Towards Fair Graph Federated Learning via Incentive Mechanisms

Authors: Chenglu Pan, Jiarong Xu, Yue Yu, Ziqi Yang, Qingbiao Wu, Chunping Wang, Lei Chen, Yang Yang

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

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments show that our model achieves the best trade-off between accuracy and the fairness of model gradient, as well as superior payoff fairness.
Researcher Affiliation Collaboration 1Zhejiang University 2Fudan University 3ZJU-Hangzhou Global Scientific and Technological Innovation Center 4Fin Volution Group
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide an explicit statement or link confirming the release of open-source code for the described methodology.
Open Datasets Yes We use three graph classification datasets: PROTEINS, DD, and IMDB-BINARY.
Dataset Splits No The paper states: 'We retain 10% of all the graphs as the global test set for the server, and the remaining graphs are distributed to 10 agents. In each agent, we randomly split 90% for training and 10% for testing.' No explicit validation split information is provided for local agent data.
Hardware Specification No The paper does not provide specific hardware details (e.g., exact GPU/CPU models, memory amounts) used for running its experiments.
Software Dependencies No The paper mentions using a GIN network and Adam optimizer, and a motif extraction method from a cited paper, but does not provide specific version numbers for any software dependencies or libraries.
Experiment Setup Yes We set the parameters β1 and β2 in Eq. (10) as 0.05 and 1, the parameter λ in Eq. (14) as 0.1, β in Eq. (4) as 1, and the budget B of payoff as 1... We utilized a three-layer GIN network with a hidden size of 64 and a dropout rate of 0.5... An Adam optimizer with a learning rate of 0.001 and weight decay of 5e 4 is employed. The communication round is 200, the epoch of local training on agents is 1 and the batch size is 128.