Explore the cutting-edge field of self-supervised learning in computer vision through this comprehensive 1 hour 36 minute lecture by Ishan Misra. Delve into various advanced techniques including Pretext-invariant representation learning (PIRL), Swapping assignments between views (SwAV), Audiovisual instance discrimination with cross model agreement (AVID + CMA), and Barlow Twins for redundancy reduction. Gain insights into the latest developments and applications of these self-supervised learning methods in computer vision. Access additional resources such as course website, slides, and a full YouTube playlist to enhance your understanding of this complex topic. Conclude the session with a live chat to address any questions or discussions arising from the lecture content.
Self-Supervised Learning in Computer Vision
Overview
Syllabus
– Welcome to class
– Self-supervised learning in computer vision
– Pretext-invariant representation learning PIRL
– Swapping assignments between views SwAV
– Audiovisual instance discrimination with cross model agreement AVID + CMA
– Barlow Twins: self-supervised learning via redundancy reduction
– Live chat
Taught by
Alfredo Canziani
