MLFCIL decomposes forgetting into local, cross-task, and federation sources and mitigates them via class-reweighted loss, knowledge distillation with replay and drift compensation, class-aware aggregation, and dual-granularity coordination.
Communication-Efficient Learning of Deep Networks from Decentralized Data,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.NI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MLFCIL: A Multi-Level Forgetting Mitigation Framework for Federated Class-Incremental Learning in LEO Satellites
MLFCIL decomposes forgetting into local, cross-task, and federation sources and mitigates them via class-reweighted loss, knowledge distillation with replay and drift compensation, class-aware aggregation, and dual-granularity coordination.