Multi-label out-of-distribution detection with spectral normalized joint energy

In today’s interconnected world, achieving reliable out-of-distribution (OOD) detection poses a significant challenge for machine learning models. While numerous studies have introduced improved approaches for multi-class OOD detection tasks, the investigation into multi-label OOD detection tasks has been notably limited. We introduce Spectral Normalized Joint Energy (SNoJoE), a method that consolidates label-specific information across multiple labels through the theoretically justified concept of an energy-based function. Throughout the training process, we employ spectral normalization to manage the model’s feature space, thereby enhancing model efficacy and generalization, in addition to bolstering robustness. Our findings indicate that the application of spectral normalization to joint energy scores notably amplifies the model’s capability for OOD detection. We perform OOD detection experiments utilizing PASCAL-VOC or MS-COCO as the in-distribution dataset and ImageNet-22K or Texture as the out-of-distribution datasets. Our experimental results reveal that, in comparison to prior top performances, SNoJoE achieves 11% and 54% relative reductions in FPR95 on ImageNet-22K and Texture, respectively, when using PASCAL-VOC as the in-distribution dataset. Similarly, with MS-COCO as the in-distribution dataset, SNoJoE achieves 11.3% and 42.58% relative reductions on ImageNet-22K and Texture. These improvements establish a new state of the art in OOD detection and further validate the effectiveness of incorporating spectral normalization.