{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,9]],"date-time":"2026-05-09T05:25:47Z","timestamp":1778304347720,"version":"3.51.4"},"reference-count":64,"publisher":"Association for Computing Machinery (ACM)","issue":"1","funder":[{"name":"Natural Sciences and Engineering Research Council (NSERC) of Canada under Discovery","award":["RGPIN-2019-06348, RGPIN-2020-05410, RGPIN2021-02970 and DGECR-2021-00187"],"award-info":[{"award-number":["RGPIN-2019-06348, RGPIN-2020-05410, RGPIN2021-02970 and DGECR-2021-00187"]}]},{"name":"Guangdong Pearl River Talent Recruitment Program","award":["2019ZT08X603"],"award-info":[{"award-number":["2019ZT08X603"]}]},{"name":"Guangdong Pearl River Talent Plan","award":["2019JC01X235"],"award-info":[{"award-number":["2019JC01X235"]}]},{"name":"UBC PMC-Sierra Professorship in Networking and Communications"}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["ACM Trans. Web"],"published-print":{"date-parts":[[2026,2,28]]},"abstract":"<jats:p>Traditional centralized model training assumes that data samples are readily available and can be processed without constraints. In contrast, decentralized machine learning (DML) addresses the limitation by collaborative model training and inference directly on distributed data sources. The transformation from data centralization to decentralization helps comply with data regulations and improves system scalability with reduced reliance on cloud servers. However, a tradeoff between model personalization and generalization exists: the fine-tuning of local training data distribution sacrifices model generalization on the testing data distribution that differs from the training data distribution. To improve the tradeoff, we propose a DML framework that can inherently make model personalization and generalization easier by selecting a model among multiple ones judiciously. We develop a scalable selector for model selection and use blockchain to achieve model consensus. The personalized model selector is then proposed for test-time adaptation. Using computer simulations, we show that our method not only outperforms competitive personalization benchmarks but also generalizes well for new data distributions with various shifts.<\/jats:p>","DOI":"10.1145\/3764936","type":"journal-article","created":{"date-parts":[[2025,8,29]],"date-time":"2025-08-29T11:47:32Z","timestamp":1756468052000},"page":"1-26","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":2,"title":["Decentralized Model Selection for Test-Time Adaptation in Heterogeneous Connected Systems"],"prefix":"10.1145","volume":"20","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-9086-5467","authenticated-orcid":false,"given":"Yao","family":"Du","sequence":"first","affiliation":[{"name":"Department of Electrical and Computer Engineering, The University of British Columbia - Vancouver Campus","place":["Vancouver, Canada"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9911-2069","authenticated-orcid":false,"given":"Cyril","family":"Leung","sequence":"additional","affiliation":[{"name":"Department of Electrical and Computer Engineering, The University of British Columbia - Vancouver Campus","place":["Vancouver, Canada"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9040-847X","authenticated-orcid":false,"given":"Zehua","family":"Wang","sequence":"additional","affiliation":[{"name":"School of Computer Science and Technology, China University of Mining and Technology - Xuzhou Campus","place":["Xuzhou, China"]},{"name":"Department of Electrical and Computer Engineering, The University of British Columbia - Vancouver Campus","place":["Xuzhou, China"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8833-0244","authenticated-orcid":false,"given":"Xiaoxiao","family":"Li","sequence":"additional","affiliation":[{"name":"Electrical and Computer Engineering, The University of British Columbia - Vancouver