arXiv subjects

Jianmin Wang

Publications and source records attributed to Jianmin Wang.

At least 145 records · Page 8Linked to original sources

arXiv2018

Partial Adversarial Domain Adaptation

Domain adversarial learning aligns the feature distributions across the source and target domains in a two-player minimax game. Existing domain adversarial networks generally assume identical label space across different domains. In the presence of big data, there is strong motivation of transferring deep models from existing big domains to unknown small domains. This paper introduces partial domain adaptation as a new domain adaptation scenario, which relaxes the fully shared label space assumption to that the source label space subsumes the target label space. Previous methods typically match the whole source domain to the target domain, which are vulnerable to negative transfer for the partial domain adaptation problem due to the large mismatch between label spaces. We present Partial Adversarial Domain Adaptation (PADA), which simultaneously alleviates negative transfer by down-weighing the data of outlier source classes for training both source classifier and domain adversary, and promotes positive transfer by matching the feature distributions in the shared label space. Experiments show that PADA exceeds state-of-the-art results for partial domain adaptation tasks on several datasets.

cs.CV↗

arXiv2018

Matching Consecutive Subpatterns Over Streaming Time Series

Pattern matching of streaming time series with lower latency under limited computing resource comes to a critical problem, especially as the growth of Industry 4.0 and Industry Internet of Things. However, against traditional single pattern matching model, a pattern may contain multiple subpatterns representing different physical meanings in the real world. Hence, we formulate a new problem, called "consecutive subpatterns matching", which allows users to specify a pattern containing several consecutive subpatterns with various specified thresholds. We propose a novel representation Equal-Length Block (ELB) together with two efficient implementations, which work very well under all Lp-Norms without false dismissals. Extensive experiments are performed on synthetic and real-world datasets to illustrate that our approach outperforms the brute-force method and MSM, a multi-step filter mechanism over the multi-scaled representation by orders of magnitude.

cs.DB↗

arXiv2017

Transfer Adversarial Hashing for Hamming Space Retrieval

Hashing is widely applied to large-scale image retrieval due to the storage and retrieval efficiency. Existing work on deep hashing assumes that the database in the target domain is identically distributed with the training set in the source domain. This paper relaxes this assumption to a transfer retrieval setting, which allows the database and the training set to come from different but relevant domains. However, the transfer retrieval setting will introduce two technical difficulties: first, the hash model trained on the source domain cannot work well on the target domain due to the large distribution gap; second, the domain gap makes it difficult to concentrate the database points to be within a small Hamming ball. As a consequence, transfer retrieval performance within Hamming Radius 2 degrades significantly in existing hashing methods. This paper presents Transfer Adversarial Hashing (TAH), a new hybrid deep architecture that incorporates a pairwise $t$-distribution cross-entropy loss to learn concentrated hash codes and an adversarial network to align the data distributions between the source and target domains. TAH can generate compact transfer hash codes for efficient image retrieval on both source and target domains. Comprehensive experiments validate that TAH yields state of the art Hamming space retrieval performance on standard datasets.

cs.CV↗

arXiv2017

Fine-grained Pattern Matching Over Streaming Time Series

Pattern matching of streaming time series with lower latency under limited computing resource comes to a critical problem, especially as the growth of Industry 4.0 and Industry Internet of Things. However, against traditional single pattern matching problem, a pattern may contain multiple segments representing different statistical properties or physical meanings for more precise and expressive matching in real world. Hence, we formulate a new problem, called "fine-grained pattern matching", which allows users to specify varied granularities of matching deviation to different segments of a given pattern, and fuzzy regions for adaptive breakpoints determination between consecutive segments. In this paper, we propose a novel two-phase approach. In the pruning phase, we introduce Equal-Length Block (ELB) representation together with Block-Skipping Pruning (BSP) policy, which guarantees low cost feature calculation, effective pruning and no false dismissals. In the post-processing phase, a delta-function is proposed to enable us to conduct exact matching in linear complexity. Extensive experiments are conducted to evaluate on synthetic and real-world datasets, which illustrates that our algorithm outperforms the brute-force method and MSM, a multi-step filter mechanism over the multi-scaled representation.

