The next topic is ‘probabilistic representations’ presented by Mate Lengyel and Wayne Soo.
Mate will begin by introducing and comparing different types of neural representations of uncertainty. Wayne will then present a very recent paper that delves into such representations by analysing monkey V1 recordings: Representation of visual uncertainty through neural gain variability
https://www.nature.com/articles/s41467-020-15533-0
Wayne will then discuss a new and rapidly evolving research direction that makes use of the advent of AI in the context of probabilistic representations for constructing neural network models: Efficient probabilistic inference in generic neural networks trained with non-probabilistic feedback
https://www.nature.com/articles/s41467-017-00181-8