arXiv subjects

Chao Li

Publications and source records attributed to Chao Li.

At least 505 records · Page 28Linked to original sources

arXiv2019

Deep Concept-wise Temporal Convolutional Networks for Action Localization

Existing action localization approaches adopt shallow temporal convolutional networks (\ie, TCN) on 1D feature map extracted from video frames. In this paper, we empirically find that stacking more conventional temporal convolution layers actually deteriorates action classification performance, possibly ascribing to that all channels of 1D feature map, which generally are highly abstract and can be regarded as latent concepts, are excessively recombined in temporal convolution. To address this issue, we introduce a novel concept-wise temporal convolution (CTC) layer as an alternative to conventional temporal convolution layer for training deeper action localization networks. Instead of recombining latent concepts, CTC layer deploys a number of temporal filters to each concept separately with shared filter parameters across concepts. Thus can capture common temporal patterns of different concepts and significantly enrich representation ability. Via stacking CTC layers, we proposed a deep concept-wise temporal convolutional network (C-TCN), which boosts the state-of-the-art action localization performance on THUMOS'14 from 42.8 to 52.1 in terms of mAP(\%), achieving a relative improvement of 21.7\%. Favorable result is also obtained on ActivityNet.

cs.CV↗

arXiv2019

On the Differential-Linear Connectivity Table of Vectorial Boolean Functions

Vectorial Boolean functions are crucial building-blocks in symmetric ciphers. Different known attacks on block ciphers have resulted in diverse cryptographic criteria for vectorial Boolean functions, such as differential uniformity and nonlinearity. Very recently, Bar-On et al. introduced at Eurocrypt'19 a new tool, called the differential-linear connectivity table (DLCT), which allows for taking into account the dependency between the two subciphers $E_0$ and $E_1$ involved in differential-linear attacks. This new notion leads to significant improvements of differential-linear attacks on several ciphers. This paper presents a theoretical characterization of the DLCT of vectorial Boolean functions and also investigates this new criterion for some families of functions with specific forms. More precisely, we firstly reveal the connection between the DLCT and the autocorrelation of vectorial Boolean functions, we characterize properties of the DLCT by means of the Walsh transform of the function and of its differential distribution table, and we present generic bounds on the highest magnitude occurring in the DLCT of vectorial Boolean functions, which coincides (up to a factor~$2$) with the well-established notion of absolute indicator. Next, we investigate the invariance property of the DLCT of vectorial Boolean functions under the affine, extended-affine, and Carlet-Charpin-Zinoviev (CCZ) equivalence and exhaust the DLCT spectra of optimal $4$-bit S-boxes under affine equivalence. Furthermore, we study the DLCT of APN, plateaued and AB functions and establish its connection with other cryptographic criteria. Finally, we investigate the DLCT and the absolute indicator of some specific polynomials with optimal or low differential uniformity, including monomials, cubic functions, quadratic functions and inverses of quadratic permutations.

cs.IT↗

arXiv2019

Hour-scale persistent negative anomaly of atmospheric electrostatic field near the epicenter before earthquake

Although earthquake prediction is a big challenge in the world, some simple observational tools can capture many physical signals and demonstrate that an earthquake (EQ) may be forthcoming in short period. Many researchers have studied the significant variation of atmospheric electrostatic field related to the forthcoming earthquake. However, until now, there is not a compelling physical mechanism which can explain why atmospheric electrostatic abnormal signal could appear just before an earthquake. Here we present a precursor signal and propose a brief physical interpretation. Under fair air conditions, if the near-surface atmospheric electrostatic field Ez (oriented down when it is positive) presents a very stable negative anomaly (from -100 V/m to -5000 V/m), it will forebode that an earthquake (seismicity from 3-8) would take place in the next several to tens of hours within a distance less than 100 km. We name this prediction technique as "DC Ez Determination"(DED). In addition, the mechanism of such abnormal quasi-static electric field before a forthcoming earthquake has been proposed here: (1) Radon gas releases from the rock clefts near the epicenter during seismogenic process. (2) α particles are produced due to the radioactive decay of Radon gas in the air. (3) α particle ionizing radiation creates more positive particles in the air, which is much more effective than that β and γ particles produced during Radon radioactive decay. (4) The new positive particles change formal positive atmospheric electric field (Ez) into stably negative near the earth-surface. (5) The closer the instrument is to the epicenter, the more likely it is to observe the negative Ez signal related to the earthquake. It is recommended to establish an instrument network to capture the reliable precursor and make some warnings before the earthquake disaster.

