Dipartimento di Informatica

Date: November 24, 2021
Time: 17:00
Venue: https://meet.google.com/yjz-eogn-gas
Speaker: Fabio De Gaspari

Title: Machine Learning and Cyberseurity

Abstract: Advancements in machine learning (ML), and in particular deep neural networks, are pushing us towards an increasingly automated cyberspace. Intelligent systems are applied throughout all fields of IT and, in recent years, also to cybersecurity. Intelligent cyber defense agents promise to replace human experts in the cyber security domain, providing near instant incident response and adaptive security for the systems. However, while extremely successful, machine learning techniques are not designed to work in adversarial settings. Several works show how it is possible to hijack machine learning classifiers to evade classification or leak sensitive information. This talk discusses applications of ML to the cyberspace and the challenges of using ML techniques in adversarial settings.