An Information-Theoretic Perspective on Overfitting and Underfitting
Abstract
We present an information-theoretic framework for understanding overfitting and underfitting in machine learning and prove the formal undecidability of determining whether an arbitrary classification algorithm will overfit a dataset. Measuring algorithm capacity via the information transferred from datasets to models, we consider mismatches between algorithm capacities and datasets to provide a signature for when a model can overfit or underfit a dataset. We present results upper-bounding algorithm capacity, establish its relationship to quantities in the algorithmic search framework for machine learning, and relate our work to recent information-theoretic approaches to generalization.
Cite
@article{arxiv.2010.06076,
title = {An Information-Theoretic Perspective on Overfitting and Underfitting},
author = {Daniel Bashir and George D. Montanez and Sonia Sehra and Pedro Sandoval Segura and Julius Lauw},
journal= {arXiv preprint arXiv:2010.06076},
year = {2020}
}
Comments
Accepted for presentation at The 33rd Australasian Joint Conference on Artificial Intelligence (AJCAI 2020), November 29-30, 2020