Campus","place":["Vancouver, Canada"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3529-2640","authenticated-orcid":false,"given":"Victor","family":"Leung","sequence":"additional","affiliation":[{"name":"Artificial Intelligence Research Institute, Shenzhen MSU-BIT University","place":["Shenzhen, China"]},{"name":"Electrical and Computer Engineering, The University of British Columbia - Vancouver Campus","place":["Shenzhen, China"]}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"320","published-online":{"date-parts":[[2026,2,18]]},"reference":[{"key":"e_1_3_2_2_2","doi-asserted-by":"publisher","DOI":"10.1109\/SP40000.2020.00044"},{"key":"e_1_3_2_3_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2015.7298911"},{"key":"e_1_3_2_4_2","unstructured":"Juan Benet. 2014. IPFS\u2014content addressed versioned P2P file system. (2014). arXiv:1407.3561. Retrieved from https:\/\/arxiv.org\/abs\/1407.3561"},{"key":"e_1_3_2_5_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.00816"},{"key":"e_1_3_2_6_2","first-page":"173","volume-title":"USENIX Symposium on Operating Systems Design and Implementation","author":"Castro Miguel","year":"1999","unstructured":"Miguel Castro, Barbara Liskov, et\u00a0al. 1999. Practical byzantine fault tolerance. In USENIX Symposium on Operating Systems Design and Implementation, Vol. 99. USENIX Association, New Orleans, LA, USA, 173\u2013186."},{"key":"e_1_3_2_7_2","doi-asserted-by":"publisher","DOI":"10.1109\/TII.2022.3145010"},{"key":"e_1_3_2_8_2","first-page":"4587","volume-title":"International Conference on Machine Learning","author":"Dai Rong","year":"2022","unstructured":"Rong Dai, Li Shen, Fengxiang He, Xinmei Tian, and Dacheng Tao. 2022. DisPFL: Towards communication-efficient personalized federated learning via decentralized sparse training. In International Conference on Machine Learning. PMLR, Baltimore, MD, USA, 4587\u20134604."},{"key":"e_1_3_2_9_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"e_1_3_2_10_2","first-page":"2611","volume-title":"International Conference on Machine Learning","author":"Dennis Don Kurian","year":"2021","unstructured":"Don Kurian Dennis, Tian Li, and Virginia Smith. 2021. Heterogeneity for the win: One-shot federated clustering. In International Conference on Machine Learning. PMLR, Virtual Conference, 2611\u20132620."},{"key":"e_1_3_2_11_2","first-page":"1","volume-title":"International Conference on Learning Representations","author":"Dhillon Guneet Singh","year":"2020","unstructured":"Guneet Singh Dhillon, Pratik Chaudhari, Avinash Ravichandran, and Stefano Soatto. 2020. A baseline for few-shot image classification. In International Conference on Learning Representations. OpenReview.net, Virtual Conference, 1\u201320. Retrieved from https:\/\/openreview.net\/forum?id=rylXBkrYDS"},{"key":"e_1_3_2_12_2","doi-asserted-by":"publisher","DOI":"10.1109\/TII.2022.3192297"},{"key":"e_1_3_2_13_2","first-page":"3557","volume-title":"Advances in Neural Information Processing Systems","author":"Fallah Alireza","year":"2020","unstructured":"Alireza Fallah, Aryan Mokhtari, and Asuman Ozdaglar. 2020. Personalized federated learning with theoretical guarantees: A model-agnostic meta-learning approach. In Advances in Neural Information Processing Systems, Vol. 33. NeurIPS Foundation, Vancouver, BC, Canada, 3557\u20133568."},{"key":"e_1_3_2_14_2","doi-asserted-by":"publisher","DOI":"10.1088\/1361-6560\/ac97d9"},{"key":"e_1_3_2_15_2","first-page":"1050","volume-title":"International Conference on Machine Learning","author":"Gal Yarin","year":"2016","unstructured":"Yarin Gal and Zoubin Ghahramani. 2016. Dropout as a bayesian approximation: Representing model uncertainty in deep learning. In International Conference on Machine Learning. PMLR, New York City, NY, USA, 1050\u20131059."},{"key":"e_1_3_2_16_2","doi-asserted-by":"publisher","DOI":"10.1145\/3543873.3587583"},{"key":"e_1_3_2_17_2","first-page":"19586","volume-title":"Advances in Neural Information Processing Systems","author":"Ghosh Avishek","year":"2020","unstructured":"Avishek Ghosh, Jichan Chung, Dong Yin, and Kannan Ramchandran. 