Oct 27, 2023 Friday
Abstract:
Computer vision is witnessed significant progress recently and is widely used in applications in the real 3D world, such as autonomous driving, robots, etc. However, they systems have many security risks in the 3D world. On one hand, the visual recognition models are not generalizable under data distribution shifts (such as viewpoint changes, data noises, etc.). This talk will first introduce methods to evaluate and improve the natural robustness of the models in 3D environment. On the other hand, the visual recognition models also have insufficient robustness under adversarial attacks in the physical world. This talk will introduce attack methods to generate more natural and effective 3D adversarial examples in the physical world. Finally, I will introduce our recent work on robustness of multimodal foundation models.
Time:
Oct 27, 2023 Friday
11:00-11:50
Location:
RmE1-134, GZ Campus
Zoom:
628 334 1826 (PW: 234567)
Speaker Bio:
Dr. Yinpeng DONG
Postdoctoral Researcher, Department of Computer Science and Technology, Tsinghua University
Yinpeng Dong is a postdoctoral researcher in the Department of Computer Science and Technology, Tsinghua University. He received his BE and PhD degrees from Tsinghua University in 2017 and 2022, advised by Prof. Jun Zhu. His research interest includes machine learning, deep learning, and especially the adversarial robustness of deep learning. Yinpeng has published over 40 papers in the prestigious conferences and journals, including TPAMI, IJCV, CVPR, NeurIPS. He served as reviewers for ICML, NeurIPS, ICLR, CVPR, ICCV, TPAMI, etc. He received CCF Outstanding Doctoral Dissertation Award, Microsoft Research Asia Fellowship, Baidu Fellowship, ByteDance Scholarship, etc. He is supported by the Shuimu Tsinghua Scholar Program and China National Postdoctoral Program for Innovative Talents.