Salvato in:
Dettagli Bibliografici
Autori principali: Perera, David, Letzelter, Victor, Mariotte, Théo, Cortés, Adrien, Chen, Mickael, Essid, Slim, Richard, Gaël
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:https://arxiv.org/abs/2407.15580
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866912191388581888
author Perera, David
Letzelter, Victor
Mariotte, Théo
Cortés, Adrien
Chen, Mickael
Essid, Slim
Richard, Gaël
author_facet Perera, David
Letzelter, Victor
Mariotte, Théo
Cortés, Adrien
Chen, Mickael
Essid, Slim
Richard, Gaël
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2407_15580
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Annealed Multiple Choice Learning: Overcoming limitations of Winner-takes-all with annealing
Perera, David
Letzelter, Victor
Mariotte, Théo
Cortés, Adrien
Chen, Mickael
Essid, Slim
Richard, Gaël
Machine Learning
Sound
Audio and Speech Processing
Probability
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.
title Annealed Multiple Choice Learning: Overcoming limitations of Winner-takes-all with annealing
topic Machine Learning
Sound
Audio and Speech Processing
Probability
url https://arxiv.org/abs/2407.15580