Reassessing How to Compare and Improve the Calibration of Machine Learning Models

📅 2024-06-06
🏛️ arXiv.org
📈 Citations: 1
Influential: 0
📄 PDF

career value

198K/year
🤖 AI Summary
This paper addresses the problem that conventional calibration evaluation of deep learning models is vulnerable to spurious recalibration—i.e., trivial post-hoc adjustments that improve calibration metrics without enhancing generalization. To tackle this, we propose a novel joint evaluation paradigm integrating calibration and generalization. First, we derive a Bregman-divergence-based decomposition of calibration error, establishing the first theoretical connection between calibration metrics and generalization objectives (e.g., negative log-likelihood). Second, we design a new reliability diagram that jointly visualizes calibration bias and estimated generalization error. Third, we characterize multiple “pseudo-optimal” calibration phenomena and provide theoretically grounded, detectable criteria for identifying trivial recalibration. Experiments on standard benchmarks demonstrate that our approach significantly improves model diagnostic capability, yielding a more reliable and interpretable evaluation framework for calibration research.

Technology Category

Application Category

📝 Abstract
A machine learning model is calibrated if its predicted probability for an outcome matches the observed frequency for that outcome conditional on the model prediction. This property has become increasingly important as the impact of machine learning models has continued to spread to various domains. As a result, there are now a dizzying number of recent papers on measuring and improving the calibration of (specifically deep learning) models. In this work, we reassess the reporting of calibration metrics in the recent literature. We show that there exist trivial recalibration approaches that can appear seemingly state-of-the-art unless calibration and prediction metrics (i.e. test accuracy) are accompanied by additional generalization metrics such as negative log-likelihood. We then derive a calibration-based decomposition of Bregman divergences that can be used to both motivate a choice of calibration metric based on a generalization metric, and to detect trivial calibration. Finally, we apply these ideas to develop a new extension to reliability diagrams that can be used to jointly visualize calibration as well as the estimated generalization error of a model.
Problem

Research questions and friction points this paper is trying to address.

Reassessing calibration metrics in machine learning
Developing visualization for calibration and generalization error
Proving relationship between full and confidence calibration error
Innovation

Methods, ideas, or system contributions that make the work stand out.

Calibration-based decomposition of Bregman divergences
New extension to reliability diagrams
Consistency of kernel regression estimator
🔎 Similar Papers