{"id":7287,"date":"2025-02-25T11:25:23","date_gmt":"2025-02-25T16:25:23","guid":{"rendered":"https:\/\/carleton.ca\/cuids\/?post_type=cu-events&p=7287"},"modified":"2025-03-13T11:55:34","modified_gmt":"2025-03-13T15:55:34","slug":"putting-test-time-adaptation-to-the-test-updates-for-robust-visual-recognition","status":"publish","type":"cu-events","link":"https:\/\/carleton.ca\/cuids\/cu-events\/putting-test-time-adaptation-to-the-test-updates-for-robust-visual-recognition\/","title":{"rendered":"[RECORDED] Putting Test-Time Adaptation to the Test: Updates for Robust Visual Recognition"},"content":{"rendered":"

CUIDS Distinguished Speaker Series<\/h2>\n

Putting Test-Time Adaptation to the Test: Updates for Robust Visual Recognition<\/h3>\n

When the data shifts between training and testing the accuracy of predictions can degrade. The problem is that the data changes, but our\u00a0systems remain the same. Does it have to stay this way? In this talk we will examine how visual recognition can adapt and generalize to new and different data during testing. We will cover natural shifts, such as image corruptions, to highlight opportunities to update models (by entropy minimization and parameter mixing) and inputs (by diffusion) when the data differs. Then we will take a critical look at adversarial attacks through our case study of test-time adaptive defenses. On this front more adaptation is still needed in practice for empirical robustness, but there is progress in theory for certified robustness with our adaptive extension of randomized smoothing. As a last step, we will discuss what is next for test-time updates.<\/p>\n

\"PhotoEvan Shelhamer<\/a>\u00a0is an assistant professor at UBC in Vancouver, member of the Vector Institute, and senior research scientist at Google DeepMind. His research is on visual recognition, self-supervised learning without annotations, and robustness by adaptation. He earned his PhD at UC Berkeley advised by Prof. Trevor Darrell. He was the lead developer of the Caffe open-source deep learning framework from version 0.1 to 1.0. His research and service have received awards including the best paper honorable mention at CVPR’15 for fully convolutional networks and the Mark Everingham award at ICCV’17, the open-source award at MM’14, and the test-of-time award at MM’24 for Caffe. He is new to Canada and is excited to learn more about all of its provinces and seasons!<\/p>\n

Moderator: Prof. Jim Green<\/a>, Department of Systems and Computer Engineering<\/p>\n

RSVP today for the CUIDS Distinguished Speaker Series.<\/p>\n\n

\n\t

\n\t\tRSVP<\/a>\n\t<\/p>\n<\/div>\n\n