{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,15]],"date-time":"2026-03-15T04:48:18Z","timestamp":1773550098424,"version":"3.50.1"},"reference-count":52,"publisher":"Association for Computing Machinery (ACM)","issue":"3","funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"crossref","award":["U23A20387"],"award-info":[{"award-number":["U23A20387"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"crossref"}]},{"name":"Beijing Natural Science Foundation","award":["L252032"],"award-info":[{"award-number":["L252032"]}]}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["ACM Trans. Multimedia Comput. Commun. Appl."],"published-print":{"date-parts":[[2026,3,31]]},"abstract":"<jats:p>\n                    The deployment scenarios often include conditions not anticipated during training. Therefore, Out-of-Distribution (OOD) detection is essential for ensuring the reliability and security of neural networks. However, many existing OOD detectors suffer from instability, with performance degrading significantly when the dataset or model changes. This challenge highlights the need to approach OOD detection by examining intrinsic differences between In-Distribution (ID) and OOD samples in terms of model capacities, rather than relying on their observable characteristics. In this article, we propose Gradient-based Attribution Reliability for OOD Detection (GAROD), a novel method grounded in the capacity of invariance to irrelevant inputs, an important property linked to model generalization. We hypothesize that models exhibit such properties with ID samples, and samples for which the model lacks this invariance are classified as OOD. Specifically, GAROD leverages gradient-based attribution to separate relevant and irrelevant pixels in the input samples and observes how a model\u2019s decisions change after removing irrelevant pixels. The approach most closely related to ours is attribution reliability evaluation (e.g., Insertion or Deletion metrics). However, these methods have never been applied to OOD detection. Moreover, directly using classical reliability metrics does not yield effective results. We identify two key issues: (1) model outputs are insufficient to capture decision changes effectively, and (2) using Insertion or Deletion metrics individually lacks comprehensiveness. In GAROD, we address these by observing final features instead, fusing both metrics to achieve robust OOD detection. Extensive experiments on CIFAR and ImageNet benchmarks demonstrate GAROD\u2019s superiority over state-of-the-art\n                    <jats:italic toggle=\"yes\">post hoc<\/jats:italic>\n                    methods, as well as its resilience to performance degradation under dataset\/model variations. Code:\n                    <jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" ext-link-type=\"uri\" xlink:href=\"https:\/\/github.com\/iceshade000\/GAROD\">https:\/\/github.com\/iceshade000\/GAROD<\/jats:ext-link>\n                    .\n                  <\/jats:p>","DOI":"10.1145\/3787859","type":"journal-article","created":{"date-parts":[[2026,2,14]],"date-time":"2026-02-14T14:27:41Z","timestamp":1771079261000},"page":"1-17","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["GAROD: Delve into Gradient-Based Attribution Reliability for Out-of-Distribution Detection"],"prefix":"10.1145","volume":"22","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-6878-971X","authenticated-orcid":false,"given":"Guanhua","family":"Zheng","sequence":"first","affiliation":[{"name":"University of Science and Technology of China, Hefei, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0699-3205","authenticated-orcid":false,"given":"Jitao","family":"Sang","sequence":"additional","affiliation":[{"name":"The School of Computer and Information Technology and the Beijing Key Laboratory of Traffic Data Analysis and Mining, Beijing Jiaotong