physics.geo-ph↗

arXiv2019

On the Differential Linear Connectivity Table of Vectorial Boolean Functions

Vectorial Boolean functions are crucial building blocks in symmetric ciphers. Different known attacks on block ciphers have resulted in diverse cryptographic criteria of vectorial Boolean functions,such as differential distribution table and nonlinearity. Very recently, Bar-On et al. introduced at Eurocrypt'19 a new tool, called the Differential-Linear Connectivity Table (DLCT).This paper is a follow-up work, which presents further theoretical characterization of the DLCT of vectorial Boolean functions and also investigates this new criterion of functions with certain forms. In this paper we introduce a generalized concept of the additive autocorrelation, which is extended from Boolean functions to the vectorial Boolean functions, and use it as a main tool to investigate the DLCT property of vectorial Boolean functions. Firstly, by establishing a connection between the DLCT and the additive autocorrelation, we characterize properties of DLCT by means of the Walsh transform and the differential distribution table, and present generic lower bounds on the differential-linear uniformity (DLU) of vectorial Boolean functions. Furthermore, we investigate the DLCT property of monomials, APN, plateaued and AB functions. Our study reveals that the DLCT of these special functions are closely related to other cryptographic criteria. Next, we prove that the DLU of vectorial Boolean functions is invariant underthe EA equivalence but not invariant under the CCZ equivalence, and that the DLCT spectrum is only invariant under affine equivalence. In addition, under affine equivalence, we exhaust the DLCT spectra and DLU of optimal S-boxes with $4$ bit by Magma. Finally, we investigate the DLCT spectra and DLU of some polynomials over $F_{2^n}$, including the inverse, Gold, Bracken-Leander power functions and all quadratic polynomials.

cs.IT↗

arXiv2019

Best Practices for Learning Domain-Specific Cross-Lingual Embeddings

Cross-lingual embeddings aim to represent words in multiple languages in a shared vector space by capturing semantic similarities across languages. They are a crucial component for scaling tasks to multiple languages by transferring knowledge from languages with rich resources to low-resource languages. A common approach to learning cross-lingual embeddings is to train monolingual embeddings separately for each language and learn a linear projection from the monolingual spaces into a shared space, where the mapping relies on a small seed dictionary. While there are high-quality generic seed dictionaries and pre-trained cross-lingual embeddings available for many language pairs, there is little research on how they perform on specialised tasks. In this paper, we investigate the best practices for constructing the seed dictionary for a specific domain. We evaluate the embeddings on the sequence labelling task of Curriculum Vitae parsing and show that the size of a bilingual dictionary, the frequency of the dictionary words in the domain corpora and the source of data (task-specific vs generic) influence the performance. We also show that the less training data is available in the low-resource language, the more the construction of the bilingual dictionary matters, and demonstrate that some of the choices are crucial in the zero-shot transfer learning case.

cs.CL↗

arXiv2019

Multi-Objective Pruning for CNNs Using Genetic Algorithm

In this work, we propose a heuristic genetic algorithm (GA) for pruning convolutional neural networks (CNNs) according to the multi-objective trade-off among error, computation and sparsity. In our experiments, we apply our approach to prune pre-trained LeNet across the MNIST dataset, which reduces 95.42% parameter size and achieves 16$\times$ speedups of convolutional layer computation with tiny accuracy loss by laying emphasis on sparsity and computation, respectively. Our empirical study suggests that GA is an alternative pruning approach for obtaining a competitive compression performance. Additionally, compared with state-of-the-art approaches, GA is capable of automatically pruning CNNs based on the multi-objective importance by a pre-defined fitness function.

