loaction
SNU Professor Byung-Gon Chun's Team Receives the 'Google Research Award' for Their World-Renown AI System Research
-
작성자
관리자
-
등록일
2020.12.31
-
조회수
463
SNU Professor Byung-Gon Chun's Team Receives the 'Google Research Award' for Their World-Renown AI System Research
- The award is to be announced in October this year, along with the funding Of 30,000 USD for research.
- Recognized for research results that have enhanced the performance and usability of AI systems
- Recognized for research results that have enhanced the performance and usability of AI systems
▲ (From Left) M.S. student Yunmo Koo, Ph.D. student Eunji Jeong, Professor Byung-Gon Chung of the SNU Department of Computer Science and Engineering, M.S. student Taebum Kim
Professor Byung-Gon Chun's research team of the SNU Department of Computer Science and Engineering won the 'Google Research Award' in October for their research on AI systems. Google, which was founded at Stanford University, considers technological interaction within the academia to be of high importance and selects to award research teams among many research groups around the world that have demonstrated outstanding research work.
In addition to the award, Professor Chun's research team will receive 30,000 USD worth of research funds from Google.
The underlying research behind this award involves the study of increasing the performance and usability of artificial intelligence systems. Recently, artificial intelligence technology has shown rapid development, yielding performances that are closely comparable to humans in many fields such as in image and voice processing. The role of artificial intelligence systems to swiftly and easily learn artificial neural network models is critical for further developments to take place.
However, in artificial intelligence systems, the facile representation of models and quick learning are two concepts that are at odds with each other. There are two main ways to execute mathematical operations of artificial neural network models, the first being symbolic graph-based systems (a method involving the formation and use of graphs consisting of symbolized calculations) which can rapidly learn artificial neural network models of fixed structures but show difficulties in the representation of models with diverse structures.
Conversely, in an imperative system (where calculations are immediately executed), various artificial neural network models can be easily created, but it takes longer for them to be learned.
Professor Chun's research team has been working on combining the advantages of both systems since 2017. Their research work has been conducted with the support of global companies such as Amazon and Samsung Electronics, and in 2019, they have developed a system called 'Janus' that can easily conduct artificial intelligence research while dramatically reducing the time it takes to conduct their experiments. Using the award-winning and the interaction with Google as a launch pad, the research team plans to develop a new system by further upgrading their research content.