Disentangled Variational Information Bottleneck for Multiview Representation Learning

TL;DR

This paper introduces a novel disentangled variational information bottleneck method for multiview representation learning, effectively separating shared and private features to improve interpretability and classification robustness.

Contribution

The work proposes DVIB, a new framework that explicitly disentangles shared and private multiview features using information bottleneck principles, with efficient mutual information optimization.

Findings

Shared and private representations preserve common and view-specific labels.

DVIB achieves comparable classification performance on corrupted images.

The method effectively disentangles multiview features for better interpretability.

Abstract

Multiview data contain information from multiple modalities and have potentials to provide more comprehensive features for diverse machine learning tasks. A fundamental question in multiview analysis is what is the additional information brought by additional views and can quantitatively identify this additional information. In this work, we try to tackle this challenge by decomposing the entangled multiview features into shared latent representations that are common across all views and private representations that are specific to each single view. We formulate this feature disentanglement in the framework of information bottleneck and propose disentangled variational information bottleneck (DVIB). DVIB explicitly defines the properties of shared and private representations using constrains from mutual information. By deriving variational upper and lower bounds of mutual information terms, representations are efficiently optimized. We demonstrate the shared and private representations learned by DVIB well preserve the common labels shared between two views and unique labels corresponding to each single view, respectively. DVIB also shows comparable performance in classification task on images with corruptions. DVIB implementation is available at https://github.com/feng-bao-ucsf/DVIB.