University, Beijing, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8343-9665","authenticated-orcid":false,"given":"Changsheng","family":"Xu","sequence":"additional","affiliation":[{"name":"National Lab of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences, Beijing, China"}]}],"member":"320","published-online":{"date-parts":[[2026,2,27]]},"reference":[{"key":"e_1_3_1_2_2","doi-asserted-by":"publisher","DOI":"10.1109\/ITA.2018.8503149"},{"key":"e_1_3_1_3_2","unstructured":"Julius Adebayo Justin Gilmer Michael Muelly Ian Goodfellow Moritz Hardt and Been Kim. 2018. Sanity checks for saliency maps. arXiv:1810.03292. Retrieved from https:\/\/arxiv.org\/abs\/1810.03292"},{"key":"e_1_3_1_4_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.173"},{"key":"e_1_3_1_5_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPRW56347.2022.00031"},{"key":"e_1_3_1_6_2","first-page":"79946","volume-title":"Proceedings of the Advances in Neural Information Processing Systems","volume":"36","author":"Chen Jinggang","year":"2023","unstructured":"Jinggang Chen, Junjie Li, Xiaoyang Qu, Jianzong Wang, Jiguang Wan, and Jing Xiao. 2023. GAIA: Delving into gradient-based attribution abnormality for out-of-distribution detection. In Proceedings of the Advances in Neural Information Processing Systems. A. Oh, T. Naumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine (Eds.), Vol. 36. Curran Associates, Inc., 79946\u201379958. Retrieved from https:\/\/proceedings.neurips.cc\/paper_files\/paper\/2023\/file\/fcdccd419c4dc471fa3b73ec97b53789-Paper-Conference.pdf"},{"key":"e_1_3_1_7_2","first-page":"1","volume-title":"Proceedings of the 37th International Conference on Machine Learning (ICML\u201920)","volume":"149","author":"Chen Ting","year":"2020","unstructured":"Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. 2020. A simple framework for contrastive learning of visual representations. In Proceedings of the 37th International Conference on Machine Learning (ICML\u201920). JMLR.org, Article 149, 1\u201311."},{"key":"e_1_3_1_8_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2014.461"},{"key":"e_1_3_1_9_2","doi-asserted-by":"crossref","unstructured":"Weijian Deng Gould Stephen and Liang Zheng. 2022. On the strong correlation between model invariance and generalization. In Proceedings of the Advances in Neural Information Processing Systems. S. Koyejo S. Mohamed A. Agarwal D. Belgrave K. Cho and A. Oh (Eds.) Vol. 35 Curran Associates Inc. 28052\u201328067. Retrieved from https:\/\/proceedings.neurips.cc\/paper_files\/paper\/2022\/file\/b3847cda0c8cc0cfcdacf462dc122214-Paper-Conference.pdf","DOI":"10.52202\/068431-2034"},{"key":"e_1_3_1_10_2","volume-title":"Proceedings of the 11th International Conference on Learning Representations","author":"Djurisic Andrija","year":"2023","unstructured":"Andrija Djurisic, Nebojsa Bozanic, Arjun Ashok, and Rosanne Liu. 2023. Extremely simple activation shaping for out-of-distribution detection. In Proceedings of the 11th International Conference on Learning Representations. Retrieved from https:\/\/openreview.net\/forum?id=ndYXTEL6cZz"},{"key":"e_1_3_1_11_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR42600.2020.00975"},{"key":"e_1_3_1_12_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"e_1_3_1_13_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00013"},{"key":"e_1_3_1_14_2","volume-title":"Proceedings of the International Conference on Learning Representations","author":"Hendrycks Dan","year":"2017","unstructured":"Dan Hendrycks and Kevin Gimpel. 