arXiv subjects

Bin Wang

Publications and source records attributed to Bin Wang.

At least 685 records · Page 38Linked to original sources

arXiv2021

Pre-training with Meta Learning for Chinese Word Segmentation

Recent researches show that pre-trained models (PTMs) are beneficial to Chinese Word Segmentation (CWS). However, PTMs used in previous works usually adopt language modeling as pre-training tasks, lacking task-specific prior segmentation knowledge and ignoring the discrepancy between pre-training tasks and downstream CWS tasks. In this paper, we propose a CWS-specific pre-trained model METASEG, which employs a unified architecture and incorporates meta learning algorithm into a multi-criteria pre-training task. Empirical results show that METASEG could utilize common prior segmentation knowledge from different existing criteria and alleviate the discrepancy between pre-trained models and downstream CWS tasks. Besides, METASEG can achieve new state-of-the-art performance on twelve widely-used CWS datasets and significantly improve model performance in low-resource settings.

cs.CL↗

arXiv2021

Progressive residual learning for single image dehazing

The recent physical model-free dehazing methods have achieved state-of-the-art performances. However, without the guidance of physical models, the performances degrade rapidly when applied to real scenarios due to the unavailable or insufficient data problems. On the other hand, the physical model-based methods have better interpretability but suffer from multi-objective optimizations of parameters, which may lead to sub-optimal dehazing results. In this paper, a progressive residual learning strategy has been proposed to combine the physical model-free dehazing process with reformulated scattering model-based dehazing operations, which enjoys the merits of dehazing methods in both categories. Specifically, the global atmosphere light and transmission maps are interactively optimized with the aid of accurate residual information and preliminary dehazed restorations from the initial physical model-free dehazing process. The proposed method performs favorably against the state-of-the-art methods on public dehazing benchmarks with better model interpretability and adaptivity for complex hazy data.

eess.IV↗

arXiv2021

Black holes of $(2+1)$-dimensional $f(R)$ gravity coupled to a scalar field

We consider $f(R)$ gravity theories in the presence of a scalar field minimally coupled to gravity with a self-interacting potential in $(2+1)$-dimensions. Without specifying the form of the $f(R)$ function, we first obtain an exact black hole solution dressed with scalar hair with the scalar charge to appear in the $f(R)$ function and we discuss its thermodynamics. This solution at large distances gives a hairy BTZ black hole, and it reduces to the BTZ black hole when the scalar field decouples. In a pure $f(R)$ gravity supported by the scalar field, we find an exact hairy black hole similar to the BTZ black hole with phantom hair and an analytic $f(R)$ form and discuss its thermodynamics.

gr-qc↗

arXiv2021

Towards Speeding up Adversarial Training in Latent Spaces

Adversarial training is wildly considered as one of the most effective way to defend against adversarial examples. However, existing adversarial training methods consume unbearable time, due to the fact that they need to generate adversarial examples in the large input space. To speed up adversarial training, we propose a novel adversarial training method that does not need to generate real adversarial examples. By adding perturbations to logits to generate Endogenous Adversarial Examples (EAEs) -- the adversarial examples in the latent space, the time consuming gradient calculation can be avoided. Extensive experiments are conducted on CIFAR-10 and ImageNet, and the results show that comparing to state-of-the-art methods, our EAE adversarial training not only shortens the training time, but also enhances the robustness of the model and has less impact on the accuracy of clean examples than the existing methods.

cs.LG↗

arXiv2021

The collider tests of a leptophilic scalar for the anomalous magnetic moments

We study the anomalous muon and electron magnetic moments by introducing a scalar with CP-violating Yukawa couplings to the lepton sector. By fitting these two magnetic moments with the recent experimental measurements, we find that such a leptophilic scalar in the mass range of $\mathcal{O}(10)- \mathcal{O}(1000 )\,\rm GeV$ can be a possible source for the current experimental deviations from the Standard Model (SM) predictions, with $\mathcal{O}(0.1) - \mathcal{O}(1)$ Yukawa couplings. The current electron and muon EDM constraints to the general CP-violating Yukawa couplings are discussed. We propose to search such a leptophilic scalar mediated at the future high-luminosity LHC (HL-LHC) runs, as well as the high-energy lepton colliders, including the CEPC and the muon collider. Our results show that the leptophilic scalar in the mass range of $\mathcal{O}(10)- \mathcal{O}(1000 )\,\rm GeV$ can be fully probed by the future experimental searches at the HL-LHC and the lepton colliders at their early stages.