2020. An efficient framework for clustered federated learning. In Advances in Neural Information Processing Systems, Vol. 33. NeurIPS Foundation, Vancouver, BC, Canada, 19586\u201319597."},{"key":"e_1_3_2_18_2","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v37i12.26732"},{"key":"e_1_3_2_19_2","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM53939.2023.10229027"},{"key":"e_1_3_2_20_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-020-2649-2"},{"key":"e_1_3_2_21_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"e_1_3_2_22_2","unstructured":"Tzu-Ming Harry Hsu Hang Qi and Matthew Brown. 2019. Measuring the effects of non-identical data distribution for federated visual classification. (2019). arXiv:1909.06335. Retrieved from https:\/\/arxiv.org\/abs\/1909.06335"},{"key":"e_1_3_2_23_2","doi-asserted-by":"publisher","DOI":"10.1007\/s102070100002"},{"key":"e_1_3_2_24_2","doi-asserted-by":"publisher","DOI":"10.1109\/MWC.001.2100255"},{"key":"e_1_3_2_25_2","doi-asserted-by":"publisher","DOI":"10.1142\/12037"},{"key":"e_1_3_2_26_2","volume-title":"Learning Multiple Layers of Features from Tiny Images","author":"Krizhevsky Alex","year":"2009","unstructured":"Alex Krizhevsky and Geoffrey Hinton. 2009. Learning Multiple Layers of Features from Tiny Images. Technical Report TR-2009. University of Toronto, Toronto, ON, Canada. Retrieved from https:\/\/www.cs.toronto.edu\/kriz\/learning-features-2009-TR.pdfTechnical Report."},{"key":"e_1_3_2_27_2","first-page":"957","volume-title":"International Conference on Machine Learning","author":"Kusner Matt","year":"2015","unstructured":"Matt Kusner, Yu Sun, Nicholas Kolkin, and Kilian Weinberger. 2015. From word embeddings to document distances. In International Conference on Machine Learning. PMLR, Lille, France, 957\u2013966."},{"key":"e_1_3_2_28_2","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v35i18.18013"},{"key":"e_1_3_2_29_2","doi-asserted-by":"publisher","DOI":"10.1109\/5.726791"},{"key":"e_1_3_2_30_2","first-page":"6357","volume-title":"International Conference on Machine Learning","author":"Li Tian","year":"2021","unstructured":"Tian Li, Shengyuan Hu, Ahmad Beirami, and Virginia Smith. 2021. Ditto: Fair and robust federated learning through personalization. In International Conference on Machine Learning. PMLR, Virtual Conference, 6357\u20136368."},{"key":"e_1_3_2_31_2","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2020.3042392"},{"key":"e_1_3_2_32_2","first-page":"6028","volume-title":"International Conference on Machine Learning","author":"Liang Jian","year":"2020","unstructured":"Jian Liang, Dapeng Hu, and Jiashi Feng. 2020. Do we really need to access the source data? Source hypothesis transfer for unsupervised domain adaptation. In International Conference on Machine Learning. PMLR, Virtual Conference, 6028\u20136039."},{"key":"e_1_3_2_33_2","doi-asserted-by":"publisher","DOI":"10.1109\/TNSE.2022.3168025"},{"key":"e_1_3_2_34_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01170"},{"key":"e_1_3_2_35_2","first-page":"3384","volume-title":"International Conference on Machine Learning","author":"Madras David","year":"2018","unstructured":"David Madras, Elliot Creager, Toniann Pitassi, and Richard Zemel. 2018. Learning adversarially fair and transferable representations. In International Conference on Machine Learning. PMLR, Stockholm, Sweden, 3384\u20133393."},{"key":"e_1_3_2_36_2","first-page":"139","article-title":"One-class SVMs for document classification","volume":"2","author":"Manevitz Larry M","year":"2001","unstructured":"Larry M Manevitz and Malik Yousef. 