cs.NE↗

arXiv2019

A polyhedron comparison theorem for 3-manifolds with positive scalar curvature

The study of comparison theorems in geometry has a rich history. In this paper, we establish a comparison theorem for polyhedra in 3-manifolds with nonnegative scalar curvature, answering affirmatively a dihedral rigidity conjecture by Gromov. For a large collections of polyhedra with interior non-negative scalar curvature and mean convex faces, we prove the dihedral angles along its edges cannot be everywhere less or equal than those of the corresponding Euclidean model, unless it is a isometric to a flat polyhedron.

math.DG↗

arXiv2019

Anomaly Detection for an E-commerce Pricing System

Online retailers execute a very large number of price updates when compared to brick-and-mortar stores. Even a few mis-priced items can have a significant business impact and result in a loss of customer trust. Early detection of anomalies in an automated real-time fashion is an important part of such a pricing system. In this paper, we describe unsupervised and supervised anomaly detection approaches we developed and deployed for a large-scale online pricing system at Walmart. Our system detects anomalies both in batch and real-time streaming settings, and the items flagged are reviewed and actioned based on priority and business impact. We found that having the right architecture design was critical to facilitate model performance at scale, and business impact and speed were important factors influencing model selection, parameter choice, and prioritization in a production environment for a large-scale system. We conducted analyses on the performance of various approaches on a test set using real-world retail data and fully deployed our approach into production. We found that our approach was able to detect the most important anomalies with high precision.

cs.LG↗

arXiv2019

Interaction between high intensity beam and structure resonances

With the help of the linearized perturbation theory, the collective beam instability due to the space charge, modelled by the Vlasov-Poisson equation, is well studied in the former research theoretically in a general sense [Chao Li and R. A. Jameson PRAB, 21, 024204 (2018)], where the related instability is depicted by the "structure resonance". As an extension of the theoretical study, this paper aims to verify the validity of the structure resonance with numerical simulations and to show the basic principles which interpret the interaction between structure resonance and beam in a general sense. It is found that the reaction of beam to the structure resonance ($nσ\approx 180^\circ$) differs for the cases with $n= 0$ and $n\ne 0$ stop bands. The non-resonant "plasma oscillation" is also revisited; the "phase space mixing", lying in the transient plasma oscillation, is the mechanism that ties the resonant effects, non-resonant effects and macroscopic beam phenomena together.

physics.acc-ph↗

arXiv2019

POG: Personalized Outfit Generation for Fashion Recommendation at Alibaba iFashion

Increasing demand for fashion recommendation raises a lot of challenges for online shopping platforms and fashion communities. In particular, there exist two requirements for fashion outfit recommendation: the Compatibility of the generated fashion outfits, and the Personalization in the recommendation process. In this paper, we demonstrate these two requirements can be satisfied via building a bridge between outfit generation and recommendation. Through large data analysis, we observe that people have similar tastes in individual items and outfits. Therefore, we propose a Personalized Outfit Generation (POG) model, which connects user preferences regarding individual items and outfits with Transformer architecture. Extensive offline and online experiments provide strong quantitative evidence that our method outperforms alternative methods regarding both compatibility and personalization metrics. Furthermore, we deploy POG on a platform named Dida in Alibaba to generate personalized outfits for the users of the online application iFashion. This work represents a first step towards an industrial-scale fashion outfit generation and recommendation solution, which goes beyond generating outfits based on explicit queries, or merely recommending from existing outfit pools. As part of this work, we release a large-scale dataset consisting of 1.01 million outfits with rich context information, and 0.28 billion user click actions from 3.57 million users. To the best of our knowledge, this dataset is the largest, publicly available, fashion related dataset, and the first to provide user behaviors relating to both outfits and fashion items.

cs.IR↗

arXiv2019

Adversarial Defense Through Network Profiling Based Path Extraction

Recently, researchers have started decomposing deep neural network models according to their semantics or functions. Recent work has shown the effectiveness of decomposed functional blocks for defending adversarial attacks, which add small input perturbation to the input image to fool the DNN models. This work proposes a profiling-based method to decompose the DNN models to different functional blocks, which lead to the effective path as a new approach to exploring DNNs' internal organization. Specifically, the per-image effective path can be aggregated to the class-level effective path, through which we observe that adversarial images activate effective path different from normal images. We propose an effective path similarity-based method to detect adversarial images with an interpretable model, which achieve better accuracy and broader applicability than the state-of-the-art technique.

cs.LG↗

arXiv2019

Adapting Image Super-Resolution State-of-the-arts and Learning Multi-model Ensemble for Video Super-Resolution

Recently, image super-resolution has been widely studied and achieved significant progress by leveraging the power of deep convolutional neural networks. However, there has been limited advancement in video super-resolution (VSR) due to the complex temporal patterns in videos. In this paper, we investigate how to adapt state-of-the-art methods of image super-resolution for video super-resolution. The proposed adapting method is straightforward. The information among successive frames is well exploited, while the overhead on the original image super-resolution method is negligible. Furthermore, we propose a learning-based method to ensemble the outputs from multiple super-resolution models. Our methods show superior performance and rank second in the NTIRE2019 Video Super-Resolution Challenge Track 1.

cs.CV↗

arXiv2019

CryptoNN: Training Neural Networks over Encrypted Data

Emerging neural networks based machine learning techniques such as deep learning and its variants have shown tremendous potential in many application domains. However, they raise serious privacy concerns due to the risk of leakage of highly privacy-sensitive data when data collected from users is used to train neural network models to support predictive tasks. To tackle such serious privacy concerns, several privacy-preserving approaches have been proposed in the literature that use either secure multi-party computation (SMC) or homomorphic encryption (HE) as the underlying mechanisms. However, neither of these cryptographic approaches provides an efficient solution towards constructing a privacy-preserving machine learning model, as well as supporting both the training and inference phases. To tackle the above issue, we propose a CryptoNN framework that supports training a neural network model over encrypted data by using the emerging functional encryption scheme instead of SMC or HE. We also construct a functional encryption scheme for basic arithmetic computation to support the requirement of the proposed CryptoNN framework. We present performance evaluation and security analysis of the underlying crypto scheme and show through our experiments that CryptoNN achieves accuracy that is similar to those of the baseline neural network models on the MNIST dataset.

cs.CR↗

arXiv2019

Multi-Interest Network with Dynamic Routing for Recommendation at Tmall

Industrial recommender systems usually consist of the matching stage and the ranking stage, in order to handle the billion-scale of users and items. The matching stage retrieves candidate items relevant to user interests, while the ranking stage sorts candidate items by user interests. Thus, the most critical ability is to model and represent user interests for either stage. Most of the existing deep learning-based models represent one user as a single vector which is insufficient to capture the varying nature of user's interests. In this paper, we approach this problem from a different view, to represent one user with multiple vectors encoding the different aspects of the user's interests. We propose the Multi-Interest Network with Dynamic routing (MIND) for dealing with user's diverse interests in the matching stage. Specifically, we design a multi-interest extractor layer based on capsule routing mechanism, which is applicable for clustering historical behaviors and extracting diverse interests. Furthermore, we develop a technique named label-aware attention to help learn a user representation with multiple vectors. Through extensive experiments on several public benchmarks and one large-scale industrial dataset from Tmall, we demonstrate that MIND can achieve superior performance than state-of-the-art methods for recommendation. Currently, MIND has been deployed for handling major online traffic at the homepage on Mobile Tmall App.

