Robust Ordinal VAE: Employing Noisy Pairwise Comparisons for Disentanglement

TL;DR

Robust Ordinal VAE (ROVAE) leverages noisy pairwise comparisons to improve disentanglement in VAEs, automatically identifying trustworthy comparisons and outperforming existing methods especially with noisy data.

Contribution

This paper introduces ROVAE, a novel VAE framework that uses noisy pairwise ordinal comparisons for disentanglement, with an automatic trustworthiness mechanism.

Findings

ROVAE outperforms existing methods in benchmark datasets.

ROVAE is more robust to noisy pairwise comparisons.

Effective in real-world applications with noisy data.

Abstract

Recent work by Locatello et al. (2018) has shown that an inductive bias is required to disentangle factors of interest in Variational Autoencoder (VAE). Motivated by a real-world problem, we propose a setting where such bias is introduced by providing pairwise ordinal comparisons between instances, based on the desired factor to be disentangled. For example, a doctor compares pairs of patients based on the level of severity of their illnesses, and the desired factor is a quantitive level of the disease severity. In a real-world application, the pairwise comparisons are usually noisy. Our method, Robust Ordinal VAE (ROVAE), incorporates the noisy pairwise ordinal comparisons in the disentanglement task. We introduce non-negative random variables in ROVAE, such that it can automatically determine whether each pairwise ordinal comparison is trustworthy and ignore the noisy comparisons. Experimental results demonstrate that ROVAE outperforms existing methods and is more robust to noisy pairwise comparisons in both benchmark datasets and a real-world application.