Can Fairness and Robustness Be Simultaneously Achieved Under Byzantine Attacks?

This hub stores the code for paper Can Fairness and Robustness Be Simultaneously Achieved Under Byzantine Attacks?

Install

1. Download the dependant packages (c.f. install.sh):

• python 3.8.10
• pytorch 1.9.0
• matplotlib 3.3.4

2. Download the dataset to the directory ./dataset and create a directory named ./record. The experiment outputs will be stored in ./record.

Runing

Run AFL

python AFL.py --aggregation trimmed-mean --attack sign_flipping --data-partition noniid --seed 1

The arguments can be

<aggregation>:

• mean
• trimmed-mean
• median
• geometric-median
• faba
• Krum
• bulyan

<attack>:

• none
• gaussian
• sign_flipping
• disguise

<data-partition>:

• iid
• noniid

Run other methods

To run other methods, such as $q$FFL, H-nobs, DRFL, one should simply run the corresponding .py file.