Herding Generalizes Diverse M -Best Solutions

TL;DR

This paper reveals that the divMbest algorithm for extracting diverse solutions from Conditional Random Fields is equivalent to Herding, and shows how this insight can improve solution diversity in semantic segmentation tasks.

Contribution

It demonstrates the equivalence between divMbest and Herding, enabling better handling of constraints and improving diverse solution extraction in practical applications.

Findings

divMbest is equivalent to Herding.

Herding properties can identify implausible constraints.

Improved semantic segmentation results using Herding-based approaches.

Abstract

We show that the algorithm to extract diverse M -solutions from a Conditional Random Field (called divMbest [1]) takes exactly the form of a Herding procedure [2], i.e. a deterministic dynamical system that produces a sequence of hypotheses that respect a set of observed moment constraints. This generalization enables us to invoke properties of Herding that show that divMbest enforces implausible constraints which may yield wrong assumptions for some problem settings. Our experiments in semantic segmentation demonstrate that seeing divMbest as an instance of Herding leads to better alternatives for the implausible constraints of divMbest.