This repository contains PyTorch implementation of the paper ''LFighter: Defending against Label-flipping Attacks in Federated Learning''.
LFighter: Defending against Label-flipping Attacks in Federated Learning
The repository contains one jupyter notebook for each benchmark which can be used to re-produce the experiments reported in the paper for that benchmark. The notebooks contain clear instructions on how to run the experiments.
MNIST and CIFAR10 will be automatically downloaded. However, IMDB requires a manual download using this link. After downloading IMDB, please save it as imdb.csv in the data folder inside the folder IMDB.
The table below shows LFighter's robustness to the label-flipping attack with 40% attackers.
The figure below shows the source class stability under the label-flipping attack with 40% attackers for the CIFAR10-ResNet18-non-IID and IMDB-BiLSTM benchmarks.
Jebreel, N. M., Domingo-Ferrer, J., Sánchez, D., & Blanco-Justicia, A. (2024). LFighter: Defending against the label-flipping attack in federated learning. Neural Networks, 170, 111-126.
This research was funded by the European Commission (projects H2020-871042 SoBigData++'' and H2020-101006879 MobiDataLab''), the Government of Catalonia (ICREA Acad`emia Prizes to J.Domingo-Ferrer and to D. S'anchez, FI grant to N. Jebreel), and MCIN/AEI/ 10.13039/501100011033 and ERDF A way of making Europe'' under grant PID2021-123637NB-I00 CURLING''.