hep-ph↗

arXiv2021

BLOCKEYE: Hunting For DeFi Attacks on Blockchain

Decentralized finance, i.e., DeFi, has become the most popular type of application on many public blockchains (e.g., Ethereum) in recent years. Compared to the traditional finance, DeFi allows customers to flexibly participate in diverse blockchain financial services (e.g., lending, borrowing, collateralizing, exchanging etc.) via smart contracts at a relatively low cost of trust. However, the open nature of DeFi inevitably introduces a large attack surface, which is a severe threat to the security of participants funds. In this paper, we proposed BLOCKEYE, a real-time attack detection system for DeFi projects on the Ethereum blockchain. Key capabilities provided by BLOCKEYE are twofold: (1) Potentially vulnerable DeFi projects are identified based on an automatic security analysis process, which performs symbolic reasoning on the data flow of important service states, e.g., asset price, and checks whether they can be externally manipulated. (2) Then, a transaction monitor is installed offchain for a vulnerable DeFi project. Transactions sent not only to that project but other associated projects as well are collected for further security analysis. A potential attack is flagged if a violation is detected on a critical invariant configured in BLOCKEYE, e.g., Benefit is achieved within a very short time and way much bigger than the cost. We applied BLOCKEYE in several popular DeFi projects and managed to discover potential security attacks that are unreported before. A video of BLOCKEYE is available at https://youtu.be/7DjsWBLdlQU.

cs.CR↗

arXiv2021

Study of form factors and branching ratios for $D\rightarrow S,Al\bar{ν_{l}}$ with light-cone sum rules

We systematically study the semileptonic decay process of $ D\rightarrow S,A l\bar{ν_{l}}(l=e,μ)$ by light-cone sum rules (LCSR) with chiral currents, calculate the form factors containing only the contribution of the leading twist light-cone distribution amplitudes (LCDAs). For scalar mesons $a_{0}(980)$ and $a_{0}(1450)$, we take them as $q\bar{q}$ states. For axial-vector meson, we study $a_{1}(1260)( 1^{3}p^{1})$ and $b_{1}(1235)( 1^{1}p^{1})$. Based on the results of these form factors, we further present the branching ratios of these semileptonic decay processes. The numerical results for $ D\rightarrow a_{0}(980), b_{1}(1235)l\bar{ν_{l}} $ are in good agreement with experiments and that for $ D\rightarrow a_{0}(1450)l\bar{ν_{l}}$ process are expected to be tested experimentally in the future.

hep-ph↗

arXiv2021

Task-adaptive Neural Process for User Cold-Start Recommendation

User cold-start recommendation is a long-standing challenge for recommender systems due to the fact that only a few interactions of cold-start users can be exploited. Recent studies seek to address this challenge from the perspective of meta learning, and most of them follow a manner of parameter initialization, where the model parameters can be learned by a few steps of gradient updates. While these gradient-based meta-learning models achieve promising performances to some extent, a fundamental problem of them is how to adapt the global knowledge learned from previous tasks for the recommendations of cold-start users more effectively. In this paper, we develop a novel meta-learning recommender called task-adaptive neural process (TaNP). TaNP is a new member of the neural process family, where making recommendations for each user is associated with a corresponding stochastic process. TaNP directly maps the observed interactions of each user to a predictive distribution, sidestepping some training issues in gradient-based meta-learning models. More importantly, to balance the trade-off between model capacity and adaptation reliability, we introduce a novel task-adaptive mechanism. It enables our model to learn the relevance of different tasks and customize the global knowledge to the task-related decoder parameters for estimating user preferences. We validate TaNP on multiple benchmark datasets in different experimental settings. Empirical results demonstrate that TaNP yields consistent improvements over several state-of-the-art meta-learning recommenders.

cs.IR↗

arXiv2021

Progressive Depth Learning for Single Image Dehazing

The formulation of the hazy image is mainly dominated by the reflected lights and ambient airlight. Existing dehazing methods often ignore the depth cues and fail in distant areas where heavier haze disturbs the visibility. However, we note that the guidance of the depth information for transmission estimation could remedy the decreased visibility as distances increase. In turn, the good transmission estimation could facilitate the depth estimation for hazy images. In this paper, a deep end-to-end model that iteratively estimates image depths and transmission maps is proposed to perform an effective depth prediction for hazy images and improve the dehazing performance with the guidance of depth information. The image depth and transmission map are progressively refined to better restore the dehazed image. Our approach benefits from explicitly modeling the inner relationship of image depth and transmission map, which is especially effective for distant hazy areas. Extensive results on the benchmarks demonstrate that our proposed network performs favorably against the state-of-the-art dehazing methods in terms of depth estimation and haze removal.

