Fei Zhu, Zhen Cheng, Xu-Yao Zhang, Cheng-Lin Liu
Reliable confidence estimation for the predictions is important in many safety-critical applications. However, modern deep neural networks are often overconfident for their incorrect predictions. Recently, many calibration methods have been proposed to alleviate the overconfidence problem. With calibrated confidence, a primary and practical purpose is to detect misclassification errors by filtering out low-confidence predictions (known as failure prediction). In this paper, we find a general, widely-existed but actually-neglected phenomenon that most confidence calibration methods are useless or harmful for failure prediction. We investigate this problem and reveal that popular confidence calibration methods often lead to worse confidence separation between correct and incorrect samples, making it more difficult to decide whether to trust a prediction or not. Finally, inspired by the natural connection between flat minima and confidence separation, we propose a simple hypothesis: flat minima is beneficial for failure prediction. We verify this hypothesis via extensive experiments and further boost the performance by combining two different flat minima techniques.
We run the code with torch version: 1.10.0, python version: 3.9.7
- Baseline and other calibration methods
python main_base.py
- Flat minima for failure prediction
python main_fmfp.py
A list of papers that studies out-of-distribution (OOD) detection and misclassification detection (MisD)
Our implementation references the codes in the following repositories:
Fei Zhu (zhufei2018@ia.ac.cn)