Gradient Descent Adversarial Attacks on SVM for CAP EEG signals

Abstract

Machine learning models have been widely adopted in various applications, but their vulnerability to evasion attacks has become a significant concern. Evasion attacks on machine learning models aim to manipulate the test data in a way that causes the model to make incorrect predictions. In this paper, we performed gradient-based attacks on the support vector machine (SVM) model for cyclic alternating patterns (CAP) sleep phase test dataset. Performance of the classifier is evaluated under evasion attacks and detailed analysis on robustness of model has been done.