arXiv subjects

Shuang Li

Publications and source records attributed to Shuang Li.

At least 235 records · Page 13Linked to original sources

arXiv2020

An Approach for Process Model Extraction By Multi-Grained Text Classification

Process model extraction (PME) is a recently emerged interdiscipline between natural language processing (NLP) and business process management (BPM), which aims to extract process models from textual descriptions. Previous process extractors heavily depend on manual features and ignore the potential relations between clues of different text granularities. In this paper, we formalize the PME task into the multi-grained text classification problem, and propose a hierarchical neural network to effectively model and extract multi-grained information without manually-defined procedural features. Under this structure, we accordingly propose the coarse-to-fine (grained) learning mechanism, training multi-grained tasks in coarse-to-fine grained order to share the high-level knowledge for the low-level tasks. To evaluate our approach, we construct two multi-grained datasets from two different domains and conduct extensive experiments from different dimensions. The experimental results demonstrate that our approach outperforms the state-of-the-art methods with statistical significance and further investigations demonstrate its effectiveness.

cs.CL↗

arXiv2020

A Mobile Robot Hand-Arm Teleoperation System by Vision and IMU

In this paper, we present a multimodal mobile teleoperation system that consists of a novel vision-based hand pose regression network (Transteleop) and an IMU-based arm tracking method. Transteleop observes the human hand through a low-cost depth camera and generates not only joint angles but also depth images of paired robot hand poses through an image-to-image translation process. A keypoint-based reconstruction loss explores the resemblance in appearance and anatomy between human and robotic hands and enriches the local features of reconstructed images. A wearable camera holder enables simultaneous hand-arm control and facilitates the mobility of the whole teleoperation system. Network evaluation results on a test dataset and a variety of complex manipulation tasks that go beyond simple pick-and-place operations show the efficiency and stability of our multimodal teleoperation system.

cs.RO↗

arXiv2020

Massive MIMO Asymptotics for Ray-Based Propagation Channels

Favorable propagation (FP) and channel hardening (CH) are desired properties in massive multiple-input multiple-output (MIMO) systems. To date, these properties have primarily been analyzed for classical \textit{statistical} channel models, or \textit{ray-based} models with very specific angular parameters and distributions. This paper presents a thorough mathematical analysis of the asymptotic system behavior for ray-based channels with \textit{arbitrary} ray distributions, and considers \textit{two} types of antenna array structures at the cellular base station: a uniform linear array (ULA) and a uniform planar array (UPA). In addition to FP and channel hardening, we analyze the \textit{large system potential} (LSP) which measures the asymptotic ratio of the expected power in the desired channel to the expected total interference power when both the antenna and user numbers grow. LSP is said to hold when this ratio converges to a positive constant. The results demonstrate that while FP is guaranteed in ray-based channels, CH may or may not occur depending on the nature of the model. Furthermore, we demonstrate that LSP will not normally hold as the expected interference power grows logarithmically for both ULAs and UPAs relative to the power in the desired channel as the system size increases. Nevertheless, we identify some fundamental and attractive properties of massive MIMO in this limiting regime.

cs.IT↗

arXiv2020

Generative Adversarial User Model for Reinforcement Learning Based Recommendation System

There are great interests as well as many challenges in applying reinforcement learning (RL) to recommendation systems. In this setting, an online user is the environment; neither the reward function nor the environment dynamics are clearly defined, making the application of RL challenging. In this paper, we propose a novel model-based reinforcement learning framework for recommendation systems, where we develop a generative adversarial network to imitate user behavior dynamics and learn her reward function. Using this user model as the simulation environment, we develop a novel Cascading DQN algorithm to obtain a combinatorial recommendation policy which can handle a large number of candidate items efficiently. In our experiments with real data, we show this generative adversarial user model can better explain user behavior than alternatives, and the RL policy based on this model can lead to a better long-term reward for the user and higher click rate for the system.

cs.LG↗

arXiv2019

Data-driven extraction of heavy quark diffusion in quark-gluon plasma

Heavy quark production provides a unique probe of the quark-gluon plasma transport properties in heavy ion collisions. Experimental observables like the nuclear modification factor $R_{\rm AA}$ and elliptic anisotropy $v_{2}$ of heavy flavor mesons are sensitive to the heavy quark diffusion coefficient. There now exist an extensive set of such measurements, which allow a data-driven extraction of this coefficient. In this work, we make such an attempt within our recently developed heavy quark transport modeling framework (Langevin-transport with Gluon Radiation, LGR). A question of particular interest is the temperature dependence of the diffusion coefficient, for which we test a wide range of possibility and draw constraints by comparing relevant charm meson data with model results. We find that a relatively strong increase of diffusion coefficient from crossover temperature $T_c$ toward high temperature is preferred by data. We also make predictions for Bottom meson observables for further experimental tests.

