
Saranraj Nambusubramaniyan
@memoryencoder
Generative Models & Neurorobotics; Guest @ Iris-i
ID: 2530069718
https://sarannns.github.io 28-05-2014 13:51:12
191 Tweet
317 Followers
547 Following


*Symmetry-Based Representations for Artificial and Biological General Intelligence* by Higgins Racanière Danilo J. Rezende Nice introductory paper on the role that symmetries are gaining in learning representations both in AI and neuroscience. arxiv.org/abs/2203.09250


#PSYCHE Public Lecture: You might know what having a #memory feels like, but do you know what the #brain is doing when we remember? Come join Upinder Bhalla this Sunday to decode the experience of remembering. Bhalla Lab National Centre for Biological Sciences Sign up here: bit.ly/Upinder_Bhalla



We're back after a two-year hiatus! Inviting applications for CAMP (Computational Approaches to Memory and Plasticity), an international computational neuroscience school, in-person National Centre for Biological Sciences, India. 23 Jul–7 Aug 2022. Details: camp.ncbs.res.in Please retweet!






"The Why, How, and When of Representations for Complex Systems" It's been said before, but what a terrific contribution from Dr. Leo Ann Sizemore Blevins Dani S. Bassett @tinaeliassi -- great for junior researchers entering complex systems, great for people already in the field.




Global media is speculating whether current challenges in the business sector will trip India’s ambitions to be a global economic force. I’ve lived long enough to see us face earthquakes, droughts, recessions, wars, terror attacks. All I will say is: never, ever bet against India

Harsh, but fair. For those tempted to overplay this * Demis Hassabis majored & did his PhD in cog neuro * Geoffrey Hinton majored in Psych & is founding director of the computational neuro. unit at UCL * Yann LeCun link convs to neuroscience Breakthroughs are not spontaneous.



Active Inference Institute Finding the best sequence of actions that minimize an objective function (an energy) measured over the predicted trajectory is very much like inferring latent variables. If you insist on being Bayesian, then you infer a distribution over action sequences by trading the average

