Entropy-guided MPS rank allocation compresses multimodal FL updates up to 56.8× on heterogeneous edge devices while improving accuracy over uncompressed FedAvg and cutting data-to-convergence by up to 66×.
Asynchronous federated optimization,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Entropy-Guided Tensor Compression for Multimodal Federated Learning on Edge Devices
Entropy-guided MPS rank allocation compresses multimodal FL updates up to 56.8× on heterogeneous edge devices while improving accuracy over uncompressed FedAvg and cutting data-to-convergence by up to 66×.