cs.CV↗

arXiv2021

Inductive Learning on Commonsense Knowledge Graph Completion

Commonsense knowledge graph (CKG) is a special type of knowledge graph (KG), where entities are composed of free-form text. However, most existing CKG completion methods focus on the setting where all the entities are presented at training time. Although this setting is standard for conventional KG completion, it has limitations for CKG completion. At test time, entities in CKGs can be unseen because they may have unseen text/names and entities may be disconnected from the training graph, since CKGs are generally very sparse. Here, we propose to study the inductive learning setting for CKG completion where unseen entities may present at test time. We develop a novel learning framework named InductivE. Different from previous approaches, InductiveE ensures the inductive learning capability by directly computing entity embeddings from raw entity attributes/text. InductiveE consists of a free-text encoder, a graph encoder, and a KG completion decoder. Specifically, the free-text encoder first extracts the textual representation of each entity based on the pre-trained language model and word embedding. The graph encoder is a gated relational graph convolutional neural network that learns from a densified graph for more informative entity representation learning. We develop a method that densifies CKGs by adding edges among semantic-related entities and provide more supportive information for unseen entities, leading to better generalization ability of entity embedding for unseen entities. Finally, inductiveE employs Conv-TransE as the CKG completion decoder. Experimental results show that InductiveE significantly outperforms state-of-the-art baselines in both standard and inductive settings on ATOMIC and ConceptNet benchmarks. InductivE performs especially well on inductive scenarios where it achieves above 48% improvement over present methods.

cs.AI↗

arXiv2021

Train a One-Million-Way Instance Classifier for Unsupervised Visual Representation Learning

This paper presents a simple unsupervised visual representation learning method with a pretext task of discriminating all images in a dataset using a parametric, instance-level classifier. The overall framework is a replica of a supervised classification model, where semantic classes (e.g., dog, bird, and ship) are replaced by instance IDs. However, scaling up the classification task from thousands of semantic labels to millions of instance labels brings specific challenges including 1) the large-scale softmax computation; 2) the slow convergence due to the infrequent visiting of instance samples; and 3) the massive number of negative classes that can be noisy. This work presents several novel techniques to handle these difficulties. First, we introduce a hybrid parallel training framework to make large-scale training feasible. Second, we present a raw-feature initialization mechanism for classification weights, which we assume offers a contrastive prior for instance discrimination and can clearly speed up converge in our experiments. Finally, we propose to smooth the labels of a few hardest classes to avoid optimizing over very similar negative pairs. While being conceptually simple, our framework achieves competitive or superior performance compared to state-of-the-art unsupervised approaches, i.e., SimCLR, MoCoV2, and PIC under ImageNet linear evaluation protocol and on several downstream visual tasks, verifying that full instance classification is a strong pretraining technique for many semantic visual tasks.

cs.CV↗

arXiv2021

Structure-preserving algorithms with uniform error bound and long-time energy conservation for highly oscillatory Hamiltonian systems

Structure-preserving algorithms and algorithms with uniform error bound have constituted two interesting classes of numerical methods. In this paper, we blend these two kinds of methods for solving nonlinear Hamiltonian systems with highly oscillatory solution, and the blended algorithms inherit and respect the advantage of each method. Two kinds of algorithms are presented to preserve the symplecticity and energy of the Hamiltonian systems, respectively. Moreover, the proposed algorithms are shown to have uniform error bound for the highly oscillatory structure. A numerical experiment is carried out to support the theoretical results established in this paper by showing the performance of the blended algorithms.

math.NA↗

arXiv2021

Optimal convergence and long-time conservation of exponential integration for Schrödinger equations in a normal or highly oscillatory regime

In this paper, we formulate and analyse exponential integrations when applied to nonlinear Schrödinger equations in a normal or highly oscillatory regime. A kind of exponential integrators with energy preservation, optimal convergence and long time near conservations of actions, momentum and density will be formulated and analysed. To this end, we derive continuous-stage exponential integrators and show that the integrators can exactly preserve the energy of Hamiltonian systems. Three practical energy-preserving integrators are presented. It is shown that these integrators exhibit optimal convergence and have near conservations of actions, momentum and density over long times. A numerical experiment is carried out to support all the theoretical results presented in this paper. Some applications of the integrators to other kinds of ordinary/partial differential equations are also presented.

