BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:Asia/Kolkata
BEGIN:STANDARD
DTSTART:19451014T230000
TZOFFSETFROM:+0630
TZOFFSETTO:+0530
TZNAME:IST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20240206T134541Z
UID:8278B2DE-56B2-48DD-A551-6B6958201A85
DTSTART;TZID=Asia/Kolkata:20240202T160000
DTEND;TZID=Asia/Kolkata:20240202T170000
DESCRIPTION:Attention can be deployed either voluntarily – based on task 
 goals – or captured automatically – by salient sensory stimuli. Previo
 us studies have controlled stimulus salience by altering low-level image f
 eatures (e.g.\, luminance\, or popout)\, or by inducing motion dynamics (e
 .g.\, flash\, loom)\; such salient stimuli capture attention automatically
  by driving strong neural responses in multiple visual areas. Yet\, precis
 ely what combinations of naturalistic (high-level) stimulus features drive
  the strongest neural responses\, and produce the highest behavioral salie
 nce\, remains a topic of active research. As a first step toward this goal
 \, I will describe the design of salient “super-stimuli” – high-reso
 lution\, naturalistic images tailor-made to evoke the strongest responses 
 in specific brain areas. The high dimensionality of natural images renders
  this optimization prohibitively challenging at the pixel level. To tackle
  this challenge\, we extend a recently developed framework called XDream [
 5]. This framework employs a deep generative network in combination with a
  heuristic optimization (genetic) algorithm and was recently tested to gen
 erate “super-stimuli” for the monkey visual cortex. We extend this fra
 mework with a CNN-based encoder for human functional MRI (fMRI) brain resp
 onses and design novel classes of “super-stimuli” optimized for the hu
 man brain. Specifically\, we advance the optimization algorithm to address
  the following questions: i) Given an object category (e.g.\, telephones o
 r computers) can we design a class of super-stimuli that still respects ca
 tegory boundaries? ii) Can we design “chimeric” super-stimuli that can
  combinatorially activate (or suppress) multiple brain regions (e.g.\, pri
 mary visual cortex and face area)\, at once? I will conclude by describing
  ongoing work that seeks to validate the generated images by measuring hum
 an visual cortex responses directly with fMRI and quantifying the behavior
 al salience of the generated images with psychophysical experiments.\n\nSp
 eaker(s): Prof. Sridhar Devarajan\, \n\nVirtual: https://events.vtools.iee
 e.org/m/403671
LOCATION:Virtual: https://events.vtools.ieee.org/m/403671
ORGANIZER:ieee.embs.sbc.iitkgp@gmail.com
SEQUENCE:16
SUMMARY:Designing salient\, naturalistic “super-stimuli” with deep gene
 rative models
URL;VALUE=URI:https://events.vtools.ieee.org/m/403671
X-ALT-DESC:Description: <br /><p>Attention can be deployed either voluntari
 ly &ndash\; based on task goals &ndash\; or captured automatically &ndash\
 ; by salient sensory stimuli. Previous studies have controlled stimulus sa
 lience by altering low-level image features (e.g.\, luminance\, or popout)
 \, or by inducing motion dynamics (e.g.\, flash\, loom)\; such salient sti
 muli capture attention automatically by driving strong neural responses in
  multiple visual areas. Yet\, precisely what combinations of naturalistic 
 (high-level) stimulus features drive the strongest neural responses\, and 
 produce the highest behavioral salience\, remains a topic of active resear
 ch. As a first step toward this goal\, I will describe the design of salie
 nt &ldquo\;super-stimuli&rdquo\; &ndash\; high-resolution\, naturalistic i
 mages tailor-made to evoke the strongest responses in specific brain areas
 . The high dimensionality of natural images renders this optimization proh
 ibitively challenging at the pixel level. To tackle this challenge\, we ex
 tend a recently developed framework called XDream [5]. This framework empl
 oys a deep generative network in combination with a heuristic optimization
  (genetic) algorithm and was recently tested to generate &ldquo\;super-sti
 muli&rdquo\; for the monkey visual cortex. We extend this framework with a
  CNN-based encoder for human functional MRI (fMRI) brain responses and des
 ign novel classes of &ldquo\;super-stimuli&rdquo\; optimized for the human
  brain.&nbsp\; Specifically\, we advance the optimization algorithm to add
 ress the following questions: i) Given an object category (e.g.\, telephon
 es or computers) can we design a class of super-stimuli that still respect
 s category boundaries? ii) Can we design &ldquo\;chimeric&rdquo\; super-st
 imuli that can combinatorially activate (or suppress) multiple brain regio
 ns (e.g.\, primary visual cortex and face area)\, at once? I will conclude
  by describing ongoing work that seeks to validate the generated images by
  measuring human visual cortex responses directly with fMRI and quantifyin
 g the behavioral salience of the generated images with psychophysical expe
 riments.&nbsp\;</p>
END:VEVENT
END:VCALENDAR