Meta-RL and Non-Spatial Sequences
I’ll talk about two loosely related ideas. First, I’ll describe a simple way of learning a smart (prior-dependent) reinforcement learning algorithm using recurrent networks, which we call meta-RL. Second, I’ll talk about experimental work in MEG where we found spontaneous reactivation of sequences of states in a non-spatial task. These things are related insomuch as meta-RL depends on incremental learning from a set of different tasks and needs experience to be randomized, which spontaneous reactivation could provide. More broadly, there’s a lot to learn from the relationships between all your past experiences.
Date:
30 May 2017, 13:00 (Tuesday, 6th week, Trinity 2017)
Venue:
Biology South Parks Road, South Parks Road OX1 3RB
Venue Details:
Lecture Theatre
Speaker:
Dr Zeb Kurth-Nelson (Google DeepMind)
Organising department:
Department of Experimental Psychology
Organiser:
Nils Kolling (Junior Research Fellow, Experimental Psychology, University of Oxford)
Organiser contact email address:
nils.kolling@psy.ox.ac.uk
Host:
Dr Nick Myers (University of Oxford)
Booking required?:
Not required
Audience:
Members of the University only
Editor:
Stephanie Mcclain
