MaxEnt Loss: Constrained Maximum Entropy for Calibration Under Out-of-Distribution Shift
Abstract
We present a new loss function that addresses the out-of-distribution (OOD) network calibration problem. While many objective functions have been proposed to effectively calibrate models in-distribution, our findings show that they do not always fare well OOD. Based on the Principle of Maximum Entropy, we incorporate helpful statistical constraints observed during training, delivering better model calibration without sacrificing accuracy. We provide theoretical analysis and show empirically that our method works well in practice, achieving state-of-the-art calibration on both synthetic and real-world benchmarks. Our code is available at https://github.com/dexterdley/MaxEnt-Loss.
Cite
Text
Neo et al. "MaxEnt Loss: Constrained Maximum Entropy for Calibration Under Out-of-Distribution Shift." AAAI Conference on Artificial Intelligence, 2024. doi:10.1609/AAAI.V38I19.30143Markdown
[Neo et al. "MaxEnt Loss: Constrained Maximum Entropy for Calibration Under Out-of-Distribution Shift." AAAI Conference on Artificial Intelligence, 2024.](https://mlanthology.org/aaai/2024/neo2024aaai-maxent/) doi:10.1609/AAAI.V38I19.30143BibTeX
@inproceedings{neo2024aaai-maxent,
title = {{MaxEnt Loss: Constrained Maximum Entropy for Calibration Under Out-of-Distribution Shift}},
author = {Neo, Dexter and Winkler, Stefan and Chen, Tsuhan},
booktitle = {AAAI Conference on Artificial Intelligence},
year = {2024},
pages = {21463-21472},
doi = {10.1609/AAAI.V38I19.30143},
url = {https://mlanthology.org/aaai/2024/neo2024aaai-maxent/}
}