The dominant neural machine translation (NMT) models apply unified attentional encoder-decoder neural networks for translation. Traditionally, the NMT decoders adopt recurrent neural networks (RNNs) to perform translation in a left-to-right manner, leaving the target-side contexts generated from right to left unexploited during translation. In this paper, we equip the conventional attentional encoder-decoder NMT framework with a backward decoder, in order to explore bidirectional decoding for NMT. Attending to the hidden state sequence produced by the encoder, our backward decoder first learns to generate the target-side hidden state sequence from right to left. Then, the forward decoder performs translation in the forward direction, while in each translation prediction timestep, it simultaneously applies two attention models to consider the source-side and reverse target-side hidden states, respectively. With this new architecture, our model is able to fully exploit source- and target-side contexts to improve translation quality altogether. Experimental results on NIST Chinese-English and WMT English-German translation tasks demonstrate that our model achieves substantial improvements over the conventional NMT by 3.14 and 1.38 BLEU points, respectively. The source code of this work can be obtained from https://github.com/DeepLearnXMU/ABDNMT.

Neural machine translation (NMT) heavily relies on word-level modelling to learn semantic representations of input sentences.However, for languages without natural word delimiters (e.g., Chinese) where input sentences have to be tokenized first,conventional NMT is confronted with two issues:1) it is difficult to find an optimal tokenization granularity for source sentence modelling, and2) errors in 1-best tokenizations may propagate to the encoder of NMT.To handle these issues, we propose word-lattice based Recurrent Neural Network (RNN) encoders for NMT,which generalize the standard RNN to word lattice topology.The proposed encoders take as input a word lattice that compactly encodes multiple tokenizations, and learn to generate new hidden states from arbitrarily many inputs and hidden states in preceding time steps.As such, the word-lattice based encoders not only alleviate the negative impact of tokenization errors but also are more expressive and flexible to embed input sentences.Experiment results on Chinese-English translation demonstrate the superiorities of the proposed encoders over the conventional encoder.

Recent years have witnessed an extensive popularity of convolutional neural networks (CNNs) in various computer vision and artificial intelligence applications. However, the performance gains have come at a cost of substantially intensive computation complexity, which prohibits its usage inresource-limited applications like mobile or embedded devices. While increasing attention has been paid to the acceleration of internal network structure, the redundancy of visual input is rarely considered. In this paper, we make the first attempt of reducing spatial and channel redundancy directly from the visual input for CNNs acceleration. The proposed method, termed ESPACE (Elimination of SPAtial and Channel rEdundancy), works by the following three steps: First, the 3D channel redundancy of convolutional layers is reduced by a set of low-rank approximation of convolutional filters. Second, a novel mask based selective processing scheme is proposed, which further speedups the convolution operations via skipping unsalient spatial locations of the visual input. Third, the accelerated network is fine-tuned using the training data via back-propagation. The proposed method is evaluated on ImageNet 2012 with implementations on two widely adopted CNNs, i.e. AlexNet and GoogLeNet. In comparison to several recent methods of CNN acceleration, the proposed scheme has demonstrated new state-of-the-art acceleration performance by a factor of 5.48* and 4.12* speedup on AlexNet and GoogLeNet, respectively, with a minimal decrease in classification accuracy.

Recent years have witnessed extensive attention in binary code learning, a.k.a. hashing, for nearest neighbor search problems. It has been seen that high-dimensional data points can quantize into binary codes to give an efficient similarity approximation via Hamming distance. Among the existing schemes, ranking-based hashing is recent promising that targets at preserving ordinal relations of ranking in the Hamming space to minimize retrieval loss. However, the size of the ranking tuples that show the ordinal relations, is quadratic or cubic to the size of training samples. It is so very expensive to embed such ranking tuples in binary code learning, especially given a large-scale training data set. Besides, it remains difficult to build ranking tuples efficiently for most ranking-preserving hashing, which are deployed over an ordinal graph-based setting. To handle these problems, we propose a novel ranking-preserving hashing method, dubbed Ordinal Constraint Hashing (OCH), which efficiently learns the optimal hashing functions with a graph-based approximation to embed the ordinal relations. The core idea is to reduce the size of ordinal graph with ordinal constraint projection, which preserves the ordinal relations through a small data set (such as clusters or random samples). In particular, to learn such hash functions effectively, we further relax the discrete constraints and design a specific stochastic gradient decent algorithm for optimization. Experimental results on three large-scale visual search benchmark datasets, i.e. LabelMe, Tiny100K and GIST1M, show that the proposed OCH method can achieve superior performance over the state-of-the-arts approaches.

Vehicle detection in satellite image has attracted extensive research attentions with various emerging applications.However, the detector performance has been significantly degenerated due to the low resolutions of satellite images, as well as the limited training data.In this paper, a robust domain-adaptive vehicle detection framework is proposed to bypass both problems.Our innovation is to transfer the detector learning to the high-resolution aerial image domain,where rich supervision exists and robust detectors can be trained.To this end, we first propose a super-resolution algorithm using coupled dictionary learning to ``augment'' the satellite image region being tested into the aerial domain.Notably, linear detection loss is embedded into the dictionary learning, which enforces the augmented region to be sensitive to the subsequent detector training.Second, to cope with the domain changes, we propose an instance-wised detection using Exemplar Support Vector Machines (E-SVMs), which well handles the intra-class and imaging variations like scales, rotations, and occlusions.With comprehensive experiments on large-scale satellite image collections, we demonstrate that the proposed framework can significantly boost the detection accuracy over several state-of-the-arts.

