[BIC-announce] Fwd: JULAIN Talk 17.3.2022, 4pm | Shahab Bakhtiari, MILA | Specialized parallel pathways in brains and artificial neural networks (fwd)

[Paule-Joanne Toussaint]

Dear All,

Shahab Bakhtiari -Postdoctoral Fellow with Blake Richards in the LiNC Lab, McGill U., and MILA- is giving a JULAIN Talk (Forschungszentrum Juelich) online on Thursday March 17th at 16:00h Juelich / 11:00h Montreal. Please join.

Best,
Paule-J

---------- Forwarded message ----------
Von: "Wenzel, Susanne" <s.wenzel at fz-juelich.de>
Datum: Donnerstag, 3. März 2022 um 10:48
An: ML <ML at fz-juelich.de>
Betreff: JULAIN Talk 17.3.2022, 4pm | Shahab Bakhtiari, MILA | Specialized
parallel pathways in brains and artificial neural networks

 

Dear colleagues,

We are happy to announce our next JULAIN Talk:

Shahab Bakhtiari, PhD
Montreal Institute for Learning Algorithms (MILA)

will talk about 
Specialized parallel pathways in brains and artificial neural networks

·         When: 17 March 2022, 4pm

·         Where: virtual event

·         video conf
link: https://zoom.us/j/91305203810?pwd=Qkxyd0ZOdCtoWldGdHlrMUhXOU9LUT09
Meeting ID: 913 0520 3810    Passcode: 9d4uqw

   Moderation: Timo Dickscheid, Helmholtz AI, INM-1

   Abstract
   The visual system of mammals is comprised of parallel, hierarchical
   specialized pathways. Different pathways are specialized in so far as
   they use representations that are more suitable for supporting specific
   downstream behaviours. In particular, the clearest example is the
   specialization of the ventral (" what") and dorsal (" where") pathways
   of the visual cortex. These two pathways support behaviours related to
   visual recognition and movement, respectively. To-date, deep neural
   networks have mostly been used as models of the ventral, recognition
   pathway. However, it is unknown whether both pathways can be modelled
   with a single deep ANN. In this talk, I will show how a single model
   with a single loss function can capture the properties of both the
   ventral and the dorsal pathways. I show that when we train a deep neural
   network architecture with two parallel pathways using a self-supervised
   predictive loss function, we can outperform other models in fitting the
   visual cortex. Moreover, we can model both the dorsal and ventral
   pathways. These results demonstrate that a self-supervised predictive
   learning approach applied to parallel pathway architectures can account
   for some of the functional specialization seen in mammalian visual
   systems.

Short CV

Shahab Bakhtiari is a postdoctoral researcher at Mila - Quebec AI Institute. His
research is focused on representation learning in brains and artificial neural
networks. Shahab received his PhD degree in neuroscience at McGill University.
Before that, he received his bachelor's and master's degrees in electrical
engineering from University of Tehran, Iran.

 

 

---

JULAIN – The JUeLich AI Network of ML&DL interested researchers

 

Mail: ml at fz-juelich.de

Join the mailinglist at https://lists.fz-juelich.de/mailman/listinfo/ml

Join the Journal club’s mailinglist at
https://lists.fz-juelich.de/mailman/listinfo/julain_journal_club

 

Visit us at

wiki:   https://go.fzj.de/JULAIN_WIKI

chat:  https://chat.fz-juelich.de/channel/JULAIN  

 

--
Dr.-Ing. Susanne Wenzel
Institute of Neuroscience and Medicine (INM-1)

Forschungszentrum Jülich GmbH
in the Helmholtz Association
52425 Jülich
Germany

+49-2461-61-96306
s.wenzel at fz-juelich.de