AI-Fun Seminar | Luis Muñoz-González: Machine Learning Under Attack
| Dates: | 13 March 2024 |
|---|
| Times: | 10:30 - 11:30 |
|---|
| What is it: | Seminar |
|---|
| Organiser: | Faculty of Science and Engineering |
|---|
| Who is it for: | University staff, External researchers, Current University students |
|---|
| Speaker: | Luis Muñoz-González |
|---|
|
The Manchester Centre for AI Fundamentals is hosting a series of seminars featuring expert researchers working in the fundamentals of AI.
Luis Muñoz-González is Senior Research Scientist, with roles at Imperial College London and Telefonica.
Title:
Machine Learning Under Attack
Abstract:
Despite the advances and the benefits of machine learning technologies, it has been shown that learning algorithms are vulnerable and can be compromised by attackers. Being one of the weakest components in the security chain, they are an appealing target for attackers, who can gain a significant advantage by exploiting these vulnerabilities. In this talk I will first introduce different poisoning attack strategies, aiming to compromise machine learning algorithms at training time, including formulations based on bilevel optimization and generative models, as well as defensive mechanisms to mitigate the impact of such attacks. The second part of the talk will focus on analyzing the systemic risks of machine learning models at run-time through the lens of Universal Adversarial Perturbations (UAPs), including computer vision applications and UAPs for malware detection generated in the problem space, i.e. considering adversarial manipulations of the software that result in functional malware.
Bio:
Dr Luis Muñoz-González is a senior research scientist at Telefónica Research in Barcelona, Spain. Before that, he worked as a research associate in the Department of Computing at Imperial College London, being part of the Resilient Information Systems Security (RISS) group led by Prof Emil Lupu. Dr Muñoz-González obtained a PhD in machine learning at the University Carlos III of Madrid, Spain. His PhD thesis on Gaussian Process models for nonstationary regression was recognized with the Extraordinary Doctorate Award. His current research interests lie at the intersection of machine learning and cyber security, including the security of machine learning, federated learning, and machine learning for cyber security.
Travel and Contact Information
Find event
2.60
Simon Building
Manchester
