The Division of Information Science and Engineering at KTH is looking for a doctoral student in the following area: Information theory, with specialization in AI and machine learning.
The goal of this project is to analyze machine learning techniques by using information theoretic tools, to characterize their performance, and in this way, to explain their properties as well as to contribute to the development of information theory in general and with applications in AI and machine learning in particular. Recent examples within the area are the information bottleneck method applied to deep learning, as well as new information theoretic results that relate to generalization, information leakage and robustness. The project will mainly focus on fundamental mathematical tools and results.
Supervision: The doctoral student will be supervised by: Professor Mikael Skoglund
This call is part of a joint call for doctoral students in Math/AI. It is done in several steps, where finding doctoral students is the second step. The first step, the project call, can be found here: https://wasp-sweden.org/positions/call-for-phd-projects-in-mathematics-for-ai/