Security and Privacy Attacks on Machine Learning Algorithms

Machine Learning has gained a significant increase in popularity in the recent times. ML models are being used in almost every other field including medicine, finance and many more. As machine learning has increasingly been deployed in critical real-world applications, the dangers of manipulation and misuse of these models has become of paramount importance to public safety and user privacy. In applications such as online content recognition to financial analytics to autonomous vehicles all have shown to be vulnerable to adversaries wishing to manipulate the models or mislead models to their malicious ends. Technical community's understanding of the nature and extent of these vulnerabilities remains limited even though there has been a growth in recognition that ML exposes new vulnerabilities in software systems. Identifying various types of privacy and security attacks possible on ML models and demonstrating those attacks is the focus of the project. For security part adversarial attacks on Machine Learning models will be introduced and privacy part model inversion attack and membership inference attack will be performed to show that ML models leak information.