Gabriel Laberge

Gabriel Laberge

PhD working on explainability of black-box machine learning models

Partial Order in Chaos: Consensus on Feature Attributions in the Rashomon Set

Published in Journal of Machine Learning Research, 2023

This paper tackles the challenge of deriving insights from post-hoc explanations of ML models, given the existence of an infinite set of models with good empirical performance. It is proposed to take a consensus on the local/global feature importance across all models from the Rashomon Set to provide partial orders of feature importance.

houses

compas

Paper

Github

Recommended citation: Laberge, G., Pequignot, Y., Mathieu, A., Khomh, F., & Marchand, M. (2023). Partial Order in Chaos: Consensus on Feature Attributions in the Rashomon Set. Journal of Machine Learning Research, 24(364), 1-50.