- Date : 15/11/2018
- Place : 2 rue Simone Iff 75012 - Salle Lions 1, bât C
- Guest(s) : Martin Abadi
- Organiser(s) : Gallium
As machine learning becomes more powerful and pervasive, it has an impact on the many software systems that rely on it and on the underlying software and hardware platforms. The resulting questions should concern virtually all aspects of the theory and practice of software, as they relate to software engineering, security and trust, programming, tools, foundations, and more. This talk will discuss some of those questions. In particular, it will present recent research on the semantics of programming languages suitable for learning by gradient descent and on adversarial machine learning.
(This talk is a lightly updated version of one given at ETAPS 2018.)