hep-ph↗

arXiv2019

The Landscape of Non-convex Empirical Risk with Degenerate Population Risk

The landscape of empirical risk has been widely studied in a series of machine learning problems, including low-rank matrix factorization, matrix sensing, matrix completion, and phase retrieval. In this work, we focus on the situation where the corresponding population risk is a degenerate non-convex loss function, namely, the Hessian of the population risk can have zero eigenvalues. Instead of analyzing the non-convex empirical risk directly, we first study the landscape of the corresponding population risk, which is usually easier to characterize, and then build a connection between the landscape of the empirical risk and its population risk. In particular, we establish a correspondence between the critical points of the empirical risk and its population risk without the strongly Morse assumption, which is required in existing literature but not satisfied in degenerate scenarios. We also apply the theory to matrix sensing and phase retrieval to demonstrate how to infer the landscape of empirical risk from that of the corresponding population risk.

math.OC↗

arXiv2019

Atomic Norm Denoising for Complex Exponentials with Unknown Waveform Modulations

Non-stationary blind super-resolution is an extension of the traditional super-resolution problem, which deals with the problem of recovering fine details from coarse measurements. The non-stationary blind super-resolution problem appears in many applications including radar imaging, 3D single-molecule microscopy, computational photography, etc. There is a growing interest in solving non-stationary blind super-resolution task with convex methods due to their robustness to noise and strong theoretical guarantees. Motivated by the recent work on atomic norm minimization in blind inverse problems, we focus here on the signal denoising problem in non-stationary blind super-resolution. In particular, we use an atomic norm regularized least-squares problem to denoise a sum of complex exponentials with unknown waveform modulations. We quantify how the mean square error depends on the noise variance and the true signal parameters. Numerical experiments are also implemented to illustrate the theoretical result.

cs.IT↗

arXiv2019

Iterative Hard Thresholding for Low CP-rank Tensor Models

Recovery of low-rank matrices from a small number of linear measurements is now well-known to be possible under various model assumptions on the measurements. Such results demonstrate robustness and are backed with provable theoretical guarantees. However, extensions to tensor recovery have only recently began to be studied and developed, despite an abundance of practical tensor applications. Recently, a tensor variant of the Iterative Hard Thresholding method was proposed and theoretical results were obtained that guarantee exact recovery of tensors with low Tucker rank. In this paper, we utilize the same tensor version of the Restricted Isometry Property (RIP) to extend these results for tensors with low CANDECOMP/PARAFAC (CP) rank. In doing so, we leverage recent results on efficient approximations of CP decompositions that remove the need for challenging assumptions in prior works. We complement our theoretical findings with empirical results that showcase the potential of the approach.

math.NA↗

arXiv2019

Making Sense of Audio Vibration for Liquid Height Estimation in Robotic Pouring

In this paper, we focus on the challenging perception problem in robotic pouring. Most of the existing approaches either leverage visual or haptic information. However, these techniques may suffer from poor generalization performances on opaque containers or concerning measuring precision. To tackle these drawbacks, we propose to make use of audio vibration sensing and design a deep neural network PouringNet to predict the liquid height from the audio fragment during the robotic pouring task. PouringNet is trained on our collected real-world pouring dataset with multimodal sensing data, which contains more than 3000 recordings of audio, force feedback, video and trajectory data of the human hand that performs the pouring task. Each record represents a complete pouring procedure. We conduct several evaluations on PouringNet with our dataset and robotic hardware. The results demonstrate that our PouringNet generalizes well across different liquid containers, positions of the audio receiver, initial liquid heights and types of liquid, and facilitates a more robust and accurate audio-based perception for robotic pouring.

cs.RO↗

arXiv2019

Probing the transport properties of Quark-Gluon Plasma via heavy-flavor Boltzmann and Langevin dynamics

The heavy quark propagation behavior inside the quark-gluon plasma (QGP), is usually described in terms of the Boltzmann dynamics, which can be reduced to the Langevin approach by assuming a small momentum transfer for the scattering processes between heavy quarks and the QGP constituents. In this work, the temperature and energy dependence of the transport coefficients are calculated in the framework of both Boltzmann and Langevin dynamics. The derived transport coefficients are found to be systematically larger in the Boltzmann approach as compared with the Langevin, in particular in the high temperature and high energy region. Within each of the two theoretical frameworks, we simulate the charm quark production and the subsequent evolution processes in relativistic heavy-ion collisions. We find that the total in-medium energy loss is larger from the Langevin dynamics, resulting in a smaller (larger) $R_{\rm AA}$ at high (low) $p_{\rm T}$, for both the charm quark and heavy-flavor mesons. Meanwhile, the Boltzmann model is found to induce larger $v_{\rm 2}$, in particular at moderate $p_{\rm T}$, as well as stronger broadening behavior for the azimuthal distributions. By comparing the model calculations with available experimental measurements for D-mesons, we find that the Langevin approach is more favored by the $R_{\rm AA}$ data while the Boltzmann approach is more favored favor by the $v_{\rm 2}$ data. A simultaneous description of both observables appear challenging for both models.

