Deep Temporal Sigmoid Belief Networks for Sequence Modeling

TL;DR

This paper introduces deep temporal sigmoid belief networks (TSBNs), a hierarchical generative model for sequence data that improves predictive accuracy and sequence synthesis through scalable learning algorithms.

Contribution

It proposes a novel deep hierarchical model for sequence modeling, with a recognition network for efficient inference and training, advancing the state-of-the-art in sequence prediction.

Findings

Achieves state-of-the-art predictive performance on various sequence datasets.

Demonstrates effective sequence synthesis capabilities.

Provides scalable learning algorithms for deep temporal models.

Abstract

Deep dynamic generative models are developed to learn sequential dependencies in time-series data. The multi-layered model is designed by constructing a hierarchy of temporal sigmoid belief networks (TSBNs), defined as a sequential stack of sigmoid belief networks (SBNs). Each SBN has a contextual hidden state, inherited from the previous SBNs in the sequence, and is used to regulate its hidden bias. Scalable learning and inference algorithms are derived by introducing a recognition model that yields fast sampling from the variational posterior. This recognition model is trained jointly with the generative model, by maximizing its variational lower bound on the log-likelihood. Experimental results on bouncing balls, polyphonic music, motion capture, and text streams show that the proposed approach achieves state-of-the-art predictive performance, and has the capacity to synthesize various sequences.