Learning Robust Representations for Computer Vision

TL;DR

This paper advances robust unsupervised representation learning in computer vision by introducing new methods for robust PCA and spectral clustering, effectively handling noise and outliers in real-world data.

Contribution

It presents a novel robust PCA technique for separating foreground from background and a robust spectral clustering method for high-accuracy facial image clustering.

Findings

Robust PCA outperforms standard methods in separating foreground and background.

Robust spectral clustering achieves high accuracy in facial image clustering.

Both methods demonstrate superior performance on real-world datasets.

Abstract

Unsupervised learning techniques in computer vision often require learning latent representations, such as low-dimensional linear and non-linear subspaces. Noise and outliers in the data can frustrate these approaches by obscuring the latent spaces. Our main goal is deeper understanding and new development of robust approaches for representation learning. We provide a new interpretation for existing robust approaches and present two specific contributions: a new robust PCA approach, which can separate foreground features from dynamic background, and a novel robust spectral clustering method, that can cluster facial images with high accuracy. Both contributions show superior performance to standard methods on real-world test sets.