Question classification is an important task with wide applications. However, traditional techniques treat questions as general sentences, ignoring the corresponding answer data. In order to consider answer information into question modeling, we first introduce novel group sparse autoencoders which refine question representation by utilizing group information in the answer set. We then propose novel group sparse CNNs which naturally learn question representation with respect to their answers by implanting group sparse autoencoders into traditional CNNs. The proposed model significantly outperform strong baselines on four datasets.
Deep learning (DL) training-as-a-service (TaaS) is an important emerging industrial workload. The unique challenge of TaaS is that it must satisfy a wide range of customers who have no experience and resources to tune DL hyper-parameters, and meticulous tuning for each user's dataset is prohibitively expensive. Therefore, TaaS hyper-parameters must be fixed with values that are applicable to all users. IBM Watson Natural Language Classifier (NLC) service, the most popular IBM cognitive service used by thousands of enterprise-level clients around the globe, is a typical TaaS service. By evaluating the NLC workloads, we show that only the conservative hyper-parameter setup (e.g., small mini-batch size and small learning rate) can guarantee acceptable model accuracy for a wide range of customers. We further justify theoretically why such a setup guarantees better model convergence in general. Unfortunately, the small mini-batch size causes a high volume of communication traffic in a parameter-server based system. We characterize the high communication bandwidth requirement of TaaS using representative industrial deep learning workloads and demonstrate that none of the state-of-the-art scale-up or scale-out solutions can satisfy such a requirement. We then present GaDei, an optimized shared-memory based scale-up parameter server design. We prove that the designed protocol is deadlock-free and it processes each gradient exactly once. Our implementation is evaluated on both commercial benchmarks and public benchmarks to demonstrate that it significantly outperforms the state-of-the-art parameter-server based implementation while maintaining the required accuracy and our implementation reaches near the best possible runtime performance, constrained only by the hardware limitation. Furthermore, to the best of our knowledge, GaDei is the only scale-up DL system that provides fault-tolerance.
Learning to remember long sequences remains a challenging task for recurrent neural networks. Register memory and attention mechanisms were both proposed to resolve the issue with either high computational cost to retain memory differentiability, or by discounting the RNN representation learning towards encoding shorter local contexts than encouraging long sequence encoding. Associative memory, which studies the compression of multiple patterns in a fixed size memory, were rarely considered in recent years. Although some recent work tries to introduce associative memory in RNN and mimic the energy decay process in Hopfield nets, it inherits the shortcoming of rule-based memory updates, and the memory capacity is limited. This paper proposes a method to learn the memory update rule jointly with task objective to improve memory capacity for remembering long sequences. Also, we propose an architecture that uses multiple such associative memory for more complex input encoding. We observed some interesting facts when compared to other RNN architectures on some well-studied sequence learning tasks.
Sentence-level classification and sequential labeling are two fundamental tasks in language understanding. While these two tasks are usually modeled separately, in reality, they are often correlated, for example in intent classification and slot filling, or in topic classification and named-entity recognition. In order to utilize the potential benefits from their correlations, we propose a jointly trained model for learning the two tasks simultaneously via Long Short-Term Memory (LSTM) networks. This model predicts the sentence-level category and the word-level label sequence from the stepwise output hidden representations of LSTM. We also introduce a novel mechanism of "sparse attention" to weigh words differently based on their semantic relevance to sentence-level classification. The proposed method outperforms baseline models on ATIS and TREC datasets.
We present a new topic model that generates documents by sampling a topic for one whole sentence at a time, and generating the words in the sentence using an RNN decoder that is conditioned on the topic of the sentence. We argue that this novel formalism will help us not only visualize and model the topical discourse structure in a document better, but also potentially lead to more interpretable topics since we can now illustrate topics by sampling representative sentences instead of bag of words or phrases. We present a variational auto-encoder approach for learning in which we use a factorized variational encoder that independently models the posterior over topical mixture vectors of documents using a feed-forward network, and the posterior over topic assignments to sentences using an RNN. Our preliminary experiments on two different datasets indicate early promise, but also expose many challenges that remain to be addressed.
We propose discriminative adversarial networks (DAN) for semi-supervised learning and loss function learning. Our DAN approach builds upon generative adversarial networks (GANs) and conditional GANs but includes the key differentiator of using two discriminators instead of a generator and a discriminator. DAN can be seen as a framework to learn loss functions for predictors that also implements semi-supervised learning in a straightforward manner. We propose instantiations of DAN for two different prediction tasks: classification and ranking. Our experimental results on three datasets of different tasks demonstrate that DAN is a promising framework for both semi-supervised learning and learning loss functions for predictors. For all tasks, the semi-supervised capability of DAN can significantly boost the predictor performance for small labeled sets with minor architecture changes across tasks. Moreover, the loss functions automatically learned by DANs are very competitive and usually outperform the standard pairwise and negative log-likelihood loss functions for both semi-supervised and supervised learning.
Relation detection is a core component for many NLP applications including Knowledge Base Question Answering (KBQA). In this paper, we propose a hierarchical recurrent neural network enhanced by residual learning that detects KB relations given an input question. Our method uses deep residual bidirectional LSTMs to compare questions and relation names via different hierarchies of abstraction. Additionally, we propose a simple KBQA system that integrates entity linking and our proposed relation detector to enable one enhance another. Experimental results evidence that our approach achieves not only outstanding relation detection performance, but more importantly, it helps our KBQA system to achieve state-of-the-art accuracy for both single-relation (SimpleQuestions) and multi-relation (WebQSP) QA benchmarks.
This paper proposes a new model for extracting an interpretable sentence embedding by introducing self-attention. Instead of using a vector, we use a 2-D matrix to represent the embedding, with each row of the matrix attending on a different part of the sentence. We also propose a self-attention mechanism and a special regularization term for the model. As a side effect, the embedding comes with an easy way of visualizing what specific parts of the sentence are encoded into the embedding. We evaluate our model on 3 different tasks: author profiling, sentiment classification, and textual entailment. Results show that our model yields a significant performance gain compared to other sentence embedding methods in all of the 3 tasks.
Many natural language understanding (NLU) tasks, such as shallow parsing (i.e., text chunking) and semantic slot filling, require the assignment of representative labels to the meaningful chunks in a sentence. Most of the current deep neural network (DNN) based methods consider these tasks as a sequence labeling problem, in which a word, rather than a chunk, is treated as the basic unit for labeling. These chunks are then inferred by the standard IOB (Inside-Outside-Beginning) labels. In this paper, we propose an alternative approach by investigating the use of DNN for sequence chunking, and propose three neural models so that each chunk can be treated as a complete unit for labeling. Experimental results show that the proposed neural sequence chunking models can achieve start-of-the-art performance on both the text chunking and slot filling tasks.
We present two novel and contrasting Recurrent Neural Network (RNN) based architectures for extractive summarization of documents. The Classifier based architecture sequentially accepts or rejects each sentence in the original document order for its membership in the final summary. The Selector architecture, on the other hand, is free to pick one sentence at a time in any arbitrary order to piece together the summary. Our models under both architectures jointly capture the notions of salience and redundancy of sentences. In addition, these models have the advantage of being very interpretable, since they allow visualization of their predictions broken up by abstract features such as information content, salience and redundancy. We show that our models reach or outperform state-of-the-art supervised models on two different corpora. We also recommend the conditions under which one architecture is superior to the other based on experimental evidence.