A Review of Adversarial Machine Learning for Cyber security in Intelligent Systems using Image Classification & Signal Classification Analysis

Machine learning models with adversarial examples must be trained to differentiate between genuine and invading data. Cyber computing technology and tools face huge intrusion attacks and malicious programs. Tampered QR code for online payments, malicious face and biometric detection, incorrect facial recognition, object and sign detection are the adversarial examples in real-life scenarios. Convolutional neural networks often fail to image classification tasks due to high vulnerability. The Intelligent systems comprising computer vision, self-driving cars, radio signals, IoT Devices, industrial production and automation are prone to cyber attack with adversarial data in machine learning models. The goal of this paper is to summarize the various Convolutional Neural Network models and Recurrent Neural Network approaches used in machine learning for malicious image classification. The analysis aims to identify the area of attacks using adversarial machine learning in cyber physical systems and intrusion detection systems.