HOSL reduces client memory up to 3.7x versus full first-order split learning while staying within 0.20-4.23% accuracy on OPT models by pairing client zeroth-order estimation with server first-order optimization.
Efficient parallel split learning over resource-constrained wireless edge networks
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
verdicts
UNVERDICTED 3roles
background 1polarities
background 1representative citing papers
Graph representation learning plus iterative augmented Lagrangian optimization creates stronger, harder-to-detect model manipulation attacks on federated LLM fine-tuning, cutting global accuracy by up to 26%.
The paper derives the first convergence upper bound for split federated learning under activation upload, gradient download, and aggregation failures, then jointly optimizes client sampling and model splitting to minimize the bound, with simulations on EMNIST and CIFAR-10.
citing papers explorer
-
HOSL: Hybrid-Order Split Learning for Memory-Constrained Edge Training
HOSL reduces client memory up to 3.7x versus full first-order split learning while staying within 0.20-4.23% accuracy on OPT models by pairing client zeroth-order estimation with server first-order optimization.
-
Graph Representation Learning Augmented Model Manipulation on Federated Fine-Tuning of LLMs
Graph representation learning plus iterative augmented Lagrangian optimization creates stronger, harder-to-detect model manipulation attacks on federated LLM fine-tuning, cutting global accuracy by up to 26%.
-
Optimizing Split Federated Learning with Unstable Client Participation
The paper derives the first convergence upper bound for split federated learning under activation upload, gradient download, and aggregation failures, then jointly optimizes client sampling and model splitting to minimize the bound, with simulations on EMNIST and CIFAR-10.