Refreshments are available starting at 10:30 a.m. The seminar will begin at 10:45 a.m.
Abstract
Recent advances in computer vision have been powered by supervised learning—models today have been trained on hundreds of thousands of hand-labeled images and examples. Unfortunately, this approach doesn’t scale when we move to new domains, where annotation requires expertise and labeled data is near impossible to find. It is also unlike how humans learn, which is through unannotated interactions with the world around them.
In this talk, Bharath Hariharan will explore an alternative approach, intelligence is built from unlabeled data. He will present case studies from his work where his team goes all the way from training feature representations from unlabeled data to discovering new concepts automatically without supervision.
Speaker Biography
Bharath Hariharan is an associate professor of computer science at Cornell University. He works on computer vision and machine learning—in particular on important problems that defy the “big data” label. He enjoys problems that require marrying advances in machine learning with insights from computer vision, geometry, and domain-specific knowledge. Hariharan’s work has been recognized with an NSF CAREER Award and an Institute of Electrical and Electronics Engineers Pattern Analysis and Machine Intelligence Young Researcher Award.