math.NA↗

arXiv2021

On the Strong Equivalences for LPMLN Programs

LPMLN is a powerful knowledge representation and reasoning tool that combines the non-monotonic reasoning ability of Answer Set Programming (ASP) and the probabilistic reasoning ability of Markov Logic Networks (MLN). In this paper, we study the strong equivalence for LPMLN programs, which is an important tool for program rewriting and theoretical investigations in the field of logic programming. First of all, we present the notion of p-strong equivalence for LPMLN and present a model-theoretical characterization for the notion. And we investigate the relationships among the p-strong equivalence and other existing notions of strong equivalences for LPMLN. Then, we investigate several properties of the p-strong equivalence from the following four aspects. Firstly, we investigate two relaxed notions of the p-strong equivalence according to practical scenarios of program rewriting, and present corresponding characterizations for the notions. Secondly, we analyze the computational complexities of deciding strong equivalences for LPMLN programs. Thirdly, we investigate the relationships among the strong equivalences of LPMLN and two extensions of ASP: ASP with weak constraints and ordered disjunctions. Finally, we investigate LPMLN program simplification via the p-strong equivalence and present some syntactic conditions that decide the p-strong equivalence between a single LPMLN rule and the empty program. The contributions of the paper are as follows. Firstly, all of the results presented in this paper provide a better understanding of LPMLN programming, which helps us further explore the properties of LPMLN. Secondly, the relationships among the strong equivalences open a way to study the strong equivalences for some logic formalisms by translating into LPMLN. Thirdly, the program simplification can be used to enhance the implementations of the LPMLN solvers ...

cs.LO↗

arXiv2021

Boosting performance for software defined networks from traffic engineering perspective

Paths selection algorithms and rate adaptation objective functions are usually studied separately. In contrast, this paper evaluates some traffic engineering (TE) systems for software defined networking obtained by combining path selection techniques with average delay and load balancing, the two most popular TE objective functions. Based on TE simulation results, the best TE system suitable for software defined networks is a system where the paths are calculated using an oblivious routing model and its adaptation rate calculated using an average delay objective function. Thus, we propose the RACKE+AD system combining path sets computed using Racke's oblivious routing and traffic splitting objective function using average delay. This model outperforms current state-of-the-art models, maximizes throughput, achieves better network resource utilization, and minimizes delay. The proposed system outperformed SMORE and SWAN by 4.2% and 9.6% respectively, achieving 27% better utilization and delivering 34% more traffic with 50% less latency compared with both systems on a GEANT network.

cs.NI↗

arXiv2021

Thermodynamics of 2+1 dimensional Coulomb-Like Black Holes from Non Linear Electrodynamics with a traceless energy momentum tensor

In this work we study thermodynamics of 2+1-dimensional static black holes with a nonlinear electric field. Besides employing the standard thermodynamic approach, we investigate the black hole thermodynamics by studying its thermodynamic geometry. We compute the Weinhold and Ruppeiner metrics and compare the thermodynamic geometry with the standard description on the black hole thermodynamics. We further consider the cosmological constant as an additional extensive thermodynamic variable. In the thermodynamic equilibrium three dimensional space, we compute the efficiency of the heat engine and show that it is possible to be built with this black hole.

gr-qc↗

arXiv2021

GraphHop: An Enhanced Label Propagation Method for Node Classification

A scalable semi-supervised node classification method on graph-structured data, called GraphHop, is proposed in this work. The graph contains attributes of all nodes but labels of a few nodes. The classical label propagation (LP) method and the emerging graph convolutional network (GCN) are two popular semi-supervised solutions to this problem. The LP method is not effective in modeling node attributes and labels jointly or facing a slow convergence rate on large-scale graphs. GraphHop is proposed to its shortcoming. With proper initial label vector embeddings, each iteration of GraphHop contains two steps: 1) label aggregation and 2) label update. In Step 1, each node aggregates its neighbors' label vectors obtained in the previous iteration. In Step 2, a new label vector is predicted for each node based on the label of the node itself and the aggregated label information obtained in Step 1. This iterative procedure exploits the neighborhood information and enables GraphHop to perform well in an extremely small label rate setting and scale well for very large graphs. Experimental results show that GraphHop outperforms state-of-the-art graph learning methods on a wide range of tasks (e.g., multi-label and multi-class classification on citation networks, social graphs, and commodity consumption graphs) in graphs of various sizes. Our codes are publicly available on GitHub (https://github.com/TianXieUSC/GraphHop).

cs.LG↗

arXiv2020

Phase transitions in four-dimensional AdS black holes with a nonlinear electrodynamics source

In this work we consider black hole solutions to Einstein theory coupled to a nonlinear power-law electromagnetic field with a fixed exponent value. We study the extended phase space thermodynamics in canonical and grand canonical ensembles where the varying cosmological constant plays the role of an effective thermodynamic pressure. We examine thermodynamical phase transitions in such black hols and find that both first and second order phase transitions can occur in the canonical ensemble, while for the grand canonical ensemble the Hawking-Page and second order phase transitions are allowed.

gr-qc↗