Paraphrase generation is an important problem in NLP , especially in question answering, information retrieval, information extraction, conversation systems, to name a few. In this paper, we address the problem of generating paraphrases automatically. Our proposed method is based on a combination of deep generative models ( VAE ) with sequence-to-sequence models ( LSTM ) to generate paraphrases, given an input sentence. Traditional VAEs when combined with recurrent neural networks can generate free text but they are not suitable for paraphrase generation for a given sentence. We address this problem by conditioning the both, encoder and decoder sides of VAE , on the original sentence, so that it can generate the given sentence's paraphrases. Unlike most existing models, our model is simple, modular and can generate multiple paraphrases, for a given sentence. Quantitative evaluation of the proposed method on a benchmark paraphrase dataset demonstrates its efficacy, and its performance improvement over the state-of-the-art methods by a significant margin, whereas qualitative human evaluation indicate that the generated paraphrases are well-formed, grammatically correct, and are relevant to the input sentence. Furthermore, we evaluate our method on a newly released question paraphrase dataset, and establish a new baseline for future research.

Copying mechanism shows effectiveness in sequence-to-sequence based neural network models for text generation tasks, such as abstractive sentence summarization and question generation. However, existing works on modeling copying or pointing mechanism only considers single word copying from the source sentences. In this paper, we propose a novel copying framework, named Sequential Copying Networks (SeqCopyNet), which not only learns to copy single words, but also copies sequences from the input sentence. It leverages the pointer networks to explicitly select a sub-span from the source side to target side, and integrates this sequential copying mechanism to the generation process in the encoder-decoder paradigm. Experiments on abstractive sentence summarization and question generation tasks show that the proposed SeqCopyNet can copy meaningful spans and outperforms the baseline models.

We propose a new approach to conditional probability estimation for ordinal labels. First, we present a specialized hierarchical softmax variant inspired by k-d trees that leverages the inherent spatial structure of (potentially-multivariate) ordinal labels. We then adapt ideas from signal processing on noisy graphs to develop a novel regularizer for such hierarchical softmax models. Both our tree structure and regularizer independently boost the sample efficiency of a deep learning model across a series of simulation studies. Furthermore, the combination of these two techniques produces additive gains and the model does not suffer from the pathologies of other approaches in the literature. We validate our approach empirically on a suite of real-world datasets, in some cases reducing the error by nearly half in comparison to other popular methods in the literature. Our results demonstrate that our method is a powerful new modeling technique for conditional probability estimation of ordinal labels, especially in the low-to-mid sample size regimes such as those often found in biological and other physical sciences.

We propose a Las Vegas transformation of Markov Chain Monte Carlo (MCMC) estimators of Restricted Boltzmann Machines (RBMs). We denote our approach Markov Chain Las Vegas (MCLV). MCLV gives statistical guarantees in exchange for random running times. MCLV uses a stopping set built from the training data and has maximum number of Markov chain steps K (referred as MCLV-K). We present a MCLV-K gradient estimator (LVS-K) for RBMs and explore the correspondence and differences between LVS-K and Contrastive Divergence (CD-K), with LVS-K significantly outperforming CD-K training RBMs over the MNIST dataset, indicating MCLV to be a promising direction in learning generative models.

Dialogue Act recognition associate dialogue acts (i.e., semantic labels) to utterances in a conversation. The problem of associating semantic labels to utterances can be treated as a sequence labeling problem. In this work, we build a hierarchical recurrent neural network using bidirectional LSTM as a base unit and the conditional random field (CRF) as the top layer to classify each utterance into its corresponding dialogue act. The hierarchical network learns representations at multiple levels, i.e., word level, utterance level, and conversation level. The conversation level representations are input to the CRF layer, which takes into account not only all previous utterances but also their dialogue acts, thus modeling the dependency among both, labels and utterances, an important consideration of natural dialogue. We validate our approach on two different benchmark data sets, Switchboard and Meeting Recorder Dialogue Act, and show performance improvement over the state-of-the-art methods by 2.2% and 4.1% absolute points, respectively. It is worth noting that the inter-annotator agreement on Switchboard data set is 84%, and our method is able to achieve the accuracy of about 79% despite being trained on the noisy data.

