Chroma-VAE: Mitigating Shortcut Learning with Generative Classifiers

TL;DR

Chroma-VAE introduces a novel method that uses a variational autoencoder to separate shortcut features from essential data features, effectively reducing shortcut learning in neural networks.

Contribution

The paper presents Chroma-VAE, a two-stage generative classifier that isolates shortcut information in a small latent space, improving robustness against shortcut learning.

Findings

Chroma-VAE effectively mitigates shortcut learning on benchmark tasks.

The approach isolates shortcut features in a dedicated latent subspace.

Manipulating latent space can interpret and control learned correlations.

Abstract

Deep neural networks are susceptible to shortcut learning, using simple features to achieve low training loss without discovering essential semantic structure. Contrary to prior belief, we show that generative models alone are not sufficient to prevent shortcut learning, despite an incentive to recover a more comprehensive representation of the data than discriminative approaches. However, we observe that shortcuts are preferentially encoded with minimal information, a fact that generative models can exploit to mitigate shortcut learning. In particular, we propose Chroma-VAE, a two-pronged approach where a VAE classifier is initially trained to isolate the shortcut in a small latent subspace, allowing a secondary classifier to be trained on the complementary, shortcut-free latent subspace. In addition to demonstrating the efficacy of Chroma-VAE on benchmark and real-world shortcut learning tasks, our work highlights the potential for manipulating the latent space of generative classifiers to isolate or interpret specific correlations.