cs.IR↗

arXiv2019

Shared Predictive Cross-Modal Deep Quantization

With explosive growth of data volume and ever-increasing diversity of data modalities, cross-modal similarity search, which conducts nearest neighbor search across different modalities, has been attracting increasing interest. This paper presents a deep compact code learning solution for efficient cross-modal similarity search. Many recent studies have proven that quantization-based approaches perform generally better than hashing-based approaches on single-modal similarity search. In this paper, we propose a deep quantization approach, which is among the early attempts of leveraging deep neural networks into quantization-based cross-modal similarity search. Our approach, dubbed shared predictive deep quantization (SPDQ), explicitly formulates a shared subspace across different modalities and two private subspaces for individual modalities, and representations in the shared subspace and the private subspaces are learned simultaneously by embedding them to a reproducing kernel Hilbert space, where the mean embedding of different modality distributions can be explicitly compared. In addition, in the shared subspace, a quantizer is learned to produce the semantics preserving compact codes with the help of label alignment. Thanks to this novel network architecture in cooperation with supervised quantization training, SPDQ can preserve intramodal and intermodal similarities as much as possible and greatly reduce quantization error. Experiments on two popular benchmarks corroborate that our approach outperforms state-of-the-art methods.

cs.CV↗

arXiv2019

Active Multi-Kernel Domain Adaptation for Hyperspectral Image Classification

Recent years have witnessed the quick progress of the hyperspectral images (HSI) classification. Most of existing studies either heavily rely on the expensive label information using the supervised learning or can hardly exploit the discriminative information borrowed from related domains. To address this issues, in this paper we show a novel framework addressing HSI classification based on the domain adaptation (DA) with active learning (AL). The main idea of our method is to retrain the multi-kernel classifier by utilizing the available labeled samples from source domain, and adding minimum number of the most informative samples with active queries in the target domain. The proposed method adaptively combines multiple kernels, forming a DA classifier that minimizes the bias between the source and target domains. Further equipped with the nested actively updating process, it sequentially expands the training set and gradually converges to a satisfying level of classification performance. We study this active adaptation framework with the Margin Sampling (MS) strategy in the HSI classification task. Our experimental results on two popular HSI datasets demonstrate its effectiveness.

cs.CV↗

arXiv2019

Active Transfer Learning Network: A Unified Deep Joint Spectral-Spatial Feature Learning Model For Hyperspectral Image Classification

Deep learning has recently attracted significant attention in the field of hyperspectral images (HSIs) classification. However, the construction of an efficient deep neural network (DNN) mostly relies on a large number of labeled samples being available. To address this problem, this paper proposes a unified deep network, combined with active transfer learning that can be well-trained for HSIs classification using only minimally labeled training data. More specifically, deep joint spectral-spatial feature is first extracted through hierarchical stacked sparse autoencoder (SSAE) networks. Active transfer learning is then exploited to transfer the pre-trained SSAE network and the limited training samples from the source domain to the target domain, where the SSAE network is subsequently fine-tuned using the limited labeled samples selected from both source and target domain by corresponding active learning strategies. The advantages of our proposed method are threefold: 1) the network can be effectively trained using only limited labeled samples with the help of novel active learning strategies; 2) the network is flexible and scalable enough to function across various transfer situations, including cross-dataset and intra-image; 3) the learned deep joint spectral-spatial feature representation is more generic and robust than many joint spectral-spatial feature representation. Extensive comparative evaluations demonstrate that our proposed method significantly outperforms many state-of-the-art approaches, including both traditional and deep network-based methods, on three popular datasets.

cs.CV↗

arXiv2019

Non-local Meets Global: An Integrated Paradigm for Hyperspectral Denoising

Non-local low-rank tensor approximation has been developed as a state-of-the-art method for hyperspectral image (HSI) denoising. Unfortunately, with more spectral bands for HSI, while the running time of these methods significantly increases, their denoising performance benefits little. In this paper, we claim that the HSI underlines a global spectral low-rank subspace, and the spectral subspaces of each full band patch groups should underlie this global low-rank subspace. This motivates us to propose a unified spatial-spectral paradigm for HSI denoising. As the new model is hard to optimize, we further propose an efficient algorithm for optimization, which is motivated by alternating minimization. This is done by first learning a low-dimensional projection and the related reduced image from the noisy HSI. Then, the non-local low-rank denoising and iterative regularization are developed to refine the reduced image and projection, respectively. Finally, experiments on synthetic and both real datasets demonstrate the superiority against the other state-of-the-arts HSI denoising methods.

cs.CV↗