• Publications
  • Influence
Hybrid Ranking Network for Text-to-SQL
A neat approach to leverage pre-trained language models in Text-to-SQL called Hybrid Ranking Network (HydraNet) which breaks down the problem into column-wise ranking and decoding and finally assembles the column- wise outputs into a SQL query by straightforward rules.
Exploiting Explicit Paths for Multi-hop Reading Comprehension
A novel, path-based reasoning approach for the multi-hop reading comprehension task where a system needs to combine facts from multiple passages to answer a question, and can be generalized to apply to the OpenBookQA dataset, matching state-of-the-art performance.
A Question-Focused Multi-Factor Attention Network for Question Answering
This paper proposes a novel end-to-end question-focused multi-factor attention network for answer extraction using tensor-based transformation that achieves significant improvements over the best prior state-of-the-art results on three large-scale challenging QA datasets, namely NewsQA, TriviaQA and SearchQA.
Speaker-aware training of LSTM-RNNS for acoustic modelling
  • T. Tan, Y. Qian, Yu Zhang
  • Computer Science
    IEEE International Conference on Acoustics…
  • 20 March 2016
This paper studies the LSTM-RNN speaker-aware training that incorporates the speaker information during model training to normalise the speaker variability, and empirically evaluates three types of speaker representation: I-vectors, bottleneck speaker vectors and speaking rate.
Joint acoustic factor learning for robust deep neural network based automatic speech recognition
This paper shows that discriminative auxiliary input features obtained using joint acoustic factor learning for DNN adaptation can be used for adaptation and thereby improve the performance of an ASR system and can be further improved on augmenting these BN vectors to conventional i-vectors.
A Nil-Aware Answer Extraction Framework for Question Answering
This paper proposes a novel nil-aware answer span extraction framework that is capable of returning Nil or a text span from the associated passage as an answer in a single step and shows that the integration of the proposed framework significantly outperforms several strong baseline systems that use pipeline or threshold-based approaches.
Adaptation of Deep Neural Network Acoustic Models for Robust Automatic Speech Recognition
This chapter will describe various methods of estimating reliable representations for feature augmentation, focusing primarily on comparing i-vectors and other bottleneck features, and present an adaptable DNN layer parameterisation scheme based on a linear interpolation structure.
Learning to Identify Follow-Up Questions in Conversational Question Answering
A three-way attentive pooling network is proposed that determines the suitability of a follow-up question by capturing pair-wise interactions between the associated passage, the conversation history, and a candidate follow- up question.
Incorporating a Generative Front-End Layer to Deep Neural Network for Noise Robust Automatic Speech Recognition
It is shown that incorporating a GFL to DNN yields 12.1% relative improvement over a baseline multi-condition DNN, and the proposed system performs significantly better than the noise aware training method, where the per-utterance estimated noise parameters are appended to the acoustic features.
A Co-Attentive Cross-Lingual Neural Model for Dialogue Breakdown Detection
A novel dialogue breakdown detection model that jointly incorporates a pretrained cross-lingual language model and a co-attention network is proposed that outperforms all previous approaches on all evaluation metrics in both the Japanese and English tracks in Dialogue Breakdown Detection Challenge 4.