[Quantum-ms] [Seminar] EE Faculty Candidate Lecture 4/3 at 11:40am

[Columbia EE Events]

[image: Left-aligned blue cuee logo.png]

*Please join us for the EE faculty candidate lecture on 4/3!*

*When: *4/3, Thursday, 11:40am - 12:40pm
*Where: *CEPSR 750, also via zoom
<https://columbiauniversity.zoom.us/j/4419920270#success >.
*Who:* Dr. Bingyi Kang
*Title: *Visually Elicited Physical Intelligence

*Abstract: * Foundation models have transformed the field of computer
vision by learning general-purpose understanding and generation models that
connect the visual world with semantic concepts. However, the ability to
recover and model the physical world from visual observations is still in
its infancy, despite being crucial for humans and animals to interact with
others and their environment. We refer to these abilities as physical
intelligence, which is fundamental to various applications including AR/VR,
robotics and AI agents. In this talk, we explore physical intelligence and
present several studies aimed at its development. First, we demonstrate the
necessity of new techniques by showing that current state-of-the-art video
generation methods fail to capture the physical attributes of the
underlying world. Then, we discuss three crucial components in pursuing
physical intelligence: 1) Physical modeling: building foundation models
that can recover geometry information from arbitrary visual inputs; 2) Data
intelligence: learning from real-world data with arbitrary distributions;
3) Decision making: training AI agents or robots with high-level planning
and low-level control. Finally, we discuss future directions for bringing
physical intelligence into reality and exciting opportunities for
interdisciplinary collaborations.

*Bio: *Bingyi Kang is currently a Research Scientist and Lead at TikTok
Research in Seattle. He received his Ph.D. from the National University of
Singapore in 2021 and was a visiting scholar at UC Berkeley, working with
Trevor Darrell. His research areas include embodied AI, computer vision and
machine learning. He aims to develop AI agents that can perceive, model,
and interact with the physical world. He led the research of the Depth
Anything series, which is widely recognized by both academia and industry,
including adoption by Apple Core ML.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ee.columbia.edu/pipermail/quantum-ms/attachments/20250401/f24664dd/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Left-aligned blue cuee logo.png
Type: image/png
Size: 197821 bytes
Desc: not available
URL: <http://lists.ee.columbia.edu/pipermail/quantum-ms/attachments/20250401/f24664dd/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Bingyi.pdf
Type: application/pdf
Size: 15536197 bytes
Desc: not available
URL: <http://lists.ee.columbia.edu/pipermail/quantum-ms/attachments/20250401/f24664dd/attachment-0001.pdf>