Binary code learning, a.k.a., hashing, has been recently popular due to its high efficiency in large-scale similarity search and recognition. It typically maps high-dimensional data points to binary codes, where data similarity can be efficiently computed via rapid Hamming distance. Most existing unsupervised hashing schemes pursue binary codes by reducing the quantization error from an original real-valued data space to a resulting Hamming space. On the other hand, most existing supervised hashing schemes constrain binary code learning to correlate with pairwise similarity labels. However, few methods consider ordinal relations in the binary code learning process, which serve as a very significant cue to learn the optimal binary codes for similarity search. In this paper, we propose a novel hashing scheme, dubbed Ordinal Embedding Hashing (OEH), which embeds given ordinal relations among data points to learn the ranking-preserving binary codes. The core idea is to construct a directed unweighted graph to capture the ordinal relations, and then train the hash functions using this ordinal graph to preserve the permutation relations in the Hamming space. To learn such hash functions effectively, we further relax the discrete constraints and design a stochastic gradient decent algorithm to obtain the optimal solution. Experimental results on two large-scale benchmark datasets demonstrate that the proposed OEH method can achieve superior performance over the state-of-the-arts approaches.At last, the evaluation on query by humming dataset demonstrates the OEH also has good performance for music retrieval by using user's humming or singing.

Metric learning has become a widespreadly used tool in machine learning. To reduce expensive costs brought in by increasing dimensionality, low-rank metric learning arises as it can be more economical in storage and computation. However, existing low-rank metric learning algorithms usually adopt nonconvex objectives, and are hence sensitive to the choice of a heuristic low-rank basis. In this paper, we propose a novel low-rank metric learning algorithm to yield bilinear similarity functions. This algorithm scales linearly with input dimensionality in both space and time, therefore applicable to high-dimensional data domains. A convex objective free of heuristics is formulated by leveraging trace norm regularization to promote low-rankness. Crucially, we prove that all globally optimal metric solutions must retain a certain low-rank structure, which enables our algorithm to decompose the high-dimensional learning task into two steps: an SVD-based projection and a metric learning problem with reduced dimensionality. The latter step can be tackled efficiently through employing a linearized Alternating Direction Method of Multipliers. The efficacy of the proposed algorithm is demonstrated through experiments performed on four benchmark datasets with tens of thousands of dimensions.

Salient object detection provides an alternative solution to various image semantic understanding tasks such as object recognition, adaptive compression and image retrieval. Recently, low-rank matrix recovery (LR) theory has been introduced into saliency detection, and achieves impressed results. However, the existing LR-based models neglect the underlying structure of images, and inevitably degrade the associated performance. In this paper, we propose a Low-rank and Structured sparse Matrix Decomposition (LSMD) model for salient object detection. In the model, a tree-structured sparsity-inducing norm regularization is firstly introduced to provide a hierarchical description of the image structure to ensure the completeness of the extracted salient object. The similarity of saliency values within the salient object is then guaranteed by the $\ell _\infty$-norm. Finally, high-level priors are integrated to guide the matrix decomposition and enhance the saliency detection. Experimental results on the largest public benchmark database show that our model outperforms existing LR-based approaches and other state-of-the-art methods, which verifies the effectiveness and robustness of the structure cues in our model.

Coming with the ever growing computational power of mobile devices, mobile visual search have undergone an evolution in techniques and applications. A significant trend is low bit rate visual search, where compact visual descriptors are extracted directly over a mobile and delivered as queries rather than raw images to reduce the query transmission latency. In this article, we introduce our work on low bit rate mobile landmark search, in which a compact yet discriminative landmark image descriptor is extracted by using location context such as GPS, crowd-sourced hotspot WLAN, and cell tower locations. The compactness originates from the bag-of-words image representation, with an offline learning from geotagged photos from online photo sharing websites including Flickr and Panoramio. The learning process involves segmenting the landmark photo collection by discrete geographical regions using Gaussian mixture model, and then boosting a ranking sensitive vocabulary within each region, with an “entropy” based descriptor compactness feedback to refine both phases iteratively. In online search, when entering a geographical region, the codebook in a mobile device are downstream adapted to generate extremely compact descriptors with promising discriminative ability. We have deployed landmark search apps to both HTC and iPhone mobile phones, working over the database of million scale images in typical areas like Beijing, New York, and Barcelona, and others. Our descriptor outperforms alternative compact descriptors (Chen et al. 2009; Chen et al., 2010; Chandrasekhar et al. 2009a; Chandrasekhar et al. 2009b) with significant margins. Beyond landmark search, this article will summarize the MPEG standarization progress of compact descriptor for visual search (CDVS) (Yuri et al. 2010; Yuri et al. 2011) towards application interoperability.

Nonnegative Matrix Factorization (NMF) is a widely used technique in many applications such as face recognition, motion segmentation, etc. It approximates the nonnegative data in an original high dimensional space with a linear representation in a low dimensional space by using the product of two nonnegative matrices. In many applications data are often partially corrupted with large additive noise. When the positions of noise are known, some existing variants of NMF can be applied by treating these corrupted entries as missing values. However, the positions are often unknown in many real world applications, which prevents the usage of traditional NMF or other existing variants of NMF. This paper proposes a Robust Nonnegative Matrix Factorization (RobustNMF) algorithm that explicitly models the partial corruption as large additive noise without requiring the information of positions of noise. In practice, large additive noise can be used to model outliers. In particular, the proposed method jointly approximates the clean data matrix with the product of two nonnegative matrices and estimates the positions and values of outliers/noise. An efficient iterative optimization algorithm with a solid theoretical justification has been proposed to learn the desired matrix factorization. Experimental results demonstrate the advantages of the proposed algorithm.