{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/arxiv-2603-18872","title":"DriftGuard: Mitigating Asynchronous Data Drift in Federated Learning","arxiv_id":"2603.18872","date":"2026-03-19","proceeding":null,"authors":["Yizhou Han","Di Wu","Blesson Varghese"],"abstract":"In real-world Federated Learning (FL) deployments, data distributions on devices that participate in training evolve over time. This leads to asynchronous data drift, where different devices shift at different times and toward different distributions. Mitigating such drift is challenging: frequent retraining incurs high computational cost on resource-constrained devices, while infrequent retraining degrades performance on drifting devices. We propose DriftGuard, a federated continual learning framework that efficiently adapts to asynchronous data drift. DriftGuard adopts a Mixture-of-Experts (MoE) inspired architecture that separates shared parameters, which capture globally transferable knowledge, from local parameters that adapt to group-specific distributions. This design enables two complementary retraining strategies: (i) global retraining, which updates the shared parameters when system-wide drift is identified, and (ii) group retraining, which selectively updates local parameters for clusters of devices identified via MoE gating patterns, without sharing raw data. Experiments across multiple datasets and models show that DriftGuard matches or exceeds state-of-the-art accuracy while reducing total retraining cost by up to 83%. As a result, it achieves the highest accuracy per unit retraining cost, improving over the strongest baseline by up to 2.3x. DriftGuard is available for download from https://github.com/blessonvar/DriftGuard.","url_abs":"https://arxiv.org/abs/2603.18872","url_pdf":"https://arxiv.org/pdf/2603.18872","source":{"archive":null,"snapshot":"2025-07-28","note":"not in the Papers with Code archive (frozen at the snapshot)","row_kind":"graph","title_abstract_authors_date":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)"},"code_links":[],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2603.18872","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2603.18872"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"mentioned_in_github":null,"is_official":null,"provenance":"deterministic:regex_extraction","mentioned_in_paper":null,"url":"https://github.com/blessonvar/DriftGuard","reach":null}],"summary":{"unverified":6},"by_repo_kind":{"found_in_text":{"samples":6,"ran":0,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":6,"samples":[{"code_sha256_prefix":"3cd04526777d345f","entry":"get_inference_transform","repo":"blessonvar/DriftGuard","repo_kind":"found_in_text","path":"src/driftguard/model/dataset.py","file_url":"https://github.com/blessonvar/DriftGuard/blob/HEAD/src/driftguard/model/dataset.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"3cd04526777d345f"}},{"code_sha256_prefix":"553cb8aa327118fa","entry":"get_logger","repo":"blessonvar/DriftGuard","repo_kind":"found_in_text","path":"src/driftguard/config.py","file_url":"https://github.com/blessonvar/DriftGuard/blob/HEAD/src/driftguard/config.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"553cb8aa327118fa"}},{"code_sha256_prefix":"9803a48aea9238b7","entry":"get_total_params","repo":"blessonvar/DriftGuard","repo_kind":"found_in_text","path":"src/driftguard/model/utils.py","file_url":"https://github.com/blessonvar/DriftGuard/blob/HEAD/src/driftguard/model/utils.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"9803a48aea9238b7"}},{"code_sha256_prefix":"d20e65b74b644a45","entry":"get_train_transform","repo":"blessonvar/DriftGuard","repo_kind":"found_in_text","path":"src/driftguard/model/dataset.py","file_url":"https://github.com/blessonvar/DriftGuard/blob/HEAD/src/driftguard/model/dataset.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"d20e65b74b644a45"}},{"code_sha256_prefix":"b926afd6d3cfcd6d","entry":"get_trainable_params","repo":"blessonvar/DriftGuard","repo_kind":"found_in_text","path":"src/driftguard/model/utils.py","file_url":"https://github.com/blessonvar/DriftGuard/blob/HEAD/src/driftguard/model/utils.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"b926afd6d3cfcd6d"}},{"code_sha256_prefix":"121b079a948bb5d2","entry":"setup_logging","repo":"blessonvar/DriftGuard","repo_kind":"found_in_text","path":"src/driftguard/config.py","file_url":"https://github.com/blessonvar/DriftGuard/blob/HEAD/src/driftguard/config.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"121b079a948bb5d2"}}]},"arxiv_metadata":{"licence":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)","fields":["title","abstract","authors","date"],"primary_category":"cs.LG","source":"arxiv_2026.jsonl"},"syntology_extracted_results":null}