AI Paper Atlas

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

Lightweight Federated Incremental Learning via Decoupled Replay

Xiuying Wang, Yichen Li, Hang Su, Gaozhuo Liu, Shiwei Li, Chuang Zhao, Jiangming Shi, Imran Razzak

ICML 2026 · Federated learning and privacy › Personalized and heterogeneous FL · approach: Federated prompt and LoRA tuning

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
add mixup to uploaded features (ablation)Li-FIL · CIFAR100 (α=1.0)A(f): 33.20 -> 35.61Improved
add mixup to uploaded features (ablation)Li-FIL · CIFAR10 (α=1.0)A(f): 39.61 -> 40.29Improved
add differential privacy Laplace noise (ablation)Li-FIL · CIFAR10 (α=1.0)A(f): 40.25 -> 40.29No clear effect
add differential privacy Laplace noise (ablation)Li-FIL · CIFAR100 (α=1.0)A(f): 35.49 -> 35.61No clear effect
add classifier replay loss on synthetic features (ablation)Li-FIL · CIFAR100 (α=1.0)A(f): 32.97 -> 35.61Improved
add MMD global feature alignment loss (ablation)Li-FIL · CIFAR100 (α=1.0)A(f): 33.09 -> 35.61Improved
add contrastive local-structure alignment loss (ablation)Li-FIL · CIFAR100 (α=1.0)A(f): 31.52 -> 35.61Improved
add contrastive local-structure alignment loss (ablation)Li-FIL · CIFAR10 (α=1.0)average accuracy: 60.92 -> 62.56Improved
apply mixup and DP together instead of raw featuresuploaded features · CIFAR10 feature inversion attackattack accuracy: 69.24 -> 29.3Improved
use mixup alone instead of raw featuresuploaded features · CIFAR10 feature inversion attackattack accuracy: 69.24 -> 55.7Improved
use feature replay with CVAE instead of buffer-based ReFed+federated incremental learning · CIFAR100 (α=1.0)A(f): 32.15 -> 35.61Improved
replace image-level CVAE input with ResNet18 featuresserver generator · communication and model sizetransmission cost / CVAE parameters: 3,072 -> 512 values; 3.8M -> 0.17M parametersImproved

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