luzai / LN-robust-svm

Try to reproduce Support Vector Machines Under Adversarial Label Noise

LN-robust SVM

Try to reproduce support vector machines under adversarial label noise.

About Adversarial label flip attack

• run python data.py can produce the following results.

• Follow Adversarial label flip attack algorithm,

the training result on untainted toy data is

, the training result on tainted toy data is .

with mu=0.5 and LN-svm, we learned