Leveraging Uncertainty in Deep Learning for Selective Classification

TL;DR

This paper introduces a mixed-integer programming framework that leverages uncertainty and predictive mean in deep learning to improve selective classification, achieving higher accuracy and better rejection quality, especially in cost-sensitive applications.

Contribution

It presents a novel mixed-integer programming approach for selective classification that effectively combines uncertainty and mean predictions, outperforming existing methods.

Findings

Superior non-rejected accuracy and rejection quality on multiple datasets

Effective extension to cost-sensitive scenarios in real-world applications

Outperforms industry standard methods in online fraud detection

Abstract

The wide and rapid adoption of deep learning by practitioners brought unintended consequences in many situations such as in the infamous case of Google Photos' racist image recognition algorithm; thus, necessitated the utilization of the quantified uncertainty for each prediction. There have been recent efforts towards quantifying uncertainty in conventional deep learning methods (e.g., dropout as Bayesian approximation); however, their optimal use in decision making is often overlooked and understudied. In this study, we propose a mixed-integer programming framework for classification with reject option (also known as selective classification), that investigates and combines model uncertainty and predictive mean to identify optimal classification and rejection regions. Our results indicate superior performance of our framework both in non-rejected accuracy and rejection quality on several publicly available datasets. Moreover, we extend our framework to cost-sensitive settings and show that our approach outperforms industry standard methods significantly for online fraud management in real-world settings.