• Dr. Bodla Kishor
• Mrs.Rafath Samrin
• Sathini Santhosh Kumar
• Vandhanapu Srinu

The Machine Learning (ML) systems role increasing tremendously in various technical domains day to day. Even though the applications of ML performing well in all aspects still some issues are making performance down among which security is one which is very reliable factor for all applications used by the end users. In spite of designing robust machine learning models still are vulnerable to various attacks. In this paper we conducted a strong comprehensive study on various security issues of machine learning. This study gives a better base to the future research on this area. The nature of these attacks cannot be explained properly due to stealthy in behaviour. This study gives a systematic analysis of security issues of ML by looking into existing attacks on machine learning systems related to defenses or secure learning techniques, and security evaluation methods. This survey focussing on all types of attacks from training phase to the test phase. Instead of focusing on one stage or one type of attack, this paper covers all the aspects of machine learning security from the training phase to the test phase. First, the machine learning model in the presence of adversaries is presented, and the reasons why machine learning can be attacked are analyzed. We review the state of the art approaches where ML is applicable more effectively to fulfil current real-world requirements in security. We examine different security applications perspectives where ML models play an essential role and compare, with different possible dimensions their accuracy results. We segregated these attacks in to training set poisoning, backdoors in the training set, adversarial example attacks, model theft and recovery of sensitive training data. Several suggestions on security evaluations of machine learning systems are also provided. Even with the use of current sophisticated technology and tools, attackers can evade the ML models by committing adversarial attacks.

ML, security, privacy, adversarial attacks, models, training and test phase, vulnerabilities, AI. cyber security.

"Security Issues of Machine Learning Systems: A Comprehensive Study", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.9, Issue 5, page no.f31-f42, May-2022, Available :http://www.jetir.org/papers/JETIR2205605.pdf

"Security Issues of Machine Learning Systems: A Comprehensive Study", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.9, Issue 5, page no. ppf31-f42, May-2022, Available at : http://www.jetir.org/papers/JETIR2205605.pdf