Free Energy Principle

[BioBytes SRG]

Dear all,

Our next talk will be given by Shashwat Shukla, fourth year EE undergrad who was the co-founder of our SRG. He will be speaking about his research in the field of computational neuroscience and the various applications of Reinforcement Learning and Optimal Controls.

The venue of the talk is in LC 101 on 27th March, 2019 i.e Wednesday from 7pm.

For all the enthusiasts out there, here is an abstract of the talk.

Abstract: The Free Energy Principle is a unifying theory for brain-science research that posits that the brain is minimising a variational free-energy. Active Inference is a recently proposed framework that unifies the study of action and perception under this free-energy framework, elegantly formulating them as two sides of the same coin. In this talk, I will discuss how this theory subsumes other popular paradigms such as Reinforcement Learning and Optimal Control. The question of how such a variational free-energy can be minimized in biologically plausible networks of spiking neurons will also be discussed. In the process I will also touch upon topics such as Variational Autoencoders, Particle Filtering, Expectation-Maximisation, Stochastic Approximation theory and Spike-Timing-Dependent Plasticity.

Hope to see at the next talk! 

Do look out for our poster

Best regards,

Vipul Ramtekkar and  Soham Dibyanchintan

BioBytes SRG