/LLR

Primary LanguagePython

Run training on OCT2017 dataset

run main.py

Test robust accuracy on OCT2017 dataset

run pgd_attack_oct.py

Run training on HAM10000 dataset

run main_ham10000.py

For existing experiment, we run RST (supervised version) with varying degrees of beta on the HAM10000 dataset

Test robust accuracy on HAM10000 dataset

run pgd_attack_ham10000.py

Run Simba

The folder SimBA has the scripts needed to conduct SimBA attack on the HAM10000 dataset (simple blackbox attack)

run_simba_ham.py conducts the attack. simba_eva.py evaluates the natural and adversairal accuracy under the simba attack.