fl_sim.data_processing

This module contains federated datasets and data processing utilities.

Base classes

FedDataset()	Base class for all federated datasets.
FedVisionDataset([datadir, transform, seed])	Base class for all federated vision datasets.
FedNLPDataset([datadir, seed])	Base class for all federated NLP datasets.

Vision datasets

FedCIFAR([n_class, datadir, transform, seed])	Federated CIFAR10/100 dataset.
FedCIFAR100([datadir, transform, seed])
FedEMNIST([datadir, transform, seed])	Federated EMNIST dataset.
FedMNIST([datadir, transform, seed])	MNIST is a dataset to study image classification of handwritten digits 0-9.
FedRotatedCIFAR10([datadir, num_rotations, ...])	CIFAR10 dataset with rotation augmentation.
FedRotatedMNIST([datadir, num_rotations, ...])	MNIST dataset with rotation augmentation.
FedProxFEMNIST([datadir, transform, seed])	A subset of the federeated EMNIST proposed in FedProx.
FedProxMNIST([datadir, transform, seed])	Federeated MNIST proposed in FedProx.
TinyImageNet([datadir, num_clients, alpha, ...])	Tiny ImageNet dataset.

NLP datasets

FedShakespeare([datadir, seed])	Federated Shakespeare dataset.
FedProxSent140([datadir, seed])	Federated Sentiment140 dataset used in FedProx paper.

Synthetic datasets

FedSynthetic(alpha, beta, iid, num_clients)

Federated synthetic dataset.

LibSVM datasets

FedLibSVMDataset(dataset_name, num_clients)

Dataset registry utilities

register_fed_dataset([name, override])	Decorator to register a new federated dataset.
list_fed_dataset()	List all registered federated datasets.
get_fed_dataset(name)	Get a registered federated dataset by name.