Venue
Nordita, Stockholm, Sweden
Scope
What is learning? What makes both artificial and biological neural networks so efficient at learning? To what extent features of the training data influence learning? These important questions have remained unanswered despite our ability to train deep neural networks for a variety of applications.
Most of the progress in deep learning has been achieved heuristically and it can be argued that deep learning is in a state similar to what thermodynamics was in before Carnot. However, in the last decade a number of promising ideas have emerged in Physics (statistical mechanics) and mathematics (algebraic topology). In parallel, detailed connectivity and activity of biological neural networks have become available and researchers are extracting signatures of learning in animals. Synthesis of these various theoretical and experimental developments is essential to build a theory of learning, develop appropriate mathematical tools to train artificial networks in an efficient manner and to design better experiments/hypotheses to understand neural correlates of learning biological neural networks. The goal of the proposed NORDITA program is to bring together leading Physicists, Mathematicians and Neuroscientists to discuss recent advances in the understanding of learning in artificial and biological neural networks and future perspectives.
Invited participants (confirmed)
- John Hertz (NORDITA)
- Christopher Lynn (Yale University)
- Prasad Perlekar (TIFR-Hyderabad)
- Mason Porter (NORDITA)
- Haim Sompolinsky (Harvard University)
Accommodation
Nordita may provide a limited number of rooms at the Stockholm apartment hotel Biz Apartment free of charge for participants that need support, which will be allocated after application evaluation.
Accommodation for families is possible upon request and depends on availability. No family rooms have been booked in advance
Travel Support
There may be a (very) limited number of travel grants available for participants. If you are interested in one, please apply for it by ticking the corresponding checkbox in the application form.
Application/Registration
Registration to be considered for on-site participation will close TBD. Registrants will receive an on-site/remote participation confirmation from the organizers after this date.
Sponsored by:
