Energy-based Out-of-distribution Detection
This addresses the challenge of safely deploying models in open-world settings by improving OOD detection, though it is incremental as it builds on existing classifier-based methods.
The paper tackles the problem of overconfident predictions for out-of-distribution (OOD) data in machine learning models by proposing an energy-based framework for OOD detection, which reduces the average false positive rate by 18.03% compared to traditional softmax confidence scores on a CIFAR-10 benchmark.
Determining whether inputs are out-of-distribution (OOD) is an essential building block for safely deploying machine learning models in the open world. However, previous methods relying on the softmax confidence score suffer from overconfident posterior distributions for OOD data. We propose a unified framework for OOD detection that uses an energy score. We show that energy scores better distinguish in- and out-of-distribution samples than the traditional approach using the softmax scores. Unlike softmax confidence scores, energy scores are theoretically aligned with the probability density of the inputs and are less susceptible to the overconfidence issue. Within this framework, energy can be flexibly used as a scoring function for any pre-trained neural classifier as well as a trainable cost function to shape the energy surface explicitly for OOD detection. On a CIFAR-10 pre-trained WideResNet, using the energy score reduces the average FPR (at TPR 95%) by 18.03% compared to the softmax confidence score. With energy-based training, our method outperforms the state-of-the-art on common benchmarks.