News

【演講】敬邀參加1月16日(日)14:30~16:30@工程三館427室/講者:Mr. Kuang-Huei Lee (Google Brain)/講題:Learn representations that generalize for vision and RL.

吳毅成教授邀請來自Google Brain的學者Mr. Kuang-Huei Lee來交大分享他的研究,
歡迎各位老師們及實驗室的學生抽空前往蒞臨參加,非常感謝!

時間: 111.1.16(日) 14:30~16:30
地點: 工程三館 427室
演講者:Mr. Kuang-Huei Lee (Google Brain)

演講題目:Learn representations that generalize for vision and RL.

演講摘要:In almost all ML application domains, using capable models and expressive learning objectives to absorb large amounts of diverse data has now become a common narrative of generalization success. In this talk, Kuang-Huei will explore the data and objective aspects of this narrative for vision and RL. He will discuss several ideas for self-supervised representation learning, improving uses of agent experience and simulator, the sim-to-real problem, and future challenges. The goal of the presentation is to motivate scientists to make the next generalization breakthrough by rethinking the data and objectives that they use for learning vision and RL models.
 
講者介紹:Kuang-Huei Lee is a research engineer at Google Brain. His research focuses on RL, information theory, and representation learning for perception and decision making. He has been involved in various robotics learning projects at Google. From 2016 to 2019, he was a research engineer at Microsoft where he works on computer vision and related products. He received his graduate degree from Carnegie Mellon University in computer science and his undergraduate degree from National Taiwan University.