AI Paper Atlas

Research map / Federated learning and privacy / Personalized and heterogeneous FL

LEGO-FL: Learning Heterogeneous Federated Models as a LEGO Assembly Games

Zeqi Leng, Chunxu Zhang, Guodong Long, Bo Yang

ICML 2026 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Non-IID federated learning

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
fix the root anchor to the shallowest semantic block (instead of random) (ablation)LEGO-FL · CIFAR-10 Non-IIDaccuracy (%): 75.55 -> 79.12Improved
use ordered-constraint tree search (instead of unconstrained random search) (ablation)LEGO-FL · CIFAR-10 Non-IIDaccuracy (%): 73.33 -> 79.12Improved
add dynamic pool-block completion (ablation)LEGO-FL · CIFAR-10 Non-IIDaccuracy (%): 74.20 -> 79.12Improved
complete over global unused pool (instead of ordered-constraint scope only) (ablation)LEGO-FL · CIFAR-10 Non-IIDaccuracy (%): 71.33 -> 79.12Improved
use global consensus architecture selection (instead of top-N direct matching) (ablation)LEGO-FL · CIFAR-10 Non-IIDaccuracy (%): 72.26 -> 79.12Improved
use NASWOT zero-cost proxy (instead of Jacov) (ablation)LEGO-FL · CIFAR-10 Non-IIDaccuracy (%): 79.00 -> 79.12Improved
use NASWOT zero-cost proxy (instead of Fisher) (ablation)LEGO-FL · CIFAR-10 Non-IIDaccuracy (%): 77.77 -> 79.12Improved
add personalized grafting (instead of global distillation only) (ablation)LEGO-FL · CIFAR-10 Non-IIDaccuracy (%): 77.16 -> 79.12Improved
add personalized grafting (instead of global distillation only) (ablation)LEGO-FL · CIFAR-10 IIDaccuracy (%): 78.55 -> 80.60Improved
use LEGO-FL instead of pFedClubheterogeneous FL · SVHN Non-IID, Zoo 4accuracy (%): 80.55 -> 86.48Improved
use LEGO-FL instead of FedGHheterogeneous FL · CIFAR-10 Non-IID, Zoo 4accuracy (%): 63.62 -> 79.88Improved
use LEGO-FL instead of pFedClub with 100 clientsheterogeneous FL · CIFAR-10 Non-IID, N=100accuracy (%): 67.74 -> 73.10Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy