Entropic Out-of-Distribution Detection
This work solves the seamless OOD detection problem for machine learning practitioners by enabling fast, efficient inference without extra data or hyperparameter tuning, though it appears incremental as it builds on existing loss functions.
The paper tackled the problem of out-of-distribution (OOD) detection by addressing issues like hyperparameter validation and accuracy drop, proposing the IsoMax loss and entropic score as a drop-in replacement for SoftMax loss, which significantly improved OOD detection performance without these drawbacks.
Out-of-distribution (OOD) detection approaches usually present special requirements (e.g., hyperparameter validation, collection of outlier data) and produce side effects (e.g., classification accuracy drop, slower energy-inefficient inferences). We argue that these issues are a consequence of the SoftMax loss anisotropy and disagreement with the maximum entropy principle. Thus, we propose the IsoMax loss and the entropic score. The seamless drop-in replacement of the SoftMax loss by IsoMax loss requires neither additional data collection nor hyperparameter validation. The trained models do not exhibit classification accuracy drop and produce fast energy-efficient inferences. Moreover, our experiments show that training neural networks with IsoMax loss significantly improves their OOD detection performance. The IsoMax loss exhibits state-of-the-art performance under the mentioned conditions (fast energy-efficient inference, no classification accuracy drop, no collection of outlier data, and no hyperparameter validation), which we call the seamless OOD detection task. In future work, current OOD detection methods may replace the SoftMax loss with the IsoMax loss to improve their performance on the commonly studied non-seamless OOD detection problem.