Regularization and False Alarms Quantification: Two Sides of the Explainability Coin

TL;DR

This paper explores how regularization and false alarm quantification are interconnected aspects of explainability in machine learning, emphasizing that improper estimation of either hampers the practical economic value of ML models.

Contribution

It highlights the conceptual link between regularization and false alarm costs, proposing that both are crucial for meaningful explainability and economic assessment of ML models.

Findings

Regularization impacts model interpretability and economic value.

False alarm quantification is essential for assessing model risks.

Misestimating either aspect reduces ML utility in practice.

Abstract

Regularization is a well-established technique in machine learning (ML) to achieve an optimal bias-variance trade-off which in turn reduces model complexity and enhances explainability. To this end, some hyper-parameters must be tuned, enabling the ML model to accurately fit the unseen data as well as the seen data. In this article, the authors argue that the regularization of hyper-parameters and quantification of costs and risks of false alarms are in reality two sides of the same coin, explainability. Incorrect or non-existent estimation of either quantities undermines the measurability of the economic value of using ML, to the extent that might make it practically useless.