AI Paper Atlas

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

FedBABU: Toward Enhanced Representation for Federated Image Classification

Jaehoon Oh, SangMook Kim, Se-Young Yun

ICLR 2022 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Sparse and collaboration-graph personalization · 49 citations (OpenAlex)

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
Train and aggregate only the body while keeping the classifier fixed across clientsFedAvg · CIFAR100, 100 clients, s=10, f=0.1, τ=10personalized accuracy: 62.67 -> 66.32Improved
Train and aggregate only the body while keeping the classifier fixed across clientsFedAvg · CIFAR100, 100 clients, s=50, f=0.1, τ=10personalized accuracy: 38.09 -> 42.63Improved
Train and aggregate only the body while keeping the classifier fixed across clientsFedAvg · CIFAR100, 100 clients, s=100, f=0.1, τ=10personalized accuracy: 33.13 -> 35.94Improved
Apply body-only updates and aggregation to FedProxFedProx · CIFAR100, 100 clients, s=10, f=0.1, τ=10personalized accuracy: 43.48 -> 67.15Improved
Use FedBABU rather than FedAvgFedAvg · CIFAR100, s=10, f=0.1, τ=10; one fine-tuning epochpersonalized accuracy: 27.43 -> 63.29Improved
Use FedBABU rather than FedAvgFedAvg · CIFAR10, 100 clients, s=10, f=0.1, τ=10personalized accuracy: 69.11 -> 76.25Improved
Fine-tune only the classifier during personalization (ablation)FedBABU · CIFAR100, s=10, f=0.1, τ=10personalized accuracy: 64.77 -> 66.28Improved
Fine-tune the full model rather than only the classifier (ablation)FedBABU · CIFAR100, s=10, f=0.1, τ=10personalized accuracy: 66.28 -> 66.32No clear effect
Fine-tune the full model rather than only the body (ablation)FedAvg · CIFAR100, s=10, f=0.1, τ=10personalized accuracy: 27.94 -> 62.67Improved
Use the client class templates in place of the trained classifier for representation evaluation (ablation)FedBABU · CIFAR100, s=10, f=0.1, τ=10initial accuracy: 18.50 -> 54.93Improved
Use the client class templates in place of the trained classifier for representation evaluation (ablation)FedAvg · CIFAR100, s=10, f=0.1, τ=10initial accuracy: 14.40 -> 50.25Improved
Use body-only federation instead of full-model federation (ablation)FedAvg with server-side shared data (p=0.10) · CIFAR100, s=10personalized accuracy: 49.62 -> 60.94Improved

Later papers that compared with this method

16 later accepted papers report FedBABU as a baseline in their results tables; 16 of them report a better number on most of the results they share with it.

8 more not listed.

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