Federated Continual Learning Goes Online (Online-FCL)

The repository contains the code to reproduce the results of the paper "Federated Continual Learning Goes Online: Uncertainty-Aware Memory Management for Vision Tasks and Beyond".

Requirements

The codebase is implemented in Python 3.12.2 and tested on Linux OS.

• pytorch==2.2.2
• torchvision==0.17.2
• transformers==4.40.2
• numpy==1.26.4
• scikit-learn==1.4.1
• medmnist==3.0.1
• matplotlib==3.8.0

Structure of the repo

• utils contains all the utilities to preprocess and save the data, to run the experiments, and to save the results.
• models contains the model architectures of our evaluation framework.
• configuration contains the hyperparameters (dataset, memory population strategy, memory sampling strategy, memory size, etc.) to run our experiments.

Run the experiments

The current version of the configuration file (configuration/config_jup.py) enables running an experiment for five different clients on CIFAR10 over three runs. All the parameters can be changed as desired in the main_OFCL.ipynb file. The most important hyperparameters are the followings.

For the continual learning and memory management part:

• --local_epochs: to perform multiple gradient updates per mini-batch (number of local epochs).
• --memory_size: to set the size of the memory buffer.
• --dataset_name: name of the dataset to evaluate.
• --update_strategy: to select whether we want to use reservoir (reservoir) or class-balanced update (balanced).
• --balanced_update: if balanced is used, we can select whether we want to populate the memory randomly (random) or use the uncertainty (uncertainty).
• --uncertainty_score: if uncertainty is used, we can decide the uncertainty score we want to use among the available ones (bregman, confidence, margin, entropy, rainbow, ratio).
• --balanced_step: if uncertainty is used, we can decide the sorting strategy, i.e., bottomk (bottom-k), step (step-sized), and topk (top-k).

For the federated learning part:

• --overlap: to set if the clients works on overlapping tasks (overlap (default)) or not (non-overlap).
• --burnin: number of epochs before a client contributes to the communication rounds.
• --jump: number of epochs to skip before a client contributes to the communication rounds.
• --fl_update: FedAvg (favg), weighted FedAvg (w_favg).