Splitting & Integrating: Out-of-Distribution Detection via Adversarial Gradient Attribution

Jiayu Zhang, Xinyi Wang, Zhibo Jin, Zhiyu Zhu, Jiahao Huang, Xinyi Zhang, Huaming Chen

Keywords: Out-of-Distribution Detection, Adversarial Gradient Attribution, Safety

Abstract: Out-of-distribution (OOD) detection is essential for enhancing the robustness and security of deep learning models in unknown and dynamic data environments. Gradient-based OOD detection methods, such as GAIA, analyse the explanation pattern representations of in-distribution (ID) and OOD samples by examining the sensitivity of model outputs w.r.t. model inputs, resulting in superior performance compared to traditional OOD detection methods. However, we argue that the non-zero gradient behaviors of OOD samples do not exhibit significant distinguishability, especially when ID samples are perturbed by random noise in high-dimensional spaces, which negatively impacts the accuracy of OOD detection. In this paper, we propose a novel OOD detection method called **S \& I** based on layer **S**plitting and gradient **I**ntegration via Adversarial Gradient Attribution. Specifically, our approach involves splitting the model's intermediate layers and iteratively updating adversarial examples layer-by-layer. We then integrate the attribution gradients from each intermediate layer along the attribution path from adversarial examples to the actual input, yielding true explanation pattern representations for both ID and OOD samples. Experiments demonstrate that our S \& I algorithm achieves state-of-the-art results, with the average FPR95 of 29.05\% (38.61\%) and 37.31\% on the CIFAR100 and ImageNet benchmarks, respectively. Our code is available at: https://anonymous.4open.science/r/S-I-F6F7/.

Primary Area: alignment, fairness, safety, privacy, and societal considerations

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 9432