Explore the intricacies of feedback mechanisms in the visual system through this 20-minute lecture by Grace Lindsay from University College London. Delve into masking studies that reveal the function of local feedback and learn about building a network with separated functions. Discover how two-stage training replicates behavioral findings and visualize activity in different layers. Examine classification accuracy over time and understand how recurrence affects noise classification across layers. Gain insights into the local feedback summary and its implications for visual processing.
Modeling the Influence of Feedback in the Visual System
Overview
Syllabus
Modeling the influence of feedback in the visual system
Masking studies indicate a function of local feedback
Building a network with a separation of functions
Two-stage training replicates behavioral findings
Visualizing Layer 1 Activity
Visualizing Layer 4 Activity
Classification accuracy over time
Recurrence increases noise classification at early layers and decreases it at later layers
Local Feedback Summary
Taught by
MITCBMM
