Machine learning methods such as deep neural networks (DNNs), despite their success across different domains, are known to often generate incorrect predictions with high confidence on inputs outside their training distribution. The deployment of DNNs in safety-critical domains requires detection of out-of-distribution (OOD) data so that DNNs can abstain from making predictions on those. A number of methods have been recently developed for OOD detection, but there is still room for improvement. We propose the new method iDECODe, leveraging in-distribution equivariance for conformal OOD detection. It relies on a novel base non-conformity measure and a new aggregation method, used in the inductive conformal anomaly detection framework, thereby guaranteeing a bounded false detection rate. We demonstrate the efficacy of iDECODe by experiments on image and audio datasets, obtaining state-of-the-art results. We also show that iDECODe can detect adversarial examples.
@article{arxiv.2201.02331,
title = {iDECODe: In-distribution Equivariance for Conformal Out-of-distribution Detection},
author = {Ramneet Kaur and Susmit Jha and Anirban Roy and Sangdon Park and Edgar Dobriban and Oleg Sokolsky and Insup Lee},
journal= {arXiv preprint arXiv:2201.02331},
year = {2022}
}
Comments
Association for the Advancement of Artificial Intelligence (AAAI), 2022