AI/ML Data Poisoning Attacks Explained and Analyzed-Technical

Adversarial artificial intelligence and machine learning is a growing threat in cybersecurity and data science. Algorithms are vulnerable to new types of stealthy and effective cyberattacks. Threat actors can and have altered the machine learning and artificial intelligence models without ever gaining access to their systems. This is called data poisoning or model poisoning. Find out more. Chapters: 00:00 The implications 00:56 Data Poisoning Explained 02:20 The Poisoning Threshold 04:21 Attack Objectives 05:07 Backdoor Attacks 07:27 Attack Vectors 10:38 Vulnerability 12:46 Attack Strategies 14:34 Previous Successful Attacks Sources Cited: #1. https://arxiv.org/pdf/1804.00792.pdf #2. https://github.com/aks2203/poisoning-... https://citeseerx.ist.psu.edu/viewdoc... https://www.energy.gov/sites/default/... https://arxiv.org/pdf/2009.07008.pdf https://www.usenix.org/system/files/s... https://arxiv.org/pdf/1811.00121.pdf https://arxiv.org/pdf/1811.09982.pdf https://arxiv.org/pdf/2006.12557.pdf https://citeseerx.ist.psu.edu/viewdoc... https://ieeexplore.ieee.org/stamp/sta... https://www.forbes.com/sites/naveenjo... https://blogs.microsoft.com/blog/2016... https://atlas.mitre.org/studies/AML.C... https://arxiv.org/pdf/2007.08432.pdf https://manuscriptlink-society-file.s... https://citeseerx.ist.psu.edu/viewdoc... https://arxiv.org/pdf/1910.00033.pdf https://arxiv.org/pdf/2012.03765.pdf -knearest https://elie.net/blog/ai/attacks-agai... https://people.eecs.berkeley.edu/~tyg... https://arxiv.org/pdf/1804.00792.pdf