2001. One-class SVMs for document classification. Journal of Machine Learning Research 2, Dec (2001), 139\u2013154.","journal-title":"Journal of Machine Learning Research"},{"key":"e_1_3_2_37_2","first-page":"1273","volume-title":"Artificial Intelligence and Statistics","author":"McMahan Brendan","year":"2017","unstructured":"Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, and Blaise Aguera y Arcas. 2017. Communication-efficient learning of deep networks from decentralized data. In Artificial Intelligence and Statistics. PMLR, Fort Lauderdale, FL, USA, 1273\u20131282."},{"key":"e_1_3_2_38_2","doi-asserted-by":"publisher","DOI":"10.1145\/3491101.3519779"},{"key":"e_1_3_2_39_2","first-page":"16888","volume-title":"International Conference on Machine Learning","author":"Niu Shuaicheng","year":"2022","unstructured":"Shuaicheng Niu, Jiaxiang Wu, Yifan Zhang, Yaofo Chen, Shijian Zheng, Peilin Zhao, and Mingkui Tan. 2022. Efficient test-time model adaptation without forgetting. In International Conference on Machine Learning. PMLR, Baltimore, MD, USA, 16888\u201316905."},{"key":"e_1_3_2_40_2","first-page":"305","volume-title":"USENIX Annual Technical Conference","author":"Ongaro Diego","year":"2014","unstructured":"Diego Ongaro and John Ousterhout. 2014. In search of an understandable consensus algorithm. In USENIX Annual Technical Conference. USENIX Association, Philadelphia, PA, USA, 305\u2013319."},{"key":"e_1_3_2_41_2","doi-asserted-by":"publisher","DOI":"10.23919\/IFIPNetworking52078.2021.9472790"},{"key":"e_1_3_2_42_2","first-page":"8026","volume-title":"Advances in Neural Information Processing Systems","author":"Paszke Adam","year":"2019","unstructured":"Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et\u00a0al. 2019. Pytorch: An imperative style, high-performance deep learning library. In Advances in Neural Information Processing Systems, Vol. 32. NeurIPS Foundation, Vancouver, BC, Canada, 8026\u20138037."},{"key":"e_1_3_2_43_2","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2019.00144"},{"key":"e_1_3_2_44_2","doi-asserted-by":"publisher","DOI":"10.1002\/widm.1249"},{"key":"e_1_3_2_45_2","doi-asserted-by":"publisher","DOI":"10.1109\/MIS.2006.62"},{"key":"e_1_3_2_46_2","doi-asserted-by":"publisher","DOI":"10.1002\/j.1538-7305.1948.tb01338.x"},{"key":"e_1_3_2_47_2","doi-asserted-by":"publisher","DOI":"10.1145\/3657284"},{"key":"e_1_3_2_48_2","first-page":"31269","volume-title":"International Conference on Machine Learning","author":"Shi Yifan","year":"2023","unstructured":"Yifan Shi, Li Shen, Kang Wei, Yan Sun, Bo Yuan, Xueqian Wang, and Dacheng Tao. 2023. Improving the model consistency of decentralized federated learning. In International Conference on Machine Learning. PMLR, Honolulu, HI, USA, 31269\u201331291."},{"key":"e_1_3_2_49_2","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2022.3196503"},{"key":"e_1_3_2_50_2","first-page":"21394","volume-title":"Advances in Neural Information Processing Systems","author":"Dinh Canh T","year":"2020","unstructured":"Canh T Dinh, Nguyen Tran, and Josh Nguyen. 2020. Personalized federated learning with moreau envelopes. In Advances in Neural Information Processing Systems, Vol. 33. NeurIPS Foundation, Vancouver, BC, Canada, 21394\u201321405."},{"key":"e_1_3_2_51_2","doi-asserted-by":"publisher","DOI":"10.1109\/TNNLS.2022.3160699"},{"key":"e_1_3_2_52_2","article-title":"TorchVision: PyTorch\u2019s Computer Vision library","author":"contributors TorchVision maintainers and","year":"2016","unstructured":"TorchVision maintainers and contributors. 2016. TorchVision: PyTorch\u2019s Computer Vision library. Retrieved from https:\/\/github.com\/pytorch\/vision. (2016). GitHub repository.","journal-title":"https:\/\/github.com\/pytorch\/vision"},{"key":"e_1_3_2_53_2","first-page":"1","volume-title":"International Conference on Learning Representations","author":"Wang Dequan","year":"2021","unstructured":"Dequan Wang, Evan Shelhamer, Shaoteng Liu, Bruno Olshausen, and Trevor Darrell. 