FedBCa

数据集获取

git clone https://github.com/MedcAILab/FedBCa.git

curl -L -o repo.zip https://github.com/MedcAILab/FedBCa/archive/refs/heads/main.zip
unzip repo.zip

源站 README 摘录（使用方式）

The server_classifiction.py contains the paths to the image data and label files, which can be used by modifying the paths. At the same time, server_classifiction.py contains the main function and many adjustable parameters.
The provided code from train_cls.py consists of two training functions for deep learning models:
train_step_cls:

• Trains a classification model using a given training loader, optimizer, and criterion.
• Handles GPU acceleration if available.
• Displays a progress bar for each training epoch, tracking batch times and losses.
• Supports model checkpointing by returning the learning rate, epoch loss, and the model’'s state dictionary.
  train_step_cls_prox:
• Extends the previous classification training with Federated Proximal (FedProx) optimization.
• Introduces a hyperparameter prox_mu for controlling the proximal term.
• Calculates the model parameter difference (w_diff) between the local model and a server model.
• Augments the loss function with a proximal term to penalize large parameter deviations.
• Performs training with the augmented loss and updates the model parameters accordingly.
• Also displays a progress bar for tracking training progress.
  Usage:
  python server_ classifiction.py
  Dataset download link：
  https://zenodo.org/records/10409145

数据加载示例（图像类）

from PIL import Image
import glob, os

files = (glob.glob(os.path.join(path, "**", "*.png"), recursive=True)
       + glob.glob(os.path.join(path, "**", "*.jpg"), recursive=True)
       + glob.glob(os.path.join(path, "**", "*.tif"), recursive=True))
print("图像文件数:", len(files))
img = Image.open(files[0]); print("尺寸/模式:", img.size, img.mode)

# torchvision Dataset 方式:
# from torchvision import datasets
# ds = datasets.ImageFolder(path)  # 要求 子目录=类别

目录组织与标注格式以源站说明和下载后实际文件为准。

完整仓库：github.com/MedcAILab/FedBCa