Overview
Explore the computational foundations of primate and machine vision in this lecture by Thomas Serre from Brown University. Delve into topics such as universal approximators, face recognition, image noise, and Tesla's Autopilot system. Examine the 'Same-Different' problem and investigate horizontal and top-down connections in visual processing. Gain insights into extraclassical receptive fields and their role in computational neuroscience. Analyze how breaking constraints and contextual illusions, including the tilt illusion, impact visual perception. Discover the benefits of recurrent connections in visual systems and their implications for both biological and artificial vision.
Syllabus
Intro
Universal Approximator
Face Recognition
Image Noise
Tesla Autopilot
Clever
Same Different
Horizontal and stopdown connections
Extraclassical receptive fields
Computational neuroscience
Breaking constraint
Contextual illusions
Tilt illusion
Benefits of recurrent connections
Taught by
MITCBMM