cs.CV↗

arXiv2017

Learning Multiple Tasks with Multilinear Relationship Networks

Deep networks trained on large-scale data can learn transferable features to promote learning multiple tasks. Since deep features eventually transition from general to specific along deep networks, a fundamental problem of multi-task learning is how to exploit the task relatedness underlying parameter tensors and improve feature transferability in the multiple task-specific layers. This paper presents Multilinear Relationship Networks (MRN) that discover the task relationships based on novel tensor normal priors over parameter tensors of multiple task-specific layers in deep convolutional networks. By jointly learning transferable features and multilinear relationships of tasks and features, MRN is able to alleviate the dilemma of negative-transfer in the feature layers and under-transfer in the classifier layer. Experiments show that MRN yields state-of-the-art results on three multi-task learning datasets.

cs.LG↗

arXiv2017

Deep Transfer Learning with Joint Adaptation Networks

Deep networks have been successfully applied to learn transferable features for adapting models from a source domain to a different target domain. In this paper, we present joint adaptation networks (JAN), which learn a transfer network by aligning the joint distributions of multiple domain-specific layers across domains based on a joint maximum mean discrepancy (JMMD) criterion. Adversarial training strategy is adopted to maximize JMMD such that the distributions of the source and target domains are made more distinguishable. Learning can be performed by stochastic gradient descent with the gradients computed by back-propagation in linear-time. Experiments testify that our model yields state of the art results on standard datasets.

cs.LG↗

arXiv2017

HashNet: Deep Learning to Hash by Continuation

Learning to hash has been widely applied to approximate nearest neighbor search for large-scale multimedia retrieval, due to its computation efficiency and retrieval quality. Deep learning to hash, which improves retrieval quality by end-to-end representation learning and hash encoding, has received increasing attention recently. Subject to the ill-posed gradient difficulty in the optimization with sign activations, existing deep learning to hash methods need to first learn continuous representations and then generate binary hash codes in a separated binarization step, which suffer from substantial loss of retrieval quality. This work presents HashNet, a novel deep architecture for deep learning to hash by continuation method with convergence guarantees, which learns exactly binary hash codes from imbalanced similarity data. The key idea is to attack the ill-posed gradient problem in optimizing deep networks with non-smooth binary activations by continuation method, in which we begin from learning an easier network with smoothed activation function and let it evolve during the training, until it eventually goes back to being the original, difficult to optimize, deep network with the sign activation function. Comprehensive empirical evidence shows that HashNet can generate exactly binary hash codes and yield state-of-the-art multimedia retrieval performance on standard benchmarks.

cs.LG↗

arXiv2017

Partial Transfer Learning with Selective Adversarial Networks

Adversarial learning has been successfully embedded into deep networks to learn transferable features, which reduce distribution discrepancy between the source and target domains. Existing domain adversarial networks assume fully shared label space across domains. In the presence of big data, there is strong motivation of transferring both classification and representation models from existing big domains to unknown small domains. This paper introduces partial transfer learning, which relaxes the shared label space assumption to that the target label space is only a subspace of the source label space. Previous methods typically match the whole source domain to the target domain, which are prone to negative transfer for the partial transfer problem. We present Selective Adversarial Network (SAN), which simultaneously circumvents negative transfer by selecting out the outlier source classes and promotes positive transfer by maximally matching the data distributions in the shared label space. Experiments demonstrate that our models exceed state-of-the-art results for partial transfer learning tasks on several benchmark datasets.

cs.LG↗

arXiv2017

Correlation Hashing Network for Efficient Cross-Modal Retrieval

Hashing is widely applied to approximate nearest neighbor search for large-scale multimodal retrieval with storage and computation efficiency. Cross-modal hashing improves the quality of hash coding by exploiting semantic correlations across different modalities. Existing cross-modal hashing methods first transform data into low-dimensional feature vectors, and then generate binary codes by another separate quantization step. However, suboptimal hash codes may be generated since the quantization error is not explicitly minimized and the feature representation is not jointly optimized with the binary codes. This paper presents a Correlation Hashing Network (CHN) approach to cross-modal hashing, which jointly learns good data representation tailored to hash coding and formally controls the quantization error. The proposed CHN is a hybrid deep architecture that constitutes a convolutional neural network for learning good image representations, a multilayer perception for learning good text representations, two hashing layers for generating compact binary codes, and a structured max-margin loss that integrates all things together to enable learning similarity-preserving and high-quality hash codes. Extensive empirical study shows that CHN yields state of the art cross-modal retrieval performance on standard benchmarks.

