Can we reconcile safety objectives with machine learning performances? - Archive ouverte HAL
Communication Dans Un Congrès Année : 2022

Can we reconcile safety objectives with machine learning performances?

Résumé

The strong demand for more automated transport systems with enhanced safety, in conjunction with the explosion of technologies and products implementing machine learning (ML) techniques, has led to a fundamental questioning of the trust placed in machine learning. In particular, do state-of-the-art ML models allow us to reach such safety objectives? We explore this question through two practical examples from the railway and automotive industries, showing that ML performances are currently far from those required by safety objectives. We then describe and question several techniques aimed at reducing the error rate of ML components: model diversification, monitoring, classification with a reject option, conformal prediction, and temporal redundancy. Taking inspiration from a historical example, we finally discuss when and how new ML-based technologies could be introduced.
Fichier principal
Vignette du fichier
ERTS2022_paper_39.pdf (638.43 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03765471 , version 1 (31-08-2022)

Identifiants

  • HAL Id : hal-03765471 , version 1

Citer

Lucian Alecu, Hugues Bonnin, Thomas Fel, Laurent Gardes, Sébastien Gerchinovitz, et al.. Can we reconcile safety objectives with machine learning performances?. ERTS 2022, Jun 2022, TOULOUSE, France. ⟨hal-03765471⟩
547 Consultations
409 Téléchargements

Partager

More