Papers › Improving Generalization in Federated Learning by Seeking Flat Minima

Improving Generalization in Federated Learning by Seeking Flat Minima

22 Mar 2022arXiv:2203.11834archive 2025-07-28

Debora Caldarola, Barbara Caputo, Marco Ciccone

Models trained in federated settings often suffer from degraded performances and fail at generalizing, especially when facing heterogeneous scenarios. In this work, we investigate such behavior through the lens of geometry of the loss and Hessian eigenspectrum, linking the model's lack of generalization capacity to the sharpness of the solution. Motivated by prior studies connecting the sharpness of the loss surface and the generalization gap, we show that i) training clients locally with Sharpness-Aware Minimization (SAM) or its adaptive version (ASAM) and ii) averaging stochastic weights (SWA) on the server-side can substantially improve generalization in Federated Learning and help bridging the gap with centralized models. By seeking parameters in neighborhoods having uniform low loss, the model converges towards flatter minima and its generalization significantly improves in both homogeneous and heterogeneous scenarios. Empirical results demonstrate the effectiveness of those optimizers across a variety of benchmark vision datasets (e.g. CIFAR10/100, Landmarks-User-160k, IDDA) and tasks (large scale classification, semantic segmentation, domain generalization).

PaperPDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 1 code sample harvested from 1 repository linked to this paper: 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified. Where Syntology's instrument failed: 1 where our contract draft was wrong; failures of Syntology's instrument, not of the code. Every run on this page is a failure of Syntology's instrument, not of the code.

By repository: official repository: 1 sample from 1 repository, 1 ran (0 with no instrument failure; 1 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written on 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

debcaldarola/fedsam officialmentioned in papermentioned on GitHubpytorchBSD-2-ClauseSyntology: 1 sample, 1 ran (1 instrument-failed) · 0 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

1 sample harvested: 1 ran (of which 0 constructed an object rather than computing a result; 0 with no instrument failure: 0 honoured, 0 violated, 0 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Every run on this page is a failure of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · our draft was wrong

Licence: 0 of the 1 sample are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from debcaldarola/fedsam. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: 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. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

debcaldarola/fedsam official repository 1 sample, 1 ran (1 instrument-failed) · 0 unverified
get_metrics_names models/metrics/writer.py ran · our draft was wrong BSD-2-Clause (permissive) · 7690c4d309642d3f · report fixture: signature metrics:dict/0/? · returned list · level recorded 2026-09-24

