学术沙龙:Generative Adversarial Networks as Variational Training of Energy Based Models

文:教师发展中心 / 来源:人力资源部 / 2017-01-05 / 点击量:2604

  为加强我校各学科之间的学术交流,搭建教师学术交流平台,促进教师学术水平提升和跨学科合作,教师发展中心开展跨学科学术沙龙活动。

  本次活动教师发展中心特别邀请来自纽约州立大学宾汉姆顿分校的翟双飞博士,与我校师生分享他在机器学习领域的研究心得。具体安排如下,欢迎感兴趣的师生参加:

  一、时 间:2017年1月6日(周五)上午10:00

  二、地 点:清水河校区主楼B1-104会议室

  三、主 题:Generative Adversarial Networks as Variational Training of Energy Based Models

  四、主讲人:翟双飞 博士(纽约大学宾汉姆顿分校)

  五、主持人:国家“青年千人”计划入选者  徐增林教授

  六、主办单位:人力资源部教师发展中心

  七、承办单位:计算机科学与工程学院(统计机器智能与学习实验室(SMILE Lab))  

  八、交流内容:

  We study deep generative models for effective unsupervised learning. We propose VGAN, which works by minimizing a variational lower bound of the negative log likelihood (NLL) of an energy based model (EBM), where the model density p(x) is approximated by a variational distribution q(x) that is easy to sample from. The training of VGAN takes a two step procedure: given p(x), q(x) is updated to maximize the lower bound; p(x) is then updated one step with samples drawn from q(x) to decrease the lower bound. VGAN is inspired by the generative adversarial networks (GANs), where p(x) corresponds to the discriminator and q(x) corresponds to the generator, but with several notable differences. We hence name our model variational GANs (VGANs). VGAN provides a practical solution to training deep EBMs in high dimensional space, by eliminating the need of MCMC sampling. From this view, we are also able to identify causes to the difficulty of training GANs and propose viable solutions.

  九、主讲人简介:

  Shuangfei Zhai is currently a final year Ph.D student in Multimedia Research Lab, Department of Computer Science, Binghamton Univeristy, SUNY, where he work with Prof. Zhongfei (Mark) Zhang. Before coming to Binghamton University, he obtained my B.E. in Electronic Engineering and Information Science in University of Science and Technology of China, Hefei in 2010. He  was a master student in Chinese Academy of Sciences during 2010-2012. 

  He is now on job market looking for a Machine Learning/Deep Learning Research Scientist position. 


                   人力资源部教师发展中心

                     2017年1月5日


编辑:林坤  / 审核:林坤  / 发布者:林坤