hep-ph↗

arXiv2019

Additive manufacturing of ceramics from preceramic polymers: A versatile stereolithographic approach assisted by thiol-ene click chemistry

Here we introduce a versatile stereolithographic route to produce three different kinds of Si-containing thermosets that yield high performance ceramics upon thermal treatment. Our approach is based on a fast and inexpensive thiol-ene free radical addition that can be applied for different classes of preceramic polymers with carbon-carbon double bonds. Due to the rapidity and efficiency of the thiol-ene click reactions, this additive manufacturing process can be effectively carried out using conventional light sources on benchtop printers. Through light initiated cross-linking, the liquid preceramic polymers transform into stable infusible thermosets that preserve their shape during the polymer-to-ceramic transformation. Through pyrolysis the thermosets transform into glassy ceramics with uniform shrinkage and high density. The obtained ceramic structures are nearly fully dense, have smooth surfaces, and are free from macroscopic voids and defects. A fabricated SiOC honeycomb was shown to exhibit a significantly higher compressive strength to weight ratio in comparison to other porous ceramics.

physics.app-ph↗

arXiv2019

A Sketch of Some Stochastic Models and Analysis Methods for Fiber Bundle Failure under Increasing Tensile Load

Fiber bundle models (FBM's) have been used to model the failure of fibrous composites as load-sharing systems since the 1960's when Rosen (1964 and 1965) conducted some remarkable experiments on unidirectional fibrous composites. These experiments gave seminal insights into their failure under increasing tensile load. However, over the last thirty years FBM's have been used to model catastrophic failure in other situations by the physical science community and others. The purpose of this paper is to sketch some research on load-sharing models and statistical analysis methods that have been overlooked by this community. These are illustrated by summarizing the findings regarding Rosen's Specimen A experiments and presenting the necessary results needed for this. Related research about the bundle breaking strength distribution and the joint distribution (the Gibbs measure) regarding the state (failed or unfailed) of the bundle components at a given load per component, $s$, is also given.

stat.AP↗

arXiv2019

Jet shape modifications at the LHC energies by JEWEL

Jet shape measurements are strongly suggested to explore the microscopic evolution mechanism of parton-medium interaction in ultra-relativistic heavy-ion collisions. In this paper, jet shape modifications are quantified by fragmentation function $F(z)$, relative momentum $p_{T}^{rel}$, density of charged particles $ρ(r)$, jet angularity $girth$, jet momentum dispersion $p_{T}^{Disp}$ and $LeSub$ for proton-proton collisions at $900GeV$, $5.02TeV$, $7TeV$ and $13TeV$, as well as for lead-lead collisions at $2.76TeV$ and $5.02TeV$. A differential jet shape parameters $D_{girth}$ is proposed and studied at smaller-radius jet $r<0.3$. The results indicate that medium effect is dominant for jet shape modifications, which is less dependent on center of mass energy. The jet fragmentation is enhanced significantly at very low $z<0.02$ and fragmented jet constituents are spread to larger jet-radius linearly for $p_{T}^{rel}<1$. The waveform attenuate phenomena is observed in $p_{T}^{rel}$, $girth$ and $D_{girth}$ distributions. The comparison results on $D_{girth}$ from $pp$ to $Pb+Pb$ where the wave-like distribution in $pp$ collision is ahead of $Pb+Pb$ collisions in small jet-radius intervals, is interesting to hint the medium effect.

hep-ph↗

arXiv2019

Vision-based Teleoperation of Shadow Dexterous Hand using End-to-End Deep Neural Network

In this paper, we present TeachNet, a novel neural network architecture for intuitive and markerless vision-based teleoperation of dexterous robotic hands. Robot joint angles are directly generated from depth images of the human hand that produce visually similar robot hand poses in an end-to-end fashion. The special structure of TeachNet, combined with a consistency loss function, handles the differences in appearance and anatomy between human and robotic hands. A synchronized human-robot training set is generated from an existing dataset of labeled depth images of the human hand and simulated depth images of a robotic hand. The final training set includes 400K pairwise depth images and joint angles of a Shadow C6 robotic hand. The network evaluation results verify the superiority of TeachNet, especially regarding the high-precision condition. Imitation experiments and grasp tasks teleoperated by novice users demonstrate that TeachNet is more reliable and faster than the state-of-the-art vision-based teleoperation method.

cs.RO↗

arXiv2019

PointNetGPD: Detecting Grasp Configurations from Point Sets

In this paper, we propose an end-to-end grasp evaluation model to address the challenging problem of localizing robot grasp configurations directly from the point cloud. Compared to recent grasp evaluation metrics that are based on handcrafted depth features and a convolutional neural network (CNN), our proposed PointNetGPD is lightweight and can directly process the 3D point cloud that locates within the gripper for grasp evaluation. Taking the raw point cloud as input, our proposed grasp evaluation network can capture the complex geometric structure of the contact area between the gripper and the object even if the point cloud is very sparse. To further improve our proposed model, we generate a larger-scale grasp dataset with 350k real point cloud and grasps with the YCB object set for training. The performance of the proposed model is quantitatively measured both in simulation and on robotic hardware. Experiments on object grasping and clutter removal show that our proposed model generalizes well to novel objects and outperforms state-of-the-art methods. Code and video are available at \href{https://lianghongzhuo.github.io/PointNetGPD}{https://lianghongzhuo.github.io/PointNetGPD}

cs.RO↗

arXiv2019

Progressively Growing Generative Adversarial Networks for High Resolution Semantic Segmentation of Satellite Images

Machine learning has proven to be useful in classification and segmentation of images. In this paper, we evaluate a training methodology for pixel-wise segmentation on high resolution satellite images using progressive growing of generative adversarial networks. We apply our model to segmenting building rooftops and compare these results to conventional methods for rooftop segmentation. We present our findings using the SpaceNet version 2 dataset. Progressive GAN training achieved a test accuracy of 93% compared to 89% for traditional GAN training.

cs.CV↗

arXiv2018

Scan $B$-Statistic for Kernel Change-Point Detection

Detecting the emergence of an abrupt change-point is a classic problem in statistics and machine learning. Kernel-based nonparametric statistics have been used for this task which enjoy fewer assumptions on the distributions than the parametric approach and can handle high-dimensional data. In this paper we focus on the scenario when the amount of background data is large, and propose two related computationally efficient kernel-based statistics for change-point detection, which are inspired by the recently developed $B$-statistics. A novel theoretical result of the paper is the characterization of the tail probability of these statistics using the change-of-measure technique, which focuses on characterizing the tail of the detection statistics rather than obtaining its asymptotic distribution under the null distribution. Such approximations are crucial to control the false alarm rate, which corresponds to the significance level in offline change-point detection and the average-run-length in online change-point detection. Our approximations are shown to be highly accurate. Thus, they provide a convenient way to find detection thresholds for both offline and online cases without the need to resort to the more expensive simulations or bootstrapping. We show that our methods perform well on both synthetic data and real data.

cs.LG↗

arXiv2018

Charm-strange meson production in ultra-relativistic heavy-ion collisions at the CERN-LHC energies

The nuclear modification factor ${R}_{\rm AA}$ and the elliptic flow coefficient ${v}_{\rm 2}$ of charm-strange meson $D^{+}_{s}$ is systematically studied in Pb--Pb collisions at $\sqrt{s_{\rm NN}}=5.02~{\rm TeV}$ and $2.76~{\rm TeV}$. During the modeling, the coupling strength between the injected charm quark and the incident medium constituents, is extracted from the lattice QCD calculations: $2πTD_{s}=7$ (\textbf{Model-A}) and $2πTD_{s}=1.3 + (T/T_{c})^2$ (\textbf{Model-B}). We find that, comparing ${R}_{\rm AA}(D^{+}_{s})$ with ${R}_{\rm AA}(non-strange)$, the heavy-light coalescence effect is more pronounced for the former one, resulting in an enhancement behavior in the range $2\lesssim {p}_{\rm T}\lesssim5~{\rm GeV}$. The predictions of ${R}_{\rm AA}(D^{+}_{s})$ and ${R}_{\rm AA}(non-strange)$ favor Model-A to have a better description of the measured ${p}_{\rm T}$ dependence in both energies, while their ${v}_{\rm 2}$ prefer Model-B at moderate ${p}_{\rm T}$ ($2\lesssim {p}_{\rm T}\lesssim4~{\rm GeV}$). Therefore, it is necessary to consider the temperature- and/or momentum-dependence of $2πTD_{s}$ to describe simultaneously ${R}_{\rm AA}(D^{+}_{s})$ and ${v}_{\rm 2}(D^{+}_{s})$ in different centrality classes in Pb--Pb collisions.

hep-ph↗