Data are often labeled by many different experts with each expert only labeling a small fraction of the data and each data point being labeled by several experts. This reduces the workload on individual experts and also gives a better estimate of the unobserved ground truth. When experts disagree, the standard approaches are to treat the majority opinion as the correct label or to model the correct label as a distribution. These approaches, however, do not make any use of potentially valuable information about which expert produced which label. To make use of this extra information, we propose modeling the experts individually and then learning averaging weights for combining them, possibly in sample-specific ways. This allows us to give more weight to more reliable experts and take advantage of the unique strengths of individual experts at classifying certain types of data. Here we show that our approach leads to improvements in computer-aided diagnosis of diabetic retinopathy. We also show that our method performs better than competing algorithms by Welinder and Perona (2010); Mnih and Hinton (2012). Our work offers an innovative approach for dealing with the myriad real-world settings that use expert opinions to define labels for training.

Recently, deep residual networks have been successfully applied in many computer vision and natural language processing tasks, pushing the state-of-the-art performance with deeper and wider architectures. In this work, we interpret deep residual networks as ordinary differential equations (ODEs), which have long been studied in mathematics and physics with rich theoretical and empirical success. From this interpretation, we develop a theoretical framework on stability and reversibility of deep neural networks, and derive three reversible neural network architectures that can go arbitrarily deep in theory. The reversibility property allows a memory-efficient implementation, which does not need to store the activations for most hidden layers. Together with the stability of our architectures, this enables training deeper networks using only modest computational resources. We provide both theoretical analyses and empirical results. Experimental results demonstrate the efficacy of our architectures against several strong baselines on CIFAR-10, CIFAR-100 and STL-10 with superior or on-par state-of-the-art performance. Furthermore, we show our architectures yield superior results when trained using fewer training data.

In this paper, we propose solutions to advance answer selection in Community Question Answering (CQA). Unlike previous works, we propose a hybrid attention mechanism to model question-answer pairs. Specifically, for each word, we calculate the intra-sentence attention indicating its local importance and the inter-sentence attention implying its importance to the counterpart sentence. The inter-sentence attention is based on the interactions between question-answer pairs, and the combination of these two attention mechanisms enables us to align the most informative parts in question-answer pairs for sentence matching. Additionally, we exploit user information for answer selection due to the fact that users are more likely to provide correct answers in their areas of expertise. We model users from their written answers to alleviate data sparsity problem, and then learn user representations according to the informative parts in sentences that are useful for question-answer matching task. This mean of modelling users can bridge the semantic gap between different users, as similar users may have the same way of wording their answers. The representations of users, questions and answers are learnt in an end-to-end neural network in a mean that best explains the interrelation between question-answer pairs. We validate the proposed model on a public dataset, and demonstrate its advantages over the baselines with thorough experiments.

Estimating the travel time of any path (denoted by a sequence of connected road segments) in a city is of great importance to traffic monitoring, route planning, ridesharing, taxi/Uber dispatching, etc. However, it is a very challenging problem, affected by diverse complex factors, including spatial correlations, temporal dependencies, external conditions (e.g. weather, traffic lights). Prior work usually focuses on estimating the travel times of individual road segments or sub-paths and then summing up these times, which leads to an inaccurate estimation because such approaches do not consider road intersections/traffic lights, and local errors may accumulate. To address these issues, we propose an end-to-end Deep learning framework for Travel Time Estimation called DeepTTE that estimates the travel time of the whole path directly. More specifically, we present a geo-convolution operation by integrating the geographic information into the classical convolution, capable of capturing spatial correlations. By stacking recurrent unit on the geo-convoluton layer, our DeepTTE can capture the temporal dependencies simultaneously. A multi-task learning component is given on the top of DeepTTE, that estimates the travel time of both the entire path and each local path simultaneously during the training phase. The extensive experiments on two large-scale datasets shows our DeepTTE significantly outperforms the state-of-the-art methods.

In recent years, research interest in object retrieval has shifted from 2D towards 3D data. Despite many well-designed approaches, we point out that limitations still exist and there is tremendous room for improvement, including the heavy reliance on hand-crafted features, the separated optimization of feature extraction and object retrieval, and the lack of sufficient training samples. In this work, we address the above limitations for 3D object retrieval by developing a novel end-to-end solution named Group Pair Convolutional Neural Network (GPCNN). It can jointly learn the visual features from multiple views of a 3D model and optimize towards the object retrieval task. To tackle the insufficient training data issue, we innovatively employ a pair-wise learning scheme, which learns model parameters from the similarity of each sample pair, rather than the traditional way of learning from sparse label–sample matching. Extensive experiments on three public benchmarks show that our GPCNN solution significantly outperforms the state-of-the-art methods with 3% to 42% improvement in retrieval accuracy.