Representation Learning for Sequence Data with Deep Autoencoding Predictive Components

TL;DR

This paper introduces Deep Autoencoding Predictive Components (DAPC), a self-supervised method for learning meaningful sequence representations by maximizing predictive information in the latent space, improving tasks like forecasting and speech recognition.

Contribution

The paper presents DAPC, a novel self-supervised learning approach that maximizes an exact estimate of predictive information without negative sampling, enhancing sequence data representations.

Findings

Recovers latent space of noisy dynamical systems

Extracts predictive features for forecasting

Improves speech recognition when pretrained

Abstract

We propose Deep Autoencoding Predictive Components (DAPC) -- a self-supervised representation learning method for sequence data, based on the intuition that useful representations of sequence data should exhibit a simple structure in the latent space. We encourage this latent structure by maximizing an estimate of predictive information of latent feature sequences, which is the mutual information between past and future windows at each time step. In contrast to the mutual information lower bound commonly used by contrastive learning, the estimate of predictive information we adopt is exact under a Gaussian assumption. Additionally, it can be computed without negative sampling. To reduce the degeneracy of the latent space extracted by powerful encoders and keep useful information from the inputs, we regularize predictive information learning with a challenging masked reconstruction loss. We demonstrate that our method recovers the latent space of noisy dynamical systems, extracts predictive features for forecasting tasks, and improves automatic speech recognition when used to pretrain the encoder on large amounts of unlabeled data.