250 Hutchison Rd, Rochester, NY 14620

View map

Join the Goergen Institute for Data Science for Mitigating Bias in Deep Neural Networks: Lessons from Continual Machine Learning, Medical Computer Vision, and Language Guided Scene Understanding with Christopher Kanan, Associate Professor of Computer Science at University of Rochester.

Abstract: Deep learning has been tremendously successful, but artificial intelligence (AI) still has a long way to go toward achieving the versatility of humans. My lab works toward overcoming the limitations of today’s AI systems. One of the biggest limitations is effective learning from biased datasets. Training machine learning systems, especially deep neural networks, on biased datasets often leads to systems performing well on in-distribution data, but this performance is due to relying on spurious features that fail to generalize. In this talk, I’ll discuss how we mitigated the impact of dataset bias in 1) continual learning systems that learn from ever-growing datasets, 2) systems for detecting cancer in radiology and pathology imagery, and 3) systems for language-guided scene understanding. I’ll show how “OccamNetworks” – a novel network architecture that my lab has developed – can provide a new avenue toward mitigating dataset bias in deep learning.

Bio: Christopher Kanan is an Associate Professor of Computer Science at the University of Rochester. His research focuses on deep learning, especially lifelong machine learning, where he takes inspiration from cognitive science to make neural networks capable of learning over time for large-scale vision and multi-modal perception tasks. Other recent projects cover self-supervised learning, open-world learning, and creating bias robust neural network architectures. He also works on medical applications of machine learning, especially computational pathology. Previously, he led AI R&D at the start-up Paige, leading to the first FDA approved computer vision system for helping pathologists diagnose cancer in whole slide images. He is now mentoring several former students in developing their own startups. Kanan received a PhD in computer science from the University of California at San Diego and completed his postdoc at the California Institute of Technology. Prior to joining the University of Rochester, he worked at NASA Jet Propulsion Laboratory, followed by becoming faculty at the Rochester Institute of Technology. He is an NSF CAREER award winner.

We are providing a Zoom option for this event. Please use the link below to enter the Zoom webinar.

Webinar link: https://rochester.zoom.us/j/93115149227

Webinar ID: 976 1226 3898

Event Details

User Activity

No recent activity