AI Paper Atlas

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

FeDaL: Federated Dataset Learning for General Time Series Foundation Models

Shengchao Chen, Guodong Long, Michael Blumenstein, Jing Jiang

ICLR 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
add Domain Bias Elimination (DBE) block (ablation)FeDaL · UTSD+CTSD representation learningreconstruction MSE (avg): 0.545 -> 0.489Improved
add Global Bias Elimination (GBE) (ablation)FeDaL · UTSD+CTSD representation learningreconstruction MSE (avg): 0.527 -> 0.489Improved
add bias alignment penalty in DBE (ablation)FeDaL · UTSD+CTSD representation learningreconstruction MSE (avg): 0.518 -> 0.489Improved
add core-set tuning on server (ablation)FeDaL · UTSD+CTSD representation learningreconstruction MSE (avg): 0.510 -> 0.489Improved
add gradient-level dynamic correction (ablation)FeDaL · UTSD+CTSD representation learningreconstruction MSE (avg): 0.516 -> 0.489Improved
overly large alignment weight lambda (ablation)FeDaL · UTSDreconstruction MSEHurt
larger core-set size (ablation)FeDaL · UTSDreconstruction MSENo clear effect
extreme fusion weight alpha (ablation)FeDaL · UTSDreconstruction MSEHurt
FeDaL vs FFTS federated pretrainingfederated TSFM pretraining · UTSD and CTSDreconstruction MSE: 4.16% on USTD and 8.86% on CTSDImproved
transmit core-sets to serverFeDaL · communication per roundcommunication parameters: 108.41 MB (FedAvg) -> 110.41 MBMixed
increase pretraining data size from 40B to 231B points (ablation)FeDaL · federated scaling (174 clients)downstream performanceImproved
increase client count from 30 to 174 at fixed total data (ablation)FeDaL · federated scalingdownstream performanceImproved

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