2021. Tent: Fully test-time adaptation by entropy minimization. In International Conference on Learning Representations. OpenReview.net, Virtual Conference, 1\u201315. Retrieved from https:\/\/openreview.net\/forum?id=uXl3bZLkr3c"},{"key":"e_1_3_2_54_2","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM41043.2020.9155494"},{"key":"e_1_3_2_55_2","doi-asserted-by":"crossref","unstructured":"Jianyu Wang Anit Kumar Sahu Gauri Joshi and Soummya Kar. 2022. Matcha: A matching-based link scheduling strategy to speed up distributed optimization. IEEE Transactions on Signal Processing 70 (2022) 5208\u20135221. https:\/\/ieeexplore.ieee.org\/document\/9944194","DOI":"10.1109\/TSP.2022.3212536"},{"key":"e_1_3_2_56_2","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-021-03583-3"},{"issue":"2014","key":"e_1_3_2_57_2","first-page":"1","article-title":"Ethereum: A secure decentralised generalised transaction ledger","volume":"151","author":"Wood Gavin","year":"2014","unstructured":"Gavin Wood et\u00a0al. 2014. Ethereum: A secure decentralised generalised transaction ledger. Ethereum Project Yellow Paper 151, 2014 (2014), 1\u201332.","journal-title":"Ethereum Project Yellow Paper"},{"key":"e_1_3_2_58_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPRW59228.2023.00297"},{"key":"e_1_3_2_59_2","unstructured":"Maofan Yin Dahlia Malkhi Michael K Reiter Guy Golan Gueta and Ittai Abraham. 2018. HotStuff: BFT consensus in the lens of blockchain. (2018). arXiv:1803.05069. Retrieved from https:\/\/arxiv.org\/abs\/1803.05069"},{"key":"e_1_3_2_60_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPRW59228.2023.00553"},{"key":"e_1_3_2_61_2","doi-asserted-by":"publisher","DOI":"10.1109\/JIOT.2024.3407584"},{"key":"e_1_3_2_62_2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52729.2023.01528"},{"key":"e_1_3_2_63_2","first-page":"7252","volume-title":"International Conference on Machine Learning","author":"Yurochkin Mikhail","year":"2019","unstructured":"Mikhail Yurochkin, Mayank Agarwal, Soumya Ghosh, Kristjan Greenewald, Nghia Hoang, and Yasaman Khazaeni. 2019. Bayesian nonparametric federated learning of neural networks. In International Conference on Machine Learning. PMLR, Long Beach, CA, USA, 7252\u20137261."},{"key":"e_1_3_2_64_2","first-page":"27479","volume-title":"International Conference on Machine Learning","author":"Zhu Tongtian","year":"2022","unstructured":"Tongtian Zhu, Fengxiang He, Lan Zhang, Zhengyang Niu, Mingli Song, and Dacheng Tao. 2022. Topology-aware generalization of decentralized sgd. In International Conference on Machine Learning. PMLR, Baltimore, MD, USA, 27479\u201327503."},{"key":"e_1_3_2_65_2","first-page":"3283","volume-title":"International Conference on International Joint Conferences on Artificial Intelligence","author":"Zhuang Yuan","year":"2021","unstructured":"Yuan Zhuang, Zhenguang Liu, Peng Qian, Qi Liu, Xiang Wang, and Qinming He. 2021. Smart contract vulnerability detection using graph neural networks. In International Conference on International Joint Conferences on Artificial Intelligence. IJCAI Organization, Montr\u00e9al, QC, Canada, 3283\u20133290."}],"container-title":["ACM Transactions on the Web"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3764936","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,5]],"date-time":"2026-03-05T22:18:39Z","timestamp":1772749119000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3764936"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,2,18]]},"references-count":64,"journal-issue":{"issue":"1","published-print":{"date-parts":[[2026,2,28]]}},"alternative-id":["10.1145\/3764936"],"URL":"https:\/\/doi.org\/10.1145\/3764936","relation":{},"ISSN":["1559-1131","1559-114X"],"issn-type":[{"value":"1559-1131","type":"print"},{"value":"1559-114X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,2,18]]},"assertion":[{"value":"2025-02-15","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-08-10","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-02-18","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}