Neural Lattice-to-Sequence Models for Uncertain Inputs
arXiv:1704.00559
Abstract
The input to a neural sequence-to-sequence model is often determined by an up-stream system, e.g. a word segmenter, part of speech tagger, or speech recognizer. These up-stream models are potentially error-prone. Representing inputs through word lattices allows making this uncertainty explicit by capturing alternative sequences and their posterior probabilities in a compact form. In this work, we extend the TreeLSTM (Tai et al., 2015) into a LatticeLSTM that is able to consume word lattices, and can be used as encoder in an attentional encoder-decoder model. We integrate lattice posterior scores into this architecture by extending the TreeLSTM's child-sum and forget gates and introducing a bias term into the attention mechanism. We experiment with speech translation lattices and report consistent improvements over baselines that translate either the 1-best hypothesis or the lattice without posterior scores.
arXiv commentEMNLP 2017
References in corpus (6)
- Adam: A Method for Stochastic Optimization
- Sequence to Sequence Learning with Neural Networks
- Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation
- Finding consensus in speech recognition: word error minimization and other applications of confusion networks
- DyNet: The Dynamic Neural Network Toolkit
- Tree-to-Sequence Attentional Neural Machine Translation
Cited by in corpus (5)
- Chinese NER Using Lattice LSTM
- Confusion2Vec: Towards Enriching Vector Space Word Representations with Representational Ambiguities
- Improving the Robustness of Speech Translation
- Lattice-Based Transformer Encoder for Neural Machine Translation
- Fused Acoustic and Text Encoding for Multimodal Bilingual Pretraining and Speech Translation