Zhizhong Li (?)
University of Illinois, Urbana-Champaign (UIUC)
Email: zli115 [at] illinois [dot] edu
Office: Siebel Center 3307
For machine learning algorithms such as a deep neural network, it is not uncommon to train separate networks for each vision task ad-hoc. In contrast, the human brain is simultaneously able to perform a multitude of vision tasks, such as object classification, fine-grained classification, depth estimation, etc. Knowledge in these tasks help gaining insight in one another. Constructing a learning algorithm that is aware of the extent of its visual knowledge, gradually grows it, and adapts to new circumstances is an interesting idea, and a grand vision in AI agent design.
Before coming to UIUC, I completed a MS in the Robotics Institute at Carnegie Mellon University, where I am supervised by Dr. Daniel Huber. My research focused on Infrastructure Component Recognition using Cross-domain Learning to account for dataset shift, within the Aerial Robotic Infrastructure Analyst (ARIA) project. I obtained my bachelor’s degree from Department of Automation, Tsinghua University in Beijing, China. During my undergraduate study I finished my graduation dissertation on Stick Figure Body Parts Recognition with Dr. Changshui Zhang. I also accomplished a project with a 6 DoF robot involved, under Dr. Zongying Shi‘s guidance.
Ultimately, my research interest lies in Computer Vision, especially its intersection with Machine Learning. My curriculum vitae can be found here [PDF] (updated Nov. 2018).
* My first name can be pronunced “Gee-Jone” where Jone is Jones without “s” or drone without “r”. My Chinese name is 李之仲.