Replication Package for the paper entitled "SecDefender: Detecting Low-Quality Models in Multidomain Federated Learning Systems".

Description of the content:

"SecDefender_Appendix.pdf" includes the model architectures and hyperparameters used to train the different datasets and comprehensively evaluates SecDefender performance using ASR (Attack Success Rate) and TMR (Target Misclassification Rate) metrics on the selected datasets under several attacks and defense scenarios. Moreover, it includes the analysis of hyperparameters such as batch size, momentum size, learning rate size, and finally, the AUC ROC Curve of our used datasets.

List of Authors:

Sameera K M, Arnaldo Sgueglia, Vinod P, Rafidha Rehiman K A, Corrado Aaron Visaggio, Andrea Di Sorbo, Mauro Conti.