Ganghee Jang, Oregon Tech
Will present a talk entitled:
Applying Reinforcement Learning – From Image Processing to Control
Abstract: In this talk, Dr. Jang will share his recent research in the domain of Reinforcement Learning. His journey started with a project at Korea University during one of his post-doc research positions. In the medical domain, it is not easy to get high-quality ground truth for training. The amount of ground truth becomes even more scarce despite huge data set size, for example, for connectomes. The main idea of the project is to apply Deep Reinforcement Learning for the design of automatic error detection and correction of segmentation results from human or other segmentation algorithms. Multi-agents for the detection and correction were to mimic those of humans via an environment that provided artificial errors during an episode. After the project, he moved to a company, named Cortek, where he led projects to revive the US legendary guitar processor brand, Digitech. He tightly worked with world-top OEM manufacturers and saw industry-leading automated factories. Once he started his current position at Oregon Tech, he wanted to bring the experience and train students to find their jobs in new opportunities related to highly automated factories or machinery controlled by AI. Now his research interest is Smart Edge considering applications for Smart Factory, Smart City, etc. He designed research tasks for students, for example, his recent grant from NASA for the solar panel cleaning system, and even planned to make a new course to teach Deep Reinforcement Learning algorithm development using a simulator.
Bio: Ganghee Jang received his Ph.D. from the University of California, Irvine, majoring in Electrical and Computer Engineering. His focus of study was Computer Architecture, but since graduation, he has worked at various research/industry positions before joining Oregon Tech. The projects he participated in are from various domains such as one in the domain of biometrics, a Deep Reinforcement Learning project to apply to medical imaging, a High-Performance-Computing algorithm for multi-GPGPUs, effect processors for musical instruments, etc. Also, during his role as a product manager at Cortek in South Korea, he successfully performed many tasks to revive the legendary US guitar effect processor brand, Digitech. As his teaching interest is advising students’ project classes, he believes all these experiences will help advise students’ individual projects. His current research interests are embedded platforms for AI-driven applications such as robotics, musical instruments, mobility, etc.
In his leisure time, he likes to play guitar and read articles on musical instruments. Sometimes, he plays badminton games with his family. As a father of three kids, he likes the beautiful and safe environment at Klamath Falls.
CAMP 176
Tuesday, November 19, 2024
12:15pm
*Co-Sponsored by IEEE student branch and HKN
________________________________________________________________
Electrical and Computer Engineering l CLARKSON UNIVERSITY l Potsdam, New York 13699-5720