Annealed Multiple Choice Learning: Overcoming limitations of Winner-takes-all with annealing

TL;DR

This paper presents Annealed Multiple Choice Learning (aMCL), a novel approach that integrates simulated annealing with MCL to improve diversity and avoid suboptimal convergence in ambiguous prediction tasks.

Contribution

The paper introduces aMCL, combining annealing with MCL to enhance exploration and overcome local minima issues inherent in Winner-takes-all schemes.

Findings

aMCL improves diversity of hypotheses during training

Experimental results show better performance on benchmarks

The method effectively handles ambiguous tasks

Abstract

We introduce Annealed Multiple Choice Learning (aMCL) which combines simulated annealing with MCL. MCL is a learning framework handling ambiguous tasks by predicting a small set of plausible hypotheses. These hypotheses are trained using the Winner-takes-all (WTA) scheme, which promotes the diversity of the predictions. However, this scheme may converge toward an arbitrarily suboptimal local minimum, due to the greedy nature of WTA. We overcome this limitation using annealing, which enhances the exploration of the hypothesis space during training. We leverage insights from statistical physics and information theory to provide a detailed description of the model training trajectory. Additionally, we validate our algorithm by extensive experiments on synthetic datasets, on the standard UCI benchmark, and on speech separation.