Catalyst: Out-of-Distribution Detection via Elastic Scaling
This addresses the critical need for safe deployment of deep neural networks by enhancing OOD detection, though it is incremental as it builds on existing post-hoc methods.
The paper tackled the problem of out-of-distribution detection in deep neural networks by introducing Catalyst, a post-hoc framework that uses raw channel-wise statistics from pre-pooling feature maps to compute an input-dependent scaling factor, which improves detection performance. It achieved reductions in average False Positive Rate by 32.87% on CIFAR-10, 27.94% on CIFAR-100, and 22.25% on ImageNet.
Out-of-distribution (OOD) detection is critical for the safe deployment of deep neural networks. State-of-the-art post-hoc methods typically derive OOD scores from the output logits or penultimate feature vector obtained via global average pooling (GAP). We contend that this exclusive reliance on the logit or feature vector discards a rich, complementary signal: the raw channel-wise statistics of the pre-pooling feature map lost in GAP. In this paper, we introduce Catalyst, a post-hoc framework that exploits these under-explored signals. Catalyst computes an input-dependent scaling factor ($γ$) on-the-fly from these raw statistics (e.g., mean, standard deviation, and maximum activation). This $γ$ is then fused with the existing baseline score, multiplicatively modulating it -- an ``elastic scaling'' -- to push the ID and OOD distributions further apart. We demonstrate Catalyst is a generalizable framework: it seamlessly integrates with logit-based methods (e.g., Energy, ReAct, SCALE) and also provides a significant boost to distance-based detectors like KNN. As a result, Catalyst achieves substantial and consistent performance gains, reducing the average False Positive Rate by 32.87 on CIFAR-10 (ResNet-18), 27.94% on CIFAR-100 (ResNet-18), and 22.25% on ImageNet (ResNet-50). Our results highlight the untapped potential of pre-pooling statistics and demonstrate that Catalyst is complementary to existing OOD detection approaches.