加入收藏夹
联系我们
关于本站
个人主页
西电导航
西安电子科技大学
    当前位置:首页>>学术讲座
Optimizing Device Placement in Machine Learning Workloads using Deep Reinforcement Learning
时间:2019-04-09 10:33    点击:   所属单位:通信工程学院
讲座名称 Optimizing Device Placement in Machine Learning Workloads using Deep Reinforcement Learning
讲座时间 2019-04-22 14:30:00
讲座地点 西电北校区新科技楼1012会议室
讲座人 Prof. Baochun Li
讲座人介绍
Baochun Li received his B.Engr. degree from the Department of Computer Science and Technology, Tsinghua University, China, in 1995 and his M.S. and Ph.D. degrees from the Department of Computer Science, University of Illinois at Urbana-Champaign, Urbana, in 1997 and 2000.  Since 2000, he has been with the Department of Electrical and Computer Engineering at the University of Toronto, where he is currently a Professor. He holds the Bell Canada Endowed Chair in Computer Engineering since August 2005. His research interests include cloud computing, distributed systems, datacenter networking, and wireless systems.
Dr. Li has co-authored more than 300 research papers, with a total of over 17000 citations, an H-index of 75 and an i10-index of 233, according to Google Scholar Citations. He was the recipient of the IEEE Communications Society Leonard G. Abraham Award in the Field of Communications Systems in 2000. In 2009, he was a recipient of the Multimedia Communications Best Paper Award from the IEEE Communications Society, and a recipient of the University of Toronto McLean Award. He is a member of ACM and a Fellow of IEEE.
讲座内容
Training deep neural networks requires an exorbitant amount of computation resources, including a heterogeneous mix of GPU and CPU devices. It is critical to place operations in a neural network on these devices in an optimal way, so that the training process can complete within the shortest amount of time. The state-of-the-art in the literature uses a deep reinforcement learning method based on policy gradients to solve this problem, but we believe that there remains ample room for further improvements. In this talk, I will present our recent work published in ICML 2018 and NeurIPS 2018 that uses proximal policy optimization (PPO) and cross-entropy minimization to achieve significantly better performance than the state-of-the-art. Our experiments with several popular neural network training benchmarks have demonstrated clear evidence of superior performance: with the same amount of learning time, our algorithm leads to placements that have training times up to 60% shorter. This talk will be targeting a general audience, and will therefore include a brief tutorial on basic ideas in reinforcement learning algorithms.
转载请注明出处:西安电子科技大学学术信息网
如果您有学术信息或学术动态,欢迎投稿。我们将在第一时间确认并收录,投稿邮箱: meeting@xidian.edu.cn
Copyright © 2011-2019 西安电子科技大学 
开发维护:电子工程学院网络信息中心  管理员:meeting@xidian.edu.cn 站长统计: