arXiv subjects

Heng Wang

Publications and source records attributed to Heng Wang.

At least 199 records · Page 11Linked to original sources

arXiv2019

Video Classification with Channel-Separated Convolutional Networks

Group convolution has been shown to offer great computational savings in various 2D convolutional architectures for image classification. It is natural to ask: 1) if group convolution can help to alleviate the high computational cost of video classification networks; 2) what factors matter the most in 3D group convolutional networks; and 3) what are good computation/accuracy trade-offs with 3D group convolutional networks. This paper studies the effects of different design choices in 3D group convolutional networks for video classification. We empirically demonstrate that the amount of channel interactions plays an important role in the accuracy of 3D group convolutional networks. Our experiments suggest two main findings. First, it is a good practice to factorize 3D convolutions by separating channel interactions and spatiotemporal interactions as this leads to improved accuracy and lower computational cost. Second, 3D channel-separated convolutions provide a form of regularization, yielding lower training accuracy but higher test accuracy compared to 3D convolutions. These two empirical findings lead us to design an architecture -- Channel-Separated Convolutional Network (CSN) -- which is simple, efficient, yet accurate. On Sports1M, Kinetics, and Something-Something, our CSNs are comparable with or better than the state-of-the-art while being 2-3 times more efficient.

cs.CV↗

arXiv2019

Photo-thermoelectric properties and their use in study of transport properties of both carriers from a single bulk sample

We describe a theory on photo-thermoelectric properties of a semiconductor, which include photo-conductivity, photo-Seebeck coefficient, and photo-Hall effect. We demonstrate that these properties provide a powerful tool for the study of carrier transport in semiconductors. Even though photo-carrier generation is a complicated process which often prohibits quantitative analysis as their species or numbers are not known. Using bulk samples seems even less likely as the photo-carrier only affect a thin layer. Our method will allow researchers to bypass these difficulties, to use only measured properties and determine both electron and hole mobilities as well as the ratio between electrons and holes from a bulk sample. We provide initial experiment verification of our theory in the end using two distinctively different semiconductors.

cond-mat.mtrl-sci↗

arXiv2019

Tackling Challenges in Seebeck Coefficient Measurement of Ultra-High Resistance Samples with an AC Technique

Seebeck coefficient is a widely-studied semiconductor property. Conventional Seebeck coefficient measurements are based on DC voltage measurement. Normally this is performed on samples with low resistances below a few Mohm level. Meanwhile, certain semiconductors are highly intrinsic and resistive, many examples can be found in optical and photovoltaic materials. The hybrid halide perovskites that have gained extensive attention recently are a good example. Few credible studies exist on the Seebeck coefficient of, CH3NH3PbI3, for example. We report here an AC technique based Seebeck coefficient measurement, which makes high quality voltage measurement on samples with resistances up to 100Gohm. This is achieved through a specifically designed setup to enhance sample isolation and reduce meter loading. As a demonstration, we performed Seebeck coefficient measurement of a CH3NH3PbI3 thin film at dark and found S = +550 microV/K. Such property of this material has not been successfully studied before.

physics.app-ph↗

arXiv2019

FASTER Recurrent Networks for Efficient Video Classification

Typical video classification methods often divide a video into short clips, do inference on each clip independently, then aggregate the clip-level predictions to generate the video-level results. However, processing visually similar clips independently ignores the temporal structure of the video sequence, and increases the computational cost at inference time. In this paper, we propose a novel framework named FASTER, i.e., Feature Aggregation for Spatio-TEmporal Redundancy. FASTER aims to leverage the redundancy between neighboring clips and reduce the computational cost by learning to aggregate the predictions from models of different complexities. The FASTER framework can integrate high quality representations from expensive models to capture subtle motion information and lightweight representations from cheap models to cover scene changes in the video. A new recurrent network (i.e., FAST-GRU) is designed to aggregate the mixture of different representations. Compared with existing approaches, FASTER can reduce the FLOPs by over 10x? while maintaining the state-of-the-art accuracy across popular datasets, such as Kinetics, UCF-101 and HMDB-51.

cs.CV↗

arXiv2019

Large-scale weakly-supervised pre-training for video action recognition

Current fully-supervised video datasets consist of only a few hundred thousand videos and fewer than a thousand domain-specific labels. This hinders the progress towards advanced video architectures. This paper presents an in-depth study of using large volumes of web videos for pre-training video models for the task of action recognition. Our primary empirical finding is that pre-training at a very large scale (over 65 million videos), despite on noisy social-media videos and hashtags, substantially improves the state-of-the-art on three challenging public action recognition datasets. Further, we examine three questions in the construction of weakly-supervised video action datasets. First, given that actions involve interactions with objects, how should one construct a verb-object pre-training label space to benefit transfer learning the most? Second, frame-based models perform quite well on action recognition; is pre-training for good image features sufficient or is pre-training for spatio-temporal features valuable for optimal transfer learning? Finally, actions are generally less well-localized in long videos vs. short videos; since action labels are provided at a video level, how should one choose video clips for best performance, given some fixed budget of number or minutes of videos?

cs.CV↗

arXiv2019

Defeats GAN: A Simpler Model Outperforms in Knowledge Representation Learning

The goal of knowledge representation learning is to embed entities and relations into a low-dimensional, continuous vector space. How to push a model to its limit and obtain better results is of great significance in knowledge graph's applications. We propose a simple and elegant method, Trans-DLR, whose main idea is dynamic learning rate control during training. Our method achieves remarkable improvement, compared with recent GAN-based method. Moreover, we introduce a new negative sampling trick which corrupts not only entities, but also relations, in different probabilities. We also develop an efficient way, which fully utilizes multiprocessing and parallel computing, to speed up evaluation of the model in link prediction tasks. Experiments show that our method is effective.

cs.LG↗

arXiv2019

Multi-task Learning for Chinese Word Usage Errors Detection

Chinese word usage errors often occur in non-native Chinese learners' writing. It is very helpful for non-native Chinese learners to detect them automatically when learning writing. In this paper, we propose a novel approach, which takes advantages of different auxiliary tasks, such as POS-tagging prediction and word log frequency prediction, to help the task of Chinese word usage error detection. With the help of these auxiliary tasks, we achieve the state-of-the-art results on the performances on the HSK corpus data, without any other extra data.

cs.CL↗

arXiv2019

Superconductivity above 28 K in single unit cell FeSe films interfaced with GaO$_{2-δ}$ layer on NdGaO$_{3}$(110)

We prepared single unit cell FeSe films on GaO$_{2-δ}$ terminated perovskite NdGaO$_{3}$(110) substrates and performed ex situ transport and scanning transmission electron microscopy measurements on the FeTe protected films. Our experimental measurements showed that the single unit cell FeSe films interfaced with GaO$_{2-δ}$ layer are electron doped via interface charge transfer. Most importantly, this type of heterostructure can host interface enhanced superconductivity with an onset temperature of about 28 K, which is much higher than the value of 8 K in bulk FeSe. Our work indicates that FeSe/GaO$_{2-δ}$can be a comparative platform with the FeSe/TiO$_{2-δ}$ family for understanding the mechanism of interface enhanced high temperature superconductivity.

cond-mat.supr-con↗

arXiv2019

Devon: Deformable Volume Network for Learning Optical Flow

State-of-the-art neural network models estimate large displacement optical flow in multi-resolution and use warping to propagate the estimation between two resolutions. Despite their impressive results, it is known that there are two problems with the approach. First, the multi-resolution estimation of optical flow fails in situations where small objects move fast. Second, warping creates artifacts when occlusion or dis-occlusion happens. In this paper, we propose a new neural network module, Deformable Cost Volume, which alleviates the two problems. Based on this module, we designed the Deformable Volume Network (Devon) which can estimate multi-scale optical flow in a single high resolution. Experiments show Devon is more suitable in handling small objects moving fast and achieves comparable results to the state-of-the-art methods in public benchmarks.

cs.CV↗

arXiv2019

Concept Drift Detection and Adaptation with Hierarchical Hypothesis Testing

A fundamental issue for statistical classification models in a streaming environment is that the joint distribution between predictor and response variables changes over time (a phenomenon also known as concept drifts), such that their classification performance deteriorates dramatically. In this paper, we first present a hierarchical hypothesis testing (HHT) framework that can detect and also adapt to various concept drift types (e.g., recurrent or irregular, gradual or abrupt), even in the presence of imbalanced data labels. A novel concept drift detector, namely Hierarchical Linear Four Rates (HLFR), is implemented under the HHT framework thereafter. By substituting a widely-acknowledged retraining scheme with an adaptive training strategy, we further demonstrate that the concept drift adaptation capability of HLFR can be significantly boosted. The theoretical analysis on the Type-I and Type-II errors of HLFR is also performed. Experiments on both simulated and real-world datasets illustrate that our methods outperform state-of-the-art methods in terms of detection precision, detection delay as well as the adaptability across different concept drift types.

stat.ML↗

arXiv2018

Text Generation Based on Generative Adversarial Nets with Latent Variable

In this paper, we propose a model using generative adversarial net (GAN) to generate realistic text. Instead of using standard GAN, we combine variational autoencoder (VAE) with generative adversarial net. The use of high-level latent random variables is helpful to learn the data distribution and solve the problem that generative adversarial net always emits the similar data. We propose the VGAN model where the generative model is composed of recurrent neural network and VAE. The discriminative model is a convolutional neural network. We train the model via policy gradient. We apply the proposed model to the task of text generation and compare it to other recent neural network based models, such as recurrent neural network language model and SeqGAN. We evaluate the performance of the model by calculating negative log-likelihood and the BLEU score. We conduct experiments on three benchmark datasets, and results show that our model outperforms other previous models.

cs.CL↗

arXiv2018

Overview of CAIL2018: Legal Judgment Prediction Competition

In this paper, we give an overview of the Legal Judgment Prediction (LJP) competition at Chinese AI and Law challenge (CAIL2018). This competition focuses on LJP which aims to predict the judgment results according to the given facts. Specifically, in CAIL2018 , we proposed three subtasks of LJP for the contestants, i.e., predicting relevant law articles, charges and prison terms given the fact descriptions. CAIL2018 has attracted several hundreds participants (601 teams, 1, 144 contestants from 269 organizations). In this paper, we provide a detailed overview of the task definition, related works, outstanding methods and competition results in CAIL2018.

cs.AI↗

arXiv2018

CAIL2018: A Large-Scale Legal Dataset for Judgment Prediction

In this paper, we introduce the \textbf{C}hinese \textbf{AI} and \textbf{L}aw challenge dataset (CAIL2018), the first large-scale Chinese legal dataset for judgment prediction. \dataset contains more than $2.6$ million criminal cases published by the Supreme People's Court of China, which are several times larger than other datasets in existing works on judgment prediction. Moreover, the annotations of judgment results are more detailed and rich. It consists of applicable law articles, charges, and prison terms, which are expected to be inferred according to the fact descriptions of cases. For comparison, we implement several conventional text classification baselines for judgment prediction and experimental results show that it is still a challenge for current models to predict the judgment results of legal cases, especially on prison terms. To help the researchers make improvements on legal judgment prediction, both \dataset and baselines will be released after the CAIL competition\footnote{http://cail.cipsc.org.cn/}.

cs.CL↗

arXiv2018

A Closer Look at Spatiotemporal Convolutions for Action Recognition

In this paper we discuss several forms of spatiotemporal convolutions for video analysis and study their effects on action recognition. Our motivation stems from the observation that 2D CNNs applied to individual frames of the video have remained solid performers in action recognition. In this work we empirically demonstrate the accuracy advantages of 3D CNNs over 2D CNNs within the framework of residual learning. Furthermore, we show that factorizing the 3D convolutional filters into separate spatial and temporal components yields significantly advantages in accuracy. Our empirical study leads to the design of a new spatiotemporal convolutional block "R(2+1)D" which gives rise to CNNs that achieve results comparable or superior to the state-of-the-art on Sports-1M, Kinetics, UCF101 and HMDB51.

cs.CV↗

arXiv2016

Creating arbitrary quantum vibrational states in a carbon nanotube

We theoretically study the creation of single- and multi-phonon Fock states and arbitrary superpositions of quantum phonon states in a nanomechanical carbon nanotube (CNT) resonator. In our model, a doubly clamped CNT resonator is initialized in the ground state and a single electron is trapped in a quantum dot which is formed by a electric gate potential and brought into the magnetic field of a micro-magnet. The preparation of arbitrary quantum phonon states is based on the coupling between the mechanical motion of the CNT and the electron spin which acts as a non-linearity. We assume that electrical driving pulses with different frequencies are applied on the system. The quantum information is transferred from the spin qubit to the mechanical motion by the spin-phonon coupling and the electron spin qubit can be reset by the single-electron spin resonance. We describe Wigner tomography which can be applied at the end to obtain the phase information of the prepared phonon states.

cond-mat.mes-hall↗

arXiv2015

Face Aging Effect Simulation using Hidden Factor Analysis Joint Sparse Representation

Face aging simulation has received rising investigations nowadays, whereas it still remains a challenge to generate convincing and natural age-progressed face images. In this paper, we present a novel approach to such an issue by using hidden factor analysis joint sparse representation. In contrast to the majority of tasks in the literature that handle the facial texture integrally, the proposed aging approach separately models the person-specific facial properties that tend to be stable in a relatively long period and the age-specific clues that change gradually over time. It then merely transforms the age component to a target age group via sparse reconstruction, yielding aging effects, which is finally combined with the identity component to achieve the aged face. Experiments are carried out on three aging databases, and the results achieved clearly demonstrate the effectiveness and robustness of the proposed method in rendering a face with aging effects. Additionally, a series of evaluations prove its validity with respect to identity preservation and aging effect generation.

cs.CV↗

arXiv2015

Mechanically induced two-qubit gates and maximally entangled states for single electron spins in a carbon nanotube

We theoretically analyze a system where two electrons are trapped separately in two quantum dots on a suspended carbon nanotube (CNT), subject to external ac electric driving. An indirect mechanically-induced coupling of two distant single electron spins is induced by the interaction between the spins and the mechanical motion of the CNT. We show that a two-qubit iSWAP gate and arbitrary single-qubit gates can be obtained from the intrinsic spin-orbit coupling. Combining the iSWAP gate and single-qubit gates, maximally entangled states of two spins can be generated in a single step by varying the frequency and the strength of the external electric driving field. The spin-phonon coupling can be turned off by electrostatically shifting the electron wave function on the nanotube.

cond-mat.mes-hall↗

arXiv2015

Concept Drift Detection for Streaming Data

Common statistical prediction models often require and assume stationarity in the data. However, in many practical applications, changes in the relationship of the response and predictor variables are regularly observed over time, resulting in the deterioration of the predictive performance of these models. This paper presents Linear Four Rates (LFR), a framework for detecting these concept drifts and subsequently identifying the data points that belong to the new concept (for relearning the model). Unlike conventional concept drift detection approaches, LFR can be applied to both batch and stream data; is not limited by the distribution properties of the response variable (e.g., datasets with imbalanced labels); is independent of the underlying statistical-model; and uses user-specified parameters that are intuitively comprehensible. The performance of LFR is compared to benchmark approaches using both simulated and commonly used public datasets that span the gamut of concept drift types. The results show LFR significantly outperforms benchmark approaches in terms of recall, accuracy and delay in detection of concept drifts across datasets.

stat.ML↗