Learning Meta-features for AutoML - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2022

Learning Meta-features for AutoML


This paper tackles the AutoML problem, aimed to automatically select an ML algorithm and its hyper-parameter configuration most appropriate to the dataset at hand. The proposed approach, MetaBu, learns new meta-features via an Optimal Transport procedure, aligning the manually designed meta-features with the space of distributions on the hyper-parameter configurations. MetaBu meta-features, learned once and for all, induce a topology on the set of datasets that is exploited to define a distribution of promising hyper-parameter configurations amenable to AutoML. Experiments on the OpenML CC-18 benchmark demonstrate that using MetaBu meta-features boosts the performance of state of the art AutoML systems, (Feurer et al. 2015) and Probabilistic Matrix Factorization (Fusi et al. 2018). Furthermore, the inspection of MetaBu meta-features gives some hints into when an ML algorithm does well. Finally, the topology based on MetaBu meta-features enables to estimate the intrinsic dimensionality of the OpenML benchmark w.r.t. a given ML algorithm or pipeline.
Fichier principal
Vignette du fichier
Metabu_camera_ready.pdf (2.36 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-03583789 , version 1 (22-02-2022)
hal-03583789 , version 2 (17-03-2022)


  • HAL Id : hal-03583789 , version 2


Herilalaina Rakotoarison, Louisot Milijaona, Andry Rasoanaivo, Michèle Sebag, Marc Schoenauer. Learning Meta-features for AutoML. ICLR 2022 - International Conference on Learning Representations (spotlight), Apr 2022, Virtual, United States. ⟨hal-03583789v2⟩
522 View
528 Download


Gmail Mastodon Facebook X LinkedIn More