GFM-Plus

Official codes for "Faster Gradient-Free Algorithms for Nonsmooth Nonconvex Stochastic Optimization".

Nonconvex SVM

To reproduce the experiments, run

cd ./SVM
python -u train.py --data_file a9a/w8a/covtype/ijcnn --epochs=10
python -u train.py --data_file mushrooms --epochs=50
python -u train.py --data_file phishing --epochs=200

The datasets used in our experiments are available at https://www.csie.ntu.edu.tw/~cjlin/libsvm/

Black-Box Attack on CNN

To reproduce the experiments on MNIST, please download the MNIST dataset in args.data_dir and run

cd ./Attack
python -u AttackMNIST.py

You can also download the dataset automatically by setting download=True in the following code

datasets.MNIST(root=args.data_dir, train=False, transform=test_transform, download=True)

The experiments on FashionMNIST are very similar and we do not contain the codes here.