From Pixels to Components: Eigenvector Masking for Visual Representation Learning

TL;DR

This paper introduces a novel masked image modeling approach that masks principal components instead of pixels, leading to better high-level feature learning and improved image classification performance.

Contribution

It proposes a PCA-based masking strategy that enhances the learning of global, high-level features over traditional pixel-based masking methods.

Findings

Improved image classification accuracy with component masking

Component masking captures more global information than pixel masking

Demonstrates robustness and simplicity of the PCA-based approach

Abstract

Predicting masked from visible parts of an image is a powerful self-supervised approach for visual representation learning. However, the common practice of masking random patches of pixels exhibits certain failure modes, which can prevent learning meaningful high-level features, as required for downstream tasks. We propose an alternative masking strategy that operates on a suitable transformation of the data rather than on the raw pixels. Specifically, we perform principal component analysis and then randomly mask a subset of components, which accounts for a fixed ratio of the data variance. The learning task then amounts to reconstructing the masked components from the visible ones. Compared to local patches of pixels, the principal components of images carry more global information. We thus posit that predicting masked from visible components involves more high-level features, allowing our masking strategy to extract more useful representations. This is corroborated by our empirical findings which demonstrate improved image classification performance for component over pixel masking. Our method thus constitutes a simple and robust data-driven alternative to traditional masked image modeling approaches.