cs.CV↗

arXiv2017

Unsupervised Domain Adaptation with Residual Transfer Networks

The recent success of deep neural networks relies on massive amounts of labeled data. For a target task where labeled data is unavailable, domain adaptation can transfer a learner from a different source domain. In this paper, we propose a new approach to domain adaptation in deep networks that can jointly learn adaptive classifiers and transferable features from labeled data in the source domain and unlabeled data in the target domain. We relax a shared-classifier assumption made by previous methods and assume that the source classifier and target classifier differ by a residual function. We enable classifier adaptation by plugging several layers into deep network to explicitly learn the residual function with reference to the target classifier. We fuse features of multiple layers with tensor product and embed them into reproducing kernel Hilbert spaces to match distributions for feature adaptation. The adaptation can be achieved in most feed-forward models by extending them with new residual layers and loss functions, which can be trained efficiently via back-propagation. Empirical evidence shows that the new approach outperforms state of the art methods on standard domain adaptation benchmarks.

cs.LG↗

arXiv2016

Composite Correlation Quantization for Efficient Multimodal Retrieval

Efficient similarity retrieval from large-scale multimodal database is pervasive in modern search engines and social networks. To support queries across content modalities, the system should enable cross-modal correlation and computation-efficient indexing. While hashing methods have shown great potential in achieving this goal, current attempts generally fail to learn isomorphic hash codes in a seamless scheme, that is, they embed multiple modalities in a continuous isomorphic space and separately threshold embeddings into binary codes, which incurs substantial loss of retrieval accuracy. In this paper, we approach seamless multimodal hashing by proposing a novel Composite Correlation Quantization (CCQ) model. Specifically, CCQ jointly finds correlation-maximal mappings that transform different modalities into isomorphic latent space, and learns composite quantizers that convert the isomorphic latent features into compact binary codes. An optimization framework is devised to preserve both intra-modal similarity and inter-modal correlation through minimizing both reconstruction and quantization errors, which can be trained from both paired and partially paired data in linear time. A comprehensive set of experiments clearly show the superior effectiveness and efficiency of CCQ against the state of the art hashing methods for both unimodal and cross-modal retrieval.

cs.IR↗

arXiv2015

Learning Transferable Features with Deep Adaptation Networks

Recent studies reveal that a deep neural network can learn transferable features which generalize well to novel tasks for domain adaptation. However, as deep features eventually transition from general to specific along the network, the feature transferability drops significantly in higher layers with increasing domain discrepancy. Hence, it is important to formally reduce the dataset bias and enhance the transferability in task-specific layers. In this paper, we propose a new Deep Adaptation Network (DAN) architecture, which generalizes deep convolutional neural network to the domain adaptation scenario. In DAN, hidden representations of all task-specific layers are embedded in a reproducing kernel Hilbert space where the mean embeddings of different domain distributions can be explicitly matched. The domain discrepancy is further reduced using an optimal multi-kernel selection method for mean embedding matching. DAN can learn transferable features with statistical guarantees, and can scale linearly by unbiased estimate of kernel embedding. Extensive empirical evidence shows that the proposed architecture yields state-of-the-art image classification error rates on standard domain adaptation benchmarks.

cs.LG↗

arXiv2014

A simple model clarifies the complicated relationships of complex networks

Real-world networks such as the Internet and WWW have many common traits. Until now, hundreds of models were proposed to characterize these traits for understanding the networks. Because different models used very different mechanisms, it is widely believed that these traits origin from different causes. However, we find that a simple model based on optimisation can produce many traits, including scale-free, small-world, ultra small-world, Delta-distribution, compact, fractal, regular and random networks. Moreover, by revising the proposed model, the community-structure networks are generated. By this model and the revised versions, the complicated relationships of complex networks are illustrated. The model brings a new universal perspective to the understanding of complex networks and provide a universal method to model complex networks from the viewpoint of optimisation.

