Disentangled Sequential Autoencoder

TL;DR

This paper introduces a VAE architecture that disentangles static and dynamic features in sequential data, enabling content manipulation and demonstrating improved sequence compression with stochastic RNNs.

Contribution

A novel VAE model that separates static and dynamic features in sequential data, allowing targeted content manipulation and improved long sequence modeling.

Findings

Content swapping in video and audio demonstrates effective disentanglement.

Stochastic RNNs outperform deterministic ones in sequence compression.

Empirical evidence supports stochastic RNNs' efficiency in long sequence generation.

Abstract

We present a VAE architecture for encoding and generating high dimensional sequential data, such as video or audio. Our deep generative model learns a latent representation of the data which is split into a static and dynamic part, allowing us to approximately disentangle latent time-dependent features (dynamics) from features which are preserved over time (content). This architecture gives us partial control over generating content and dynamics by conditioning on either one of these sets of features. In our experiments on artificially generated cartoon video clips and voice recordings, we show that we can convert the content of a given sequence into another one by such content swapping. For audio, this allows us to convert a male speaker into a female speaker and vice versa, while for video we can separately manipulate shapes and dynamics. Furthermore, we give empirical evidence for the hypothesis that stochastic RNNs as latent state models are more efficient at compressing and generating long sequences than deterministic ones, which may be relevant for applications in video compression.