FedST: Federated Style Transfer Learning for Non-IID Image Segmentation
Authors: Boyuan Ma, Xiang Yin, Jing Tan, Yongfeng Chen, Haiyou Huang, Hao Wang, Weihua Xue, Xiaojuan Ban
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments prove that our method achieves superior segmentation performance compared to state-of-art methods among four different Non-IID datasets in objective and subjective assessment. |
| Researcher Affiliation | Academia | 1School of Intelligence Science and Technology, Beijing Advanced Innovation Center for Materials Genome Engineering, Shunde Innovation School, Key Laboratory of Intelligent Bionic Unmanned Systems, Institute for Advanced Materials and Technology, University of Science and Technology Beijing, Beijing 100083, China 2 School of Materials Science and Technology, Liaoning Technical University, Liaoning 114051, China 3 Institute of Materials Intelligent Technology, Liaoning Academy of Materials, Shenyang 110004, China |
| Pseudocode | No | The paper does not contain any pseudocode or clearly labeled algorithm blocks. |
| Open Source Code | Yes | The code is available at https://github.com/Yofer Chen/Fed ST. |
| Open Datasets | Yes | We conduct extensive experiments on four cross-domain Non-IID tasks... including material microscopic image segmentation, medical image segmentation (CHAOS challenge (Kavur et al. 2019)), face segmentation with different races (Celeb AMask-HQ(Lee et al. 2020)), and face segmentation with different ages (All-Age-Faces (AAF) Dataset (Cheng et al. 2019)). |
| Dataset Splits | No | The paper describes the total number of patches/slices selected for the experiments and some preprocessing steps (e.g., 'selected 892 patches from each domain for the experiment', '347 slices were chosen and resized to 384 384 pixels for the experiment'). However, it does not provide explicit training, validation, and test dataset splits with percentages, sample counts, or references to predefined splits, nor does it explicitly mention a validation set. |
| Hardware Specification | No | The paper mentions 'The computing work is supported by USTB Mat Com of Beijing Advanced Innovation Center for Materials Genome Engineering,' but does not provide specific details on the hardware used, such as GPU models, CPU types, or memory. |
| Software Dependencies | No | The paper states 'We use Py Torch(Paszke et al. 2019) to implement federate style transfer and the other baselines' but does not specify the version of PyTorch or any other software dependencies with version numbers. |
| Experiment Setup | Yes | The batch size is set to 6 in material microscopic image segmentation task and medical image segmentation task while 3 in face segmentation task. The number of local epochs is set to 1 for all approaches. The number of communication rounds is set to 50 for all four tasks where all federated learning approaches have little or no accuracy gain with more communications. The federated optimizer of the proposed Fed ST is Adam. |