Undecidability of Underfitting in Learning Algorithms
Machine Learning
2021-02-11 v3 Artificial Intelligence
Formal Languages and Automata Theory
Information Theory
math.IT
Abstract
Using recent machine learning results that present an information-theoretic perspective on underfitting and overfitting, we prove that deciding whether an encodable learning algorithm will always underfit a dataset, even if given unlimited training time, is undecidable. We discuss the importance of this result and potential topics for further research, including information-theoretic and probabilistic strategies for bounding learning algorithm fit.
Cite
@article{arxiv.2102.02850,
title = {Undecidability of Underfitting in Learning Algorithms},
author = {Sonia Sehra and David Flores and George D. Montanez},
journal= {arXiv preprint arXiv:2102.02850},
year = {2021}
}
Comments
Accepted at The 2nd International Conference on Computing and Data Science (CONF-CDS 2021)