Events & Calendars Vulnerabilities of Data Analysis: From Political Spin to Data Manipulation

"Vulnerabilities of Data Analysis: From Political Spin to Data Manipulation" presented by Scott Alfeld
The Faculty Colloquium Series for 2019-20 presents a lecture titled "Vulnerabilities of Data Analysis: From Political Spin to Data Manipulation" presented by Scott Alfeld, assistant professor of computer science.

https://www.amherst.edu/mm/33214

Learn how to bypass a state-of-the-art security system with sugar cubes and a slingshot. New security vulnerabilities have arisen with the growing use of machine learning in decision-making systems. In this talk I discuss the field of Adversarial Learning — the study of using machine learning techniques when the input data may be corrupted by an attacker. I’ll cover various techniques hackers employ and defenses again them. We'll then get our hands dirty attacking a learner at training time and see what finesse is needed when manipulating even the simplest of learners.

Faculty Colloquium events are sponsored by a group of faculty colleagues who meet informally with the purpose of supporting and promoting the College’s commitment to faculty research. Colleagues interested in joining this endeavor are welcome and should contact us by email: adsarat@amherst.edu . Faculty, staff, and members of the administration are cordially invited to attend these presentations.