On April 29, 2019

Speaker

오인수 (넷마블AI센터(NARC) / 실장)


Title

Developing a Reinforcement Learning Based AI Agent for Better Games


Abstract

Many companies are making efforts to innovate their business and find new markets through artificial intelligence technology. Though the hype curve of the recent new technology is decreasing, the impact of artificial intelligence technology in the industry is growing, and games, in particular, can provide a flexible environment where AI technology can be actively applied. Netmarble, as a leader of game industry, is utilizing AI technology that can improve game fun for users.

This session provides an overview of reinforcement learning algorithms and how they are applied to the field of game production to create more efficient and better games. In addition to recent achievements of AI agent project in OpenAI and Deepmind, some projects in Netmarble will be covered with some applications of academic papers. It also will include the lessons from several use cases of reinforcement learning in game development as a user- facing AI and a testing assistant.


Bio

Insoo Oh is the head of Magellan(Game AI) under NARC at Netmarble. He received his B.S degree in Naval Architecture and Ocean Engineering from Seoul National University, and MS Eng. In Naval Architecture and Marine Engineering from University of Michigan at Ann Arbor. He has been a game server architecture, programmer and CTO in Netmarble NPark over 15 years and participated in developing MaguMagu, Magu the real, Clash of the Soccer, ASDF Hoverboard. Also, he was awarded with CJ only one grand prize for achievement and Deloitte Technology Fast 500 award. His recent main interests are in making intellectual agents with reinforcement learning, and improving game development process with machine learning.