2017. A baseline for detecting misclassified and out-of-distribution examples in neural networks. In Proceedings of the International Conference on Learning Representations. Retrieved from https:\/\/openreview.net\/forum?id=Hkg4TI9xl"},{"key":"e_1_3_1_15_2","unstructured":"Sara Hooker Dumitru Erhan Pieter-Jan Kindermans and Been Kim. 2018. A benchmark for interpretability methods in deep neural networks. arXiv:1806.10758. Retrieved from https:\/\/arxiv.org\/abs\/1806.10758"},{"key":"e_1_3_1_16_2","unstructured":"Rui Huang Andrew Geng and Yixuan Li. 2021. On the importance of gradients for detecting distributional shifts in the wild. In Proceedings of the Advances in Neural Information Processing Systems. M. Ranzato A. Beygelzimer Y. Dauphin P. S. Liang and J. Wortman Vaughan (Eds.) Vol. 34 Curran Associates Inc. 677\u2013689. Retrieved from https:\/\/proceedings.neurips.cc\/paper_files\/paper\/2021\/file\/063e26c670d07bb7c4d30e6fc69fe056-Paper.pdf"},{"key":"e_1_3_1_17_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR46437.2021.00860"},{"key":"e_1_3_1_18_2","volume-title":"Proceedings of the International Conference on Learning Representations","author":"Jacobsen Joern-Henrik","year":"2019","unstructured":"Joern-Henrik Jacobsen, Jens Behrmann, Richard Zemel, and Matthias Bethge. 2019. Excessive invariance causes adversarial vulnerability. In Proceedings of the International Conference on Learning Representations. Retrieved from https:\/\/openreview.net\/forum?id=BkfbpsAcF7"},{"key":"e_1_3_1_19_2","unstructured":"Saachi Jain Hadi Salman Eric Wong Pengchuan Zhang Vibhav Vineet Sai Vemprala and Aleksander Madry. 2022. Missingness bias in model debugging. In Proceedings of the 10th International Conference on Learning Representations (ICLR \u201922). Retrieved from https:\/\/www.microsoft.com\/en-us\/research\/publication\/missingness-bias-in-model-debugging\/"},{"key":"e_1_3_1_20_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-58558-7_29"},{"key":"e_1_3_1_21_2","unstructured":"Alex Krizhevsky. 2009. Learning Multiple Layers of Features from Tiny Images. Technical Report. Toronto University. Retrieved from http:\/\/www.cs.utoronto.ca\/~kriz\/learning-features-2009-TR.pdf"},{"key":"e_1_3_1_22_2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-58589-1_13"},{"key":"e_1_3_1_23_2","volume-title":"Proceedings of the International Conference on Machine Learning (ICML) Workshop on New Frontiers in Adversarial Machine Learning","author":"Lee Jinsol","year":"2022","unstructured":"Jinsol Lee, Mohit Prabhushankar, and Ghassan AlRegib. 2022. Gradient-based adversarial and out-of-distribution detection. In Proceedings of the International Conference on Machine Learning (ICML) Workshop on New Frontiers in Adversarial Machine Learning."},{"key":"e_1_3_1_24_2","first-page":"7167","volume-title":"Proceedings of the Advances in Neural Information Processing Systems","volume":"31","author":"Lee Kimin","year":"2018","unstructured":"Kimin Lee, Kibok Lee, Honglak Lee, and Jinwoo Shin. 2018. A simple unified framework for detecting out-of-distribution samples and adversarial attacks. In Proceedings of the Advances in Neural Information Processing Systems. S. Bengio, H. Wallach, H. Larochelle, K. Grauman, N. Cesa-Bianchi, and R. Garnett (Eds.), Vol. 31, Curran Associates, Inc., 7167\u20137177. Retrieved from https:\/\/proceedings.neurips.cc\/paper_files\/paper\/2018\/file\/abdeb6f575ac5c6676b747bca8d09cc2-Paper.pdf"},{"key":"e_1_3_1_25_2","volume-title":"Proceedings of the International Conference on Learning Representations","author":"Liang Shiyu","year":"2018","unstructured":"Shiyu Liang, Yixuan Li, and R. Srikant. 2018. Enhancing the reliability of out-of-distribution image detection in neural networks. In Proceedings of the International Conference on Learning Representations. Retrieved from https:\/\/openreview.net\/forum?id=H1VGkIxRZ"},{"key":"e_1_3_1_26_2","first-page":"21464","volume-title":"Proceedings of the Advances in Neural Information Processing Systems","volume":"33","author":"Liu Weitang","year":"2020","unstructured":"Weitang Liu, Xiaoyun Wang, John Owens, and Yixuan Li. 2020. Energy-based out-of-distribution detection. In Proceedings of the Advances in Neural Information Processing