On the Effectiveness of Low-Rank Matrix Factorization for LSTM Model Compression

Despite their ubiquity in NLP tasks, Long Short-Term Memory (LSTM) networks suffer from computational inefficiencies caused by inherent unparallelizable recurrences, which further aggravates as LSTMs require more parameters for larger memory capacity. In this paper, we propose to apply low-rank matrix factorization (MF) algorithms to different recurrences in LSTMs, and explore the effectiveness on different NLP tasks and model components. We discover that additive recurrence is more important than multiplicative recurrence, and explain this by identifying meaningful correlations between matrix norms and compression performance. We compare our approach across two settings: 1) compressing core LSTM recurrences in language models, 2) compressing biLSTM layers of ELMo evaluated in three downstream NLP tasks.

MoEL: Mixture of Empathetic Listeners

Previous research on empathetic dialogue systems has mostly focused on generating responses given certain emotions. However, being empathetic not only requires the ability of generating emotional responses, but more importantly, requires the understanding of user emotions and replying appropriately. In this paper, we propose a novel end-to-end approach for modeling empathy in dialogue systems: Mixture of Empathetic Listeners (MoEL). Our model first captures the user emotions and outputs an emotion distribution. Based on this, MoEL will softly combine the output states of the appropriate Listener(s), which are each optimized to react to certain emotions, and generate an empathetic response. Human evaluations on empathetic-dialogues (Rashkin et al., 2018) dataset confirm that MoEL outperforms multitask training baseline in terms of empathy, relevance, and fluency. Furthermore, the case study on generated responses of different Listeners shows high interpretability of our model.

Getting To Know You: User Attribute Extraction from Dialogues

User attributes provide rich and useful information for user understanding, yet structured and easy-to-use attributes are often sparsely populated. In this paper, we leverage dialogues with conversational agents, which contain strong suggestions of user information, to automatically extract user attributes. Since no existing dataset is available for this purpose, we apply distant supervision to train our proposed two-stage attribute extractor, which surpasses several retrieval and generation baselines on human evaluation. Meanwhile, we discuss potential applications (e.g., personalized recommendation and dialogue systems) of such extracted user attributes, and point out current limitations to cast light on future work.

HappyBot: Generating Empathetic Dialogue Responses by Improving User Experience Look-ahead

Recent neural conversation models that attempted to incorporate emotion and generate empathetic responses either focused on conditioning the output to a given emotion, or incorporating the current user emotional state. While these approaches have been successful to some extent in generating more diverse and seemingly engaging utterances, they do not factor in how the user would feel towards the generated dialogue response. Hence, in this paper, we advocate such look-ahead of user emotion as the key to modeling and generating empathetic dialogue responses. We thus train a Sentiment Predictor to estimate the user sentiment look-ahead towards the generated system responses, which is then used as the reward function for generating more empathetic responses. Human evaluation results show that our model outperforms other baselines in empathy, relevance, and fluency.

CAiRE_HKUST at SemEval-2019 Task 3: Hierarchical Attention for Dialogue Emotion Classification

Detecting emotion from dialogue is a challenge that has not yet been extensively surveyed. One could consider the emotion of each dialogue turn to be independent, but in this paper, we introduce a hierarchical approach to classify emotion, hypothesizing that the current emotional state depends on previous latent emotions. We benchmark several feature-based classifiers using pre-trained word and emotion embeddings, state-of-the-art end-to-end neural network models, and Gaussian processes for automatic hyper-parameter search. In our experiments, hierarchical architectures consistently give significant improvements, and our best model achieves a 76.77% F1-score on the test set.

Transferable Multi-Domain State Generator for Task-Oriented Dialogue Systems

Over-dependence on domain ontology and lack of knowledge sharing across domains are two practical and yet less studied problems of dialogue state tracking. Existing approaches generally fall short in tracking unknown slot values during inference and often have difficulties in adapting to new domains. In this paper, we propose a Transferable Dialogue State Generator (TRADE) that generates dialogue states from utterances using a copy mechanism, facilitating knowledge transfer when predicting (domain, slot, value) triplets not encountered during training. Our model is composed of an utterance encoder, a slot gate, and a state generator, which are shared across domains. Empirical results demonstrate that TRADE achieves state-of-the-art joint goal accuracy of 48.62% for the five domains of MultiWOZ, a human-human dialogue dataset. In addition, we show its transferring ability by simulating zero-shot and few-shot dialogue state tracking for unseen domains. TRADE achieves 60.58% joint goal accuracy in one of the zero-shot domains, and is able to adapt to few-shot cases without forgetting already trained domains.

Personalizing Dialogue Agents via Meta-Learning

Existing personalized dialogue models use human designed persona descriptions to improve dialogue consistency. Collecting such descriptions from existing dialogues is expensive and requires hand-crafted feature designs. In this paper, we propose to extend Model-Agnostic Meta-Learning (MAML)(Finn et al., 2017) to personalized dialogue learning without using any persona descriptions. Our model learns to quickly adapt to new personas by leveraging only a few dialogue samples collected from the same user, which is fundamentally different from conditioning the response on the persona descriptions. Empirical results on Persona-chat dataset (Zhang et al., 2018) indicate that our solution outperforms non-meta-learning baselines using automatic evaluation metrics, and in terms of human-evaluated fluency and consistency.

Learn to Code-Switch: Data Augmentation using Copy Mechanism on Language Modeling

Building large-scale datasets for training code-switching language models is challenging and very expensive. To alleviate this problem using parallel corpus has been a major workaround. However, existing solutions use linguistic constraints which may not capture the real data distribution. In this work, we propose a novel method for learning how to generate code-switching sentences from parallel corpora. Our model uses a Seq2Seq model in combination with pointer networks to align and choose words from the monolingual sentences and form a grammatical code-switching sentence. In our experiment, we show that by training a language model using the augmented sentences we improve the perplexity score by 10% compared to the LSTM baseline.

Towards End-to-end Automatic Code-Switching Speech Recognition

Speech recognition in mixed language has difficulties to adapt end-to-end framework due to the lack of data and overlapping phone sets, for example in words such as "one" in English and "w\`an" in Chinese. We propose a CTC-based end-to-end automatic speech recognition model for intra-sentential English-Mandarin code-switching. The model is trained by joint training on monolingual datasets, and fine-tuning with the mixed-language corpus. During the decoding process, we apply a beam search and combine CTC predictions and language model score. The proposed method is effective in leveraging monolingual corpus and detecting language transitions and it improves the CER by 5%.

Code-Switching Language Modeling using Syntax-Aware Multi-Task Learning

Lack of text data has been the major issue on code-switching language modeling. In this paper, we introduce multi-task learning based language model which shares syntax representation of languages to leverage linguistic information and tackle the low resource data issue. Our model jointly learns both language modeling and Part-of-Speech tagging on code-switched utterances. In this way, the model is able to identify the location of code-switching points and improves the prediction of next word. Our approach outperforms standard LSTM based language model, with an improvement of 9.7% and 7.4% in perplexity on SEAME Phase I and Phase II dataset respectively.