MNIST, CIFAR10, MedMNIST

本研究使用了三个公共图像数据集：MNIST、CIFAR10和MedMNIST。MNIST是一个广泛用于手写数字识别的数据集，包含60,000个训练样本和10,000个测试样本。CIFAR10是一个包含60,000个32x32彩色图像的数据集，分为10个类别。MedMNIST是一个医学图像数据集，包含多种类型的医学图像，用于深度学习和联邦学习研究。这些数据集经过预处理，包括尺寸调整和数据增强，以适应实验需求。研究主要关注联邦学习在非独立同分布（non-IID）数据上的应用，旨在解决数据隐私和安全问题，同时提高模型的泛化能力。

This study employs three public image datasets: MNIST, CIFAR10, and MedMNIST. MNIST is a widely adopted dataset for handwritten digit recognition, containing 60,000 training samples and 10,000 test samples. CIFAR10 comprises 60,000 32×32 color images divided into 10 distinct categories. MedMNIST is a medical imaging dataset encompassing diverse types of medical images, tailored for deep learning and federated learning research. All these datasets have undergone preprocessing including resizing and data augmentation to meet the experimental requirements. This research primarily focuses on the application of federated learning on non-independent and identically distributed (non-IID) data, aiming to resolve data privacy and security concerns while enhancing the generalization capability of models.