FedSSA: Semantic Similarity-based Aggregation for Efficient Model-Heterogeneous Personalized Federated Learning

Authors: Liping Yi, Han Yu, Zhuan Shi, Gang Wang, Xiaoguang Liu, Lizhen Cui, Xiaoxiao Li

IJCAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments present that Fed SSA achieves up to 3.62% higher accuracy, 15.54 times higher communication efficiency, and 15.52 times higher computational efficiency compared to 7 state-of-the-art MHPFL baselines.
Researcher Affiliation Academia Liping Yi1,2, Han Yu2, Zhuan Shi3, Gang Wang1, , Xiaoguang Liu1, Lizhen Cui4, , Xiaoxiao Li5 1College of Computer Science, TMCC, Sys Net, DISSec, GTIISC, Nankai University, Tianjin, China 2College of Computing and Data Science, Nanyang Technological University, Singapore 3Artificial Intelligence Laboratory, Ecole Polytechnique F ed erale de Lausanne (EPFL), Switzerland 4School of Software, Shandong University, Jinan, China 5Department of Electrical and Computer Engineering, The University of British Columbia, Vancouver, BC, Canada
Pseudocode Yes Fed SSA is detailed in Algorithm 1 (Appendix A 1).
Open Source Code Yes https://github.com/LipingYi/FedSSA
Open Datasets Yes We evaluate Fed SSA and baselines on two image classification datasets: CIFAR-10 and CIFAR-100 3 [Krizhevsky, 2009].
Dataset Splits Yes Then, each client s local data are further divided into the training set, the evaluation set, and the testing set following the ratio of 8:1:1.
Hardware Specification Yes baselines with Pytorch on four NVIDIA Ge Force RTX 3090 GPUs with 24G memory.
Software Dependencies No The paper mentions "Pytorch" but does not specify a version number or other software dependencies with their versions.
Experiment Setup Yes The epochs of local training E 2 {1, 10} and the batch size of local training B 2 {64, 128, 256, 512}. The optimizer for local training is SGD with learning rate = 0.01. We also tune special hyperparameters for the baselines and report the optimal results. We also adjust the hyperparameters µ0 and Tstable to achieve the best-performance Fed SSA. To compare Fed SSA with the baselines fairly, we set the total number of communication rounds T 2 {100, 500} to ensure all algorithms converge.