On the Value of ML Models
Machine Learning
2021-12-14 v1
Abstract
We argue that, when establishing and benchmarking Machine Learning (ML) models, the research community should favour evaluation metrics that better capture the value delivered by their model in practical applications. For a specific class of use cases -- selective classification -- we show that not only can it be simple enough to do, but that it has import consequences and provides insights what to look for in a ``good'' ML model.
Cite
@article{arxiv.2112.06775,
title = {On the Value of ML Models},
author = {Fabio Casati and Pierre-André Noël and Jie Yang},
journal= {arXiv preprint arXiv:2112.06775},
year = {2021}
}
Comments
Poster presentation at Workshop on Human and Machine Decisions at NeurIPS 2021 (WHMD 2021). https://sites.google.com/view/whmd2021