梭哈 成电新闻网梭哈

梭哈梭哈怎么玩:Deep Learning with Limited Training Data

文:教师发展中心 / 来源:人力资源部 / 2017-07-05 / 点击量:1311

  本次梭哈怎么玩活动教师发展中心特别邀请来