modeling time and high accuracy labeling.

cs.CV↗

arXiv2017

Training and Testing Object Detectors with Virtual Images

In the area of computer vision, deep learning has produced a variety of state-of-the-art models that rely on massive labeled data. However, collecting and annotating images from the real world has a great demand for labor and money investments and is usually too passive to build datasets with specific characteristics, such as small area of objects and high occlusion level. Under the framework of Parallel Vision, this paper presents a purposeful way to design artificial scenes and automatically generate virtual images with precise annotations. A virtual dataset named ParallelEye is built, which can be used for several computer vision tasks. Then, by training the DPM (Deformable Parts Model) and Faster R-CNN detectors, we prove that the performance of models can be significantly improved by combining ParallelEye with publicly available real-world datasets during the training phase. In addition, we investigate the potential of testing the trained models from a specific aspect using intentionally designed virtual datasets, in order to discover the flaws of trained models. From the experimental results, we conclude that our virtual dataset is viable to train and test the object detectors.

cs.CV↗

arXiv2017

DancingLines: An Analytical Scheme to Depict Cross-Platform Event Popularity

Nowadays, events usually burst and are propagated online through multiple modern media like social networks and search engines. There exists various research discussing the event dissemination trends on individual medium, while few studies focus on event popularity analysis from a cross-platform perspective. Challenges come from the vast diversity of events and media, limited access to aligned datasets across different media and a great deal of noise in the datasets. In this paper, we design DancingLines, an innovative scheme that captures and quantitatively analyzes event popularity between pairwise text media. It contains two models: TF-SW, a semantic-aware popularity quantification model, based on an integrated weight coefficient leveraging Word2Vec and TextRank; and wDTW-CD, a pairwise event popularity time series alignment model matching different event phases adapted from Dynamic Time Warping. We also propose three metrics to interpret event popularity trends between pairwise social platforms. Experimental results on eighteen real-world event datasets from an influential social network and a popular search engine validate the effectiveness and applicability of our scheme. DancingLines is demonstrated to possess broad application potentials for discovering the knowledge of various aspects related to events and different media.

cs.IR↗

arXiv2017

Energy dependent forward B $\rightarrow$ $J/ψ$ measurements in p+p collisions at PHENIX

The heavy flavor studies at RHIC help improve the knowledge of the bottom/charm quark production and can test Quantum Chromodynamics (QCD). Compared to the LHC/Tevatron, the RHIC heavy flavor production originates from different partonic sub-processes and has a complementary kinematic coverage. The PHENIX forward rapidity silicon vertex detector (FVTX) provides precise determination of the event vertex, tracking and the Distance of Closest Approach (DCA) of charged tracks. This detector allows direct access to the $B$ meson production via measurements of non-prompt $J/ψ\to μ^{+} + μ^{-}$ within $1.2<|y|<2.2$ rapidity in $p$+$p$ collisions at $\sqrt{s} = $ 510 and 200 GeV. Comparison among PHENIX measurements of the $B \rightarrow J/ψ$ fraction with integrated $J/ψ$ $p_{T}$ up to 5 GeV$/c$ and higher energy results at the Tevatron and the LHC presents a smooth center of mass energy dependence from 0.2 to 13 TeV in $p$+$p$ ($p$+$\bar{p}$) collisions. The Next-To-Leading order Perturbative QCD (NLO pQCD) calculations are in reasonable agreement with the extracted total $b\bar{b}$ cross section based on the $B \rightarrow J/ψ$ fraction measurements at PHENIX.

nucl-ex↗