Systems. H. Larochelle, M. Ranzato, R. Hadsell, M. F. Balcan, and H. Lin (Eds.), Vol. 33, Curran Associates, Inc., 21464\u201321475. Retrieved from https:\/\/proceedings.neurips.cc\/paper_files\/paper\/2020\/file\/f5496252609c43eb8a3d147ab9b9c006-Paper.pdf"},{"key":"e_1_3_1_27_2","volume-title":"Proceedings of the NIPS Workshop on Deep Learning and Unsupervised Feature Learning","author":"Netzer Yuval","year":"2011","unstructured":"Yuval Netzer, Tao Wang, Adam Coates, Alessandro Bissacco, and Andrew Y. Ng. 2011. Reading digits in natural images with unsupervised feature learning. In Proceedings of the NIPS Workshop on Deep Learning and Unsupervised Feature Learning."},{"key":"e_1_3_1_28_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR42600.2020.00886"},{"key":"e_1_3_1_29_2","doi-asserted-by":"publisher","DOI":"10.1145\/3632960"},{"key":"e_1_3_1_30_2","doi-asserted-by":"publisher","DOI":"10.1109\/TNNLS.2016.2599820"},{"key":"e_1_3_1_31_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2017.74"},{"key":"e_1_3_1_32_2","first-page":"3145","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Shrikumar Avanti","year":"2017","unstructured":"Avanti Shrikumar, Peyton Greenside, and Anshul Kundaje. 2017. Learning important features through propagating activation differences. In Proceedings of the International Conference on Machine Learning. PMLR, 3145\u20133153."},{"key":"e_1_3_1_33_2","unstructured":"Karen Simonyan Andrea Vedaldi and Andrew Zisserman. 2013. Deep inside convolutional networks: Visualising image classification models and saliency maps. arXiv:1312.6034. Retrieved from https:\/\/arxiv.org\/abs\/1312.6034"},{"key":"e_1_3_1_34_2","unstructured":"Daniel Smilkov Nikhil Thorat Been Kim Fernanda Vi\u00e9gas and Martin Wattenberg. 2017. SmoothGrad: Removing noise by adding noise. arXiv:1706.03825. Retrieved from https:\/\/arxiv.org\/abs\/1706.03825"},{"key":"e_1_3_1_35_2","first-page":"17885","volume-title":"Proceedings of the Advances in Neural Information Processing Systems","volume":"35","author":"Song Yue","year":"2022","unstructured":"Yue Song, Nicu Sebe, and Wei Wang. 2022. RankFeat: Rank-1 feature removal for out-of-distribution detection. In Proceedings of the Advances in Neural Information Processing Systems. S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, and A. Oh (Eds.), Vol. 35. Curran Associates, Inc., 17885\u201317898. Retrieved from https:\/\/proceedings.neurips.cc\/paper_files\/paper\/2022\/file\/71c9eb0913e6c7fda3afd69c914b1a0c-Paper-Conference.pdf"},{"key":"e_1_3_1_36_2","unstructured":"Jost Tobias Springenberg Alexey Dosovitskiy Thomas Brox and Martin Riedmiller. 2014. Striving for simplicity: The all convolutional net. arXiv:1412.6806. Retrieved from https:\/\/arxiv.org\/abs\/1412.6806"},{"key":"e_1_3_1_37_2","unstructured":"Suraj Srinivas and Fran\u00e7ois Fleuret. 2019. Full-gradient representation for neural network visualization. arXiv:1905.00780. Retrieved from https:\/\/arxiv.org\/abs\/1905.00780"},{"key":"e_1_3_1_38_2","first-page":"144","volume-title":"Proceedings of the Advances in Neural Information Processing Systems","volume":"34","author":"Sun Yiyou","year":"2021","unstructured":"Yiyou Sun, Chuan Guo, and Yixuan Li. 2021. ReAct: Out-of-distribution detection with rectified activations. In Proceedings of the Advances in Neural Information Processing Systems. M. Ranzato, A. Beygelzimer, Y. Dauphin, P. S. Liang, and J. Wortman Vaughan (Eds.), Vol. 34, Curran Associates, Inc., 144\u2013157. Retrieved from https:\/\/proceedings.neurips.cc\/paper_files\/paper\/2021\/file\/01894d6f048493d2cacde3c579c315a3-Paper.pdf"},{"key":"e_1_3_1_39_2","first-page":"20827","volume-title":"Proceedings of the 39th International Conference on Machine Learning","volume":"162","author":"Sun Yiyou","year":"2022","unstructured":"Yiyou Sun, Yifei Ming, Xiaojin Zhu, and Yixuan Li. 2022. Out-of-distribution detection with deep nearest neighbors. In Proceedings of the 39th International Conference on Machine Learning. K. Chaudhuri, S. Jegelka, L. Song, C. Szepesvari, G. Niu, and S. Sabato (Eds.), Vol. 162, PMLR, 20827\u201320840. Retrieved from https:\/\/proceedings.mlr.press\/v162\/sun22d.html"},{"key":"e_1_3_1_40_2","first-page":"3319","volume-title":"Proceedings of the International Conference on Machine Learning","author":"Sundararajan Mukund","year":"2017","unstructured":"Mukund Sundararajan, Ankur Taly, and Qiqi Yan. 2017. Axiomatic attribution for deep networks. In Proceedings of the International Conference on Machine Learning. PMLR, 3319\u20133328."},{"key":"e_1_3_1_41_2","doi-asserted-by":"publisher","DOI":"10.1109\/ITW.2015.7133169"},{"key":"e_1_3_1_42_2","first-page":"8769","volume-title":"\u2013","author":"Van Horn Grant","year":"2018","unstructured":"Grant Van Horn, Oisin Mac Aodha, Yang Song, Yin Cui, Chen Sun, Alex Shepard, Hartwig Adam, Pietro Perona, and Serge Belongie. 2018. The iNaturalist species classification and detection dataset. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR \u201918), 8769\u2013587."},{"key":"e_1_3_1_43_2","doi-asserted-by":"publisher","DOI":"10.1145\/3534678.3539438"},{"key":"e_1_3_1_44_2","doi-asserted-by":"publisher","DOI":"10.1109\/TMI.2023.3282789"},{"key":"e_1_3_1_45_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2010.5539970"},{"key":"e_1_3_1_46_2","unstructured":"Mengjiao Yang and Been Kim. 2019. Benchmarking attribution methods with relative feature importance. arXiv:1907.09701. Retrieved from https:\/\/arxiv.org\/abs\/1907.09701"},{"key":"e_1_3_1_47_2","doi-asserted-by":"publisher","DOI":"10.1145\/3578935"},{"key":"e_1_3_1_48_2","unstructured":"Fisher Yu Ari Seff Yinda Zhang Shuran Song Thomas Funkhouser and Jianxiong Xiao. 2016. LSUN: Construction of a large-scale image dataset using deep learning with humans in the loop. arXiv:1506.03365. Retrieved from https:\/\/arxiv.org\/abs\/1506.03365"},{"key":"e_1_3_1_49_2","doi-asserted-by":"publisher","DOI":"10.5244\/C.30.87"},{"key":"e_1_3_1_50_2","doi-asserted-by":"publisher","DOI":"10.1109\/TMM.2023.3324808"},{"key":"e_1_3_1_51_2","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v34i07.7000"},{"key":"e_1_3_1_52_2","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2017.2723009"},{"key":"e_1_3_1_53_2","first-page":"4328","volume-title":"Proceedings of the Advances in Neural Information Processing Systems","volume":"34","author":"Zhu Sicheng","year":"2021","unstructured":"Sicheng Zhu, Bang An, and Furong Huang. 2021. Understanding the generalization benefit of model invariance from a data perspective. In Proceedings of the Advances in Neural Information Processing Systems. M. Ranzato, A. Beygelzimer, Y. Dauphin, P. S. Liang, and J. Wortman Vaughan (Eds.), Vol. 34. Curran Associates, Inc., 4328\u20134341. Retrieved from https:\/\/proceedings.neurips.cc\/paper_files\/paper\/2021\/file\/2287c6b8641dd2d21ab050eb9ff795f3-Paper.pdf"}],"container-title":["ACM Transactions on Multimedia Computing, Communications, and Applications"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3787859","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,15]],"date-time":"2026-03-15T03:48:57Z","timestamp":1773546537000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3787859"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,2,27]]},"references-count":52,"journal-issue":{"issue":"3","published-print":{"date-parts":[[2026,3,31]]}},"alternative-id":["10.1145\/3787859"],"URL":"https:\/\/doi.org\/10.1145\/3787859","relation":{},"ISSN":["1551-6857","1551-6865"],"issn-type":[{"value":"1551-6857","type":"print"},{"value":"1551-6865","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,2,27]]},"assertion":[{"value":"2025-05-08","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-12-13","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-02-27","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}