cs.SI↗

arXiv2013

INTEGRAL & RXTE View of Gamma-ray Binaries

Gamma-ray binaries are X-ray binaries with gamma-ray emissions. Their multi-wavelength emissions range from radio, optical, X-ray and to very high energy (TeV). X-ray emissions are crucial to understand the nature of gamma-ray binaries. INTEGRAL and RXTE have covered and monitored most of the gamma-ray binaries in hard and soft X-rays. Here we report the results of several gamma-ray binaries and possible gamma-ray binaries from INTEGRAL and RXTE.

astro-ph.HE↗

arXiv2012

Latency Bounding by Trading off Consistency in NoSQL Store: A Staging and Stepwise Approach

Latency is a key service factor for user satisfaction. Consistency is in a trade-off relation with operation latency in the distributed and replicated scenario. Existing NoSQL stores guarantee either strong or weak consistencies but none provides the best consistency based on the response latency. In this paper, we introduce dConssandra, a NoSQL store enabling users to specify latency bounds for data access operations. dConssandra dynamically bounds data access latency by trading off replica consistency. dConssandra is based on Cassandra. In comparison to Cassandra's implementation, dConssandra has a staged replication strategy enabling synchronous or asynchronous replication on demand. The main idea to bound latency by trading off consistency is to decompose the replication process into minute steps and bound latency by executing only a subset of these steps. dConssandra also implements a different in-memory storage architecture to support the above features. Experimental results for dConssandra over an actual cluster demonstrate that (1) the actual response latency is bounded by the given latency constraint; (2) greater write latency bounds lead to a lower latency in reading the latest value; and, (3) greater read latency bounds lead to the return of more recently written values.

cs.DB↗

arXiv2012

Ranking the Importance of Nodes of Complex Networks by the Equivalence Classes Approach

Identifying the importance of nodes of complex networks is of interest to the research of Social Networks, Biological Networks etc.. Current researchers have proposed several measures or algorithms, such as betweenness, PageRank and HITS etc., to identify the node importance. However, these measures are based on different aspects of properties of nodes, and often conflict with the others. A reasonable, fair standard is needed for evaluating and comparing these algorithms. This paper develops a framework as the standard for ranking the importance of nodes. Four intuitive rules are suggested to measure the node importance, and the equivalence classes approach is employed to resolve the conflicts and aggregate the results of the rules. To quantitatively compare the algorithms, the performance indicators are also proposed based on a similarity measure. Three widely used real-world networks are used as the test-beds. The experimental results illustrate the feasibility of this framework and show that both algorithms, PageRank and HITS, perform well with bias when dealing with the tested networks. Furthermore, this paper uses the proposed approach to analyze the structure of the Internet, and draws out the kernel of the Internet with dense links.

cs.SI↗

arXiv2012

Unveiling the super-orbital modulation of LSI +61 303 in X-rays

We found evidence for the super-orbital modulation in the X-ray emission of LS I +61 303 from the longest monitoring date by the RXTE. The time evolution of the modulated fraction in the orbital light curves can be well fitted with a sinusoidal function having a super-orbital period of 1667 days. However, we have found a 281.8+/-44.6 day shift between the super-orbital variability found at radio frequencies and our X-ray data. We also find a super-orbital modulation in the maximum count rate of the orbital light curves, compatible with the former results, including the shift.

astro-ph.HE↗

arXiv2012

Equalitarian Societies are Economically Impossible

The inequality of wealth distribution is a universal phenomenon in the civilized nations, and it is often imputed to the Matthew effect, that is, the rich get richer and the poor get poorer. Some philosophers unjustified this phenomenon and tried to put the human civilization upon the evenness of wealth. Noticing the facts that 1) the emergence of the centralism is the starting point of human civilization, i.e., people in a society were organized hierarchically, 2) the inequality of wealth emerges simultaneously, this paper proposes a wealth distribution model based on the hidden tree structure from the viewpoint of complex network. This model considers the organized structure of people in a society as a hidden tree, and the cooperations among human beings as the transactions on the hidden tree, thereby explains the distribution of wealth. This model shows that the scale-free phenomenon of wealth distribution can be produced by the cascade controlling of human society, that is, the inequality of wealth can parasitize in the social organizations, such that any actions in eliminating the unequal wealth distribution would lead to the destroy of social or economic structures, resulting in the collapse of the economic system, therefore, would fail in vain.

physics.soc-ph↗