Outlier-Robust Convex Segmentation

TL;DR

This paper introduces a convex optimization approach for segmenting sequential data that explicitly accounts for outliers, providing algorithms and theoretical guarantees, and demonstrating improved robustness in speech segmentation tasks.

Contribution

It presents a novel convex segmentation method that explicitly models outliers, along with two algorithms and theoretical consistency results for two segments without outliers.

Findings

Algorithms outperform baseline segmentation methods

Robustness to outliers demonstrated on speech data

Consistency results established for specific cases

Abstract

We derive a convex optimization problem for the task of segmenting sequential data, which explicitly treats presence of outliers. We describe two algorithms for solving this problem, one exact and one a top-down novel approach, and we derive a consistency results for the case of two segments and no outliers. Robustness to outliers is evaluated on two real-world tasks related to speech segmentation. Our algorithms outperform baseline segmentation algorithms.