Time slot 1: Saturday 22 August, 5:30 pm - 7:00 pm (San Francisco Time)
Time slot 2: Sunday 23 August, 6:30 am - 8:00 am (San Francisco Time)
Learning with limited data or labels remains an important unsolved problem in computer vision. It prevents us from democratizing AI technology today. While humans accomplish this task seamlessly, contemporary AI algorithms struggle with it. So, how far along have we come towards solving this problem in vision? What are some new frontiers to move forward? In this turorial we will explore several emerging ideas and exciting breakthroughs that are pushing the boundaries of what is possible for this task. We will first describe exciting new accomplishments in self-supervised learning from image collections and videos, which are helping to learn a myraid of vision tasks. Next we will show emerging ideas around learning with imperfect labels and of unification across different levels of supervision. Lastly, we will present two new ideas around inverting networks to provide data or feedback, which can help to learn in limited data scenarios.
Presentation Playlist:
Speaker Introduction: Shalini De Mello
Self-Supervised Part and Viewpoint Discovery from Image Collections: Varun Jampani
Learning Visual Correspondences across Instances and Video Frames: Sifei Liu
Limitless Labels in a Labelless World: Weak Supervision with Noisy Labels: Arash Vahdat
Learning with Imperfect Labels and Visual Data: Zhiding Yu
Inverting Neural Networks for Data-free Knowledge Transfer: Pavlo Molchanov
Learning Efficiently with Biologically Inspired Feedback: Yujia Huang
