A Data-Driven Measure of Relative Uncertainty for Misclassification Detection
Authors:
- Eduardo DadaltoEDEduardo Dadalto
- Marco RomanelliMRMarco Romanelli
- Georg PichlerGPGeorg Pichler
- Pablo PiantanidaPPPablo Piantanida
Preprint
Abstract
Misclassification detection is an important problem in machine learning, as it allows for the identification of instances where the model's predictions are unreliable. However, conventional uncertainty measures such as Shannon entropy do not provide an effective way to infer the real uncertainty associated with the model's predictions. In this paper, we introduce a novel data-driven measure of uncertainty relative to an observer for misclassification detection. By learning patterns in the distribution of soft-predictions, our uncertainty measure can identify misclassified samples based on the predicted class probabilities. Interestingly, according to the proposed measure, soft-predictions corresponding to misclassified instances can carry a large amount of uncertainty, even though they may have low Shannon entropy. We demonstrate empirical improvements over multiple image classification tasks, outperforming state-of-the-art misclassification detection methods.Keywords
All Related Versions
This publication has 11 references indexed in Scilit:
- OpenMix: Exploring Outlier Samples for Misclassification Detection2023
- Rethinking Confidence Calibration for Failure Prediction2022
- Recent Advances in Open Set Recognition: A SurveyIEEE Transactions on Pattern Analysis and Machine Intelligence, 2021
- Threat of Adversarial Attacks on Deep Learning in Computer Vision: A SurveyIEEE Access, 2018
- Densely Connected Convolutional Networks2017
- Deep Residual Learning for Image Recognition2016
- Possible generalization of Boltzmann-Gibbs statisticsJournal of Statistical Physics, 1988
- Diversity and dissimilarity coefficients: A unified approachTheoretical Population Biology, 1982
- On optimum recognition error and reject tradeoffIEEE Transactions on Information Theory, 1970
- A Mathematical Theory of CommunicationBell System Technical Journal, 1948