Training Uncertainty-Aware Classifiers with Conformalized Deep Learning
This addresses the issue of unreliable uncertainty estimation in deep learning classifiers, which is crucial for applications requiring trustworthy predictions, though it is an incremental improvement building on conformal inference.
The paper tackles the problem of deep neural networks being overconfident and lacking reliable uncertainty estimates in multi-class classification by developing a novel training algorithm that produces models with more dependable uncertainty without sacrificing predictive power. Experiments show this method leads to smaller conformal prediction sets with higher conditional coverage compared to state-of-the-art alternatives.
Deep neural networks are powerful tools to detect hidden patterns in data and leverage them to make predictions, but they are not designed to understand uncertainty and estimate reliable probabilities. In particular, they tend to be overconfident. We begin to address this problem in the context of multi-class classification by developing a novel training algorithm producing models with more dependable uncertainty estimates, without sacrificing predictive power. The idea is to mitigate overconfidence by minimizing a loss function, inspired by advances in conformal inference, that quantifies model uncertainty by carefully leveraging hold-out data. Experiments with synthetic and real data demonstrate this method can lead to smaller conformal prediction sets with higher conditional coverage, after exact calibration with hold-out data, compared to state-of-the-art alternatives.