Neural Variational Inference for Text Processing

19 Nov 2015  ยท  Yishu Miao, Lei Yu, Phil Blunsom ยท

Recent advances in neural variational inference have spawned a renaissance in deep latent variable models. In this paper we introduce a generic variational inference framework for generative and conditional models of text. While traditional variational methods derive an analytic approximation for the intractable distributions over latent variables, here we construct an inference network conditioned on the discrete text input to provide the variational distribution. We validate this framework on two very different text modelling applications, generative document modelling and supervised question answering. Our neural variational document model combines a continuous stochastic document representation with a bag-of-words generative model and achieves the lowest reported perplexities on two standard test corpora. The neural answer selection model employs a stochastic representation layer within an attention mechanism to extract the semantics between a question and answer pair. On two question answering benchmarks this model exceeds all previous published benchmarks.

PDF Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Topic Models 20 Newsgroups NVDM Test perplexity 836 # 2
Question Answering QASent Attentive LSTM MAP 0.7339 # 1
MRR 0.8117 # 1
Question Answering QASent LSTM MAP 0.6436 # 5
MRR 0.7235 # 5
Question Answering QASent LSTM (lexical overlap + dist output) MAP 0.7228 # 2
MRR 0.7986 # 2
Question Answering WikiQA Attentive LSTM MAP 0.6886 # 15
MRR 0.7069 # 15
Question Answering WikiQA LSTM MAP 0.6552 # 20
MRR 0.6747 # 20
Question Answering WikiQA LSTM (lexical overlap + dist output) MAP 0.682 # 17
MRR 0.6988 # 17

Methods


No methods listed for this paper. Add relevant methods here