We invite applications for a two-year postdoc in the “Theory of Neural Networks” group led by Sebastian Goldt, starting in September 2021 or earlier if possible. The group is part of the nascent “Data Science Initiative” at the International School of Advanced Studies (SISSA) in Trieste, Italy.
Project: Neural networks underpin many of the current successes of machine learning. Yet most of this progress has come through advances in engineering, while a theoretical understanding of why they are so effective is still lacking. We combine a mix of theory and numerical experiments to advance our understanding of foundational aspects of neural networks. Our current focus is on understanding the interplay between data structure, training and architecture during learning. We analyse these questions using tools from statistical physics and statistics. Excursions into neuroscience are a possibility.
We are seeking highly motivated individuals who hold a PhD in disciplines ranging from theoretical physics, applied mathematics, information theory / signal processing to classical machine learning. Strong analytical skills and an interest in conducting numerical experiments are important.
Many of the greatest ideas and discoveries have come from a diverse mix of minds, backgrounds and experiences, and so we are committed to cultivating an inclusive work environment that is welcoming to everybody.
Required Materials: Interested applicants are invited to reach out directly to Sebastian Goldt with any questions they might have. To apply, applicants should upload the following documents:
A statement of motivation and interest in the project (<=2 pages)
Curriculum Vitae
Publication list
Names and email addresses of two referees - we don’t need the actual letters at this point.
The deadline for applications is 31.01.2020, and we encourage earlier applications. More information on the Data Science Group at SISSA can be found here. |