Tasks

Domain GeneralizationFederated LearningImage ClassificationSemantic Segmentation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Federated Learning CIFAR-100 (alpha=0, 10 clients per round) FedASAM + SWA ACC@1-100Clients 42.64 #1 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 10 clients per round) FedASAM ACC@1-100Clients 39.76 #2 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 10 clients per round) FedSAM + SWA ACC@1-100Clients 39.51 #3 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 10 clients per round) FedSAM ACC@1-100Clients 36.93 #4 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 10 clients per round) FedAvg ACC@1-100Clients 36.74 #5 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 20 clients per round) FedASAM + SWA ACC@1-100Clients 41.62 #1 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 20 clients per round) FedASAM ACC@1-100Clients 40.81 #2 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 20 clients per round) FedSAM + SWA ACC@1-100Clients 39.24 #3 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 20 clients per round) FedAvg ACC@1-100Clients 38.59 #4 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 20 clients per round) FedSAM ACC@1-100Clients 38.56 #5 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 5 clients per round) FedASAM + SWA ACC@1-100Clients 42.01 #1 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 5 clients per round) FedSAM + SWA ACC@1-100Clients 39.3 #2 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 5 clients per round) FedASAM ACC@1-100Clients 36.04 #3 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 5 clients per round) FedSAM ACC@1-100Clients 31.04 #4 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0, 5 clients per round) FedAvg ACC@1-100Clients 30.25 #5 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 10 clients per round) FedASAM + SWA ACC@1-100Clients 48.72 #1 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 10 clients per round) FedSAM + SWA ACC@1-100Clients 46.76 #2 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 10 clients per round) FedASAM ACC@1-100Clients 46.58 #3 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 10 clients per round) FedSAM ACC@1-100Clients 44.84 #4 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 10 clients per round) FedAvg ACC@1-100Clients 41.27 #5 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 20 clients per round) FedASAM + SWA ACC@1-100Clients 48.27 #1 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 20 clients per round) FedASAM ACC@1-100Clients 47.78 #2 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 20 clients per round) FedSAM + SWA ACC@1-100Clients 46.47 #3 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 20 clients per round) FedSAM ACC@1-100Clients 46.05 #4 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 20 clients per round) FedAvg ACC@1-100Clients 42.17 #5 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 5 clients per round) FedASAM + SWA ACC@1-100Clients 49.17 #1 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 5 clients per round) FedSAM + SWA ACC@1-100Clients 47.96 #2 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 5 clients per round) FedASAM ACC@1-100Clients 45.61 #3 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 5 clients per round) FedSAM ACC@1-100Clients 44.73 #4 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=0.5, 5 clients per round) FedAvg ACC@1-100Clients 40.43 #5 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 10 clients per round) FedASAM ACC@1-100Clients 54.97 #1 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 10 clients per round) FedASAM + SWA ACC@1-100Clients 54.79 #2 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 10 clients per round) FedSAM + SWA ACC@1-100Clients 53.67 #3 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 10 clients per round) FedSAM ACC@1-100Clients 53.39 #4 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 10 clients per round) FedAvg ACC@1-100Clients 50.25 #5 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 20 clients per round) FedASAM ACC@1-100Clients 54.5 #1 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 20 clients per round) FedSAM + SWA ACC@1-100Clients 54.36 #2 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 20 clients per round) FedASAM + SWA ACC@1-100Clients 54.1 #3 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 20 clients per round) FedSAM ACC@1-100Clients 53.97 #4 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 20 clients per round) FedAvg ACC@1-100Clients 50.66 #5 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 5 clients per round) FedASAM ACC@1-100Clients 54.81 #1 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 5 clients per round) FedSAM ACC@1-100Clients 54.01 #2 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 5 clients per round) FedSAM + SWA ACC@1-100Clients 53.9 #3 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 5 clients per round) FedASAM + SWA ACC@1-100Clients 53.86 #4 of 5 Archive leaderboard report
Federated Learning CIFAR-100 (alpha=1000, 5 clients per round) FedAvg ACC@1-100Clients 49.92 #5 of 5 Archive leaderboard report
Federated Learning Cityscapes heterogeneous SiloBN + ASAM mIoU 49.75 #1 of 9 Archive leaderboard report
Federated Learning Cityscapes heterogeneous SiloBN + SAM mIoU 49.1 #2 of 9 Archive leaderboard report
Federated Learning Cityscapes heterogeneous SiloBN mIoU 45.96 #3 of 9 Archive leaderboard report
Federated Learning Cityscapes heterogeneous FedSAM + SWA mIoU 43.42 #4 of 9 Archive leaderboard report
Federated Learning Cityscapes heterogeneous FedASAM + SWA mIoU 43.02 #5 of 9 Archive leaderboard report
Federated Learning Cityscapes heterogeneous FedAvg + SWA mIoU 42.48 #6 of 9 Archive leaderboard report
Federated Learning Cityscapes heterogeneous FedASAM mIoU 42.27 #7 of 9 Archive leaderboard report
Federated Learning Cityscapes heterogeneous FedSAM mIoU 41.22 #8 of 9 Archive leaderboard report
Federated Learning Cityscapes heterogeneous FedAvg mIoU 38.65 #9 of 9 Archive leaderboard report
Federated Learning Landmarks-User-160k FedASAM + SWA Acc@1-1262Clients 68.32 #1 of 6 Archive leaderboard report
Federated Learning Landmarks-User-160k FedSAM + SWA Acc@1-1262Clients 68.12 #2 of 6 Archive leaderboard report
Federated Learning Landmarks-User-160k FedAvg + SWA Acc@1-1262Clients 67.52 #3 of 6 Archive leaderboard report
Federated Learning Landmarks-User-160k FedASAM Acc@1-1262Clients 64.23 #4 of 6 Archive leaderboard report
Federated Learning Landmarks-User-160k FedSAM Acc@1-1262Clients 63.72 #5 of 6 Archive leaderboard report
Federated Learning Landmarks-User-160k FedAvg Acc@1-1262Clients 61.91 #6 of 6 Archive leaderboard report
Image Classification CIFAR-100 (alpha=0, 20 clients per round) FedAvgM + ASAM + SWA ACC@1-100Clients 51.58 #1 of 1 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Sharpness-Aware MinimizationStochastic Weight Averaging

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections