Event Representation with Sequential, Semi-Supervised Discrete Variables

2020/10/09 by Rezaee, Mehdi, Ferraro, Francis
#Computation and Language (cs.CL) #FOS: Computer and information sciences #Machine Learning (cs.LG)

paper · doi:10.48550/arxiv.2010.04361

Abstract

Within the context of event modeling and understanding, we propose a new method for neural sequence modeling that takes partially-observed sequences of discrete, external knowledge into account. We construct a sequential neural variational autoencoder, which uses Gumbel-Softmax reparametrization within a carefully defined encoder, to allow for successful backpropagation during training. The core idea is to allow semi-supervised external discrete knowledge to guide, but not restrict, the variational latent parameters during training. Our experiments indicate that our approach not only outperforms multiple baselines and the state-of-the-art in narrative script induction, but also converges more quickly.

Related

• Event2Vec: A Geometric Approach to Learning Composable Representations of Event Sequences
• Variable-rate discrete representation learning
• A Variational Autoencoder for Neural Temporal Point Processes with Dynamic Latent Graphs
• Unsupervised Learning of Disentangled and Interpretable Representations from Sequential Data
• Online Segment to Segment Neural Transduction