BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Learning Bayesian inference models of cortical visual processing -
  Rodrigo Echeveste\, sinc(i)\, Santa Fe\, Argentina
DTSTART:20240618T090000Z
DTEND:20240618T100000Z
UID:TALK218026@talks.cam.ac.uk
CONTACT:Samuel Eckmann
DESCRIPTION:In this talk I will first give a brief broad overview of my cu
 rrent research lines after leaving CBL\, to then focus on the use of machi
 ne learning tools in order to model cortical visual processing in terms of
  Bayesian inference. I will first recap previous work at CBL where we show
 ed how recurrent neural networks trained for fast sampling-based Bayesian 
 inference display stereotypical features of cortical dynamics\, such as tr
 ansients and oscillations. This work relied on counting with a generative 
 model for the domain of interest\, which was then inverted to obtain an id
 eal observer model\, which the network was asked to mimic. In current work
  we learn both the generative and inference models from the data by use of
  variational auto-encoders. I’ll show ongoing work on this line\, partic
 ularly on how to extend classical VAEs to obtain models of visual percepti
 on with well behaved uncertainty estimates. I will also discuss how these 
 extensions can prove useful in other classical machine learning domains.\n
 \nRelated papers:\n\nEcheveste et al. (2020)\, Cortical-like dynamics in r
 ecurrent circuits optimized for sampling-based probabilistic inference.  h
 ttps://www.nature.com/articles/s41593-020-0671-1\n\nCatoni et al. (2024)\,
  Uncertainty in latent representations of variational autoencoders optimiz
 ed for visual tasks. https://arxiv.org/abs/2404.15390\n\nShort bio:\n\nRod
 rigo Echeveste obtained his Bachelor and Masters Degrees in Physics from B
 alseiro Institute in Argentina\, and his PhD from the Goethe University of
  Frankfurt\, Germany. He then did a three-year postdoc at CBL. Rodrigo cur
 rently holds a permanent research position as Adjunct Researcher from Arge
 ntina's National Research Council (CONICET) at the Research Institute for 
 Signals\, Systems and Computational Intelligence\, sinc(i)\, and is an Adj
 unct Professor at the National University of Litoral (UNL). His work lies 
 at the intersection of Computational Neuroscience and Machine Learning. Ro
 drigo currently serves as Secretary of Argentina’s Society for Neuroscie
 nce Research (SAN).
LOCATION:CBL Seminar Room\, Engineering Department\, 4th floor Baker build
 ing
END:VEVENT
END:VCALENDAR