Deep Incomplete Multi-view Learning via Cyclic Permutation of VAEs

TL;DR

This paper introduces MVP, a novel multi-view learning method using cyclic permutation of VAEs to handle incomplete data, improve view consistency, and enhance multi-view clustering and generation performance.

Contribution

The paper proposes a cyclic permutation-based VAE framework for incomplete multi-view learning, enabling invariant view relationships and missing view inference.

Findings

Outperforms existing methods on seven datasets with various missing ratios.

Effectively infers missing views and improves clustering accuracy.

Enhances multi-view generation quality.

Abstract

Multi-View Representation Learning (MVRL) aims to derive a unified representation from multi-view data by leveraging shared and complementary information across views. However, when views are irregularly missing, the incomplete data can lead to representations that lack sufficiency and consistency. To address this, we propose Multi-View Permutation of Variational Auto-Encoders (MVP), which excavates invariant relationships between views in incomplete data. MVP establishes inter-view correspondences in the latent space of Variational Auto-Encoders, enabling the inference of missing views and the aggregation of more sufficient information. To derive a valid Evidence Lower Bound (ELBO) for learning, we apply permutations to randomly reorder variables for cross-view generation and then partition them by views to maintain invariant meanings under permutations. Additionally, we enhance consistency by introducing an informational prior with cyclic permutations of posteriors, which turns the regularization term into a similarity measure across distributions. We demonstrate the effectiveness of our approach on seven diverse datasets with varying missing ratios, achieving superior performance in multi-view clustering and generation tasks.