AFL derives closed-form solutions for federated learning on pre-trained models that enable single-round aggregation and produce identical results regardless of how data is partitioned across clients.
Distributed distillation for on-device learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
AFL: A Single-Round Analytic Approach for Federated Learning with Pre-trained Models
AFL derives closed-form solutions for federated learning on pre-trained models that enable single-round aggregation and produce identical results regardless of how data is partitioned across clients.