Confidence Calibration of Classifiers with Many Classes
This addresses the issue of unreliable confidence scores in multiclass classification, which is crucial for applications requiring accurate uncertainty estimates, though it is incremental as it builds on standard calibration methods.
The paper tackles the problem of confidence calibration for classifiers with many classes, where existing methods often fail, by transforming multiclass calibration into calibrating a single surrogate binary classifier, resulting in significant enhancement of existing calibration methods as evaluated on neural networks for image or text classification.
For classification models based on neural networks, the maximum predicted class probability is often used as a confidence score. This score rarely predicts well the probability of making a correct prediction and requires a post-processing calibration step. However, many confidence calibration methods fail for problems with many classes. To address this issue, we transform the problem of calibrating a multiclass classifier into calibrating a single surrogate binary classifier. This approach allows for more efficient use of standard calibration methods. We evaluate our approach on numerous neural networks used for image or text classification and show that it significantly enhances existing calibration methods.