Combating Data Imbalances in Federated Semi-supervised Learning with Dual Regulators
Authors: Sikai Bai, Shuaicheng Li, Weiming Zhuang, Jie Zhang, Kunlin Yang, Jun Hou, Shuai Yi, Shuai Zhang, Junyu Gao
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Empirically, we demonstrate that Fed Dure is superior to the existing methods across a wide range of settings, notably by more than 11% on CIFAR-10 and CINIC-10 datasets. |
| Researcher Affiliation | Collaboration | Sikai Bai1*, Shuaicheng Li2* , Weiming Zhuang3 , Jie Zhang4 , Kunlin Yang2, Jun Hou2 , Shuai Zhang2, Shuai Yi2, Junyu Gao5 1The Hong Kong University of Science and Technology 2Sense Time Research 3Sony AI 4The Hong Kong Polytechnic University 5Northwestern Polytechnical University |
| Pseudocode | No | The paper describes the optimization process and various equations, and it refers to "the pipeline of the overall optimization process in the supplementary material about the proof for analysis." However, there is no pseudocode or algorithm block labeled or formatted as such within the main paper content. |
| Open Source Code | No | The paper does not contain any explicit statements about releasing source code for the described methodology or provide a link to a code repository. |
| Open Datasets | Yes | We conduct comprehensive experiments on three datasets, including CIFAR-10 (Krizhevsky, Hinton et al. 2009), Fashion-MNIST (Xiao, Rasul, and Vollgraf 2017) and CINIC-10 (Darlow et al. 2018). |
| Dataset Splits | No | The paper states "All datasets are split according to official guidelines; we provide more dataset descriptions and split strategies in the supplementary material." While it mentions "official guidelines," it does not specify the exact split percentages or sample counts for training, validation, or test sets within the main text. |
| Hardware Specification | No | The paper does not specify any particular hardware (e.g., CPU, GPU models, or cloud computing instances) used for running the experiments. It only details software settings and hyperparameters. |
| Software Dependencies | No | The paper mentions "We use the Adam optimizer" and discusses network architectures (Res Net-9, MLP), but it does not specify any software dependencies (e.g., Python, PyTorch, TensorFlow) with their version numbers. |
| Experiment Setup | Yes | We use the Adam optimizer with momentum = 0.9, batch size = 10 and learning rates = 0.0005 for ηs, η and ηw. If there is no specified description, our default settings also include local iterations T = 1, the selected clients in each round S = 5, and the number of clients K = 100. For the DIR data configuration, we use a Dirichlet distribution Dir(γ) to generate the DIR data for all clients, where γ = 0.5 for all three datasets. We adopt the Res Net-9 network as the default backbone architecture for local models and the coarse-grained regulator, while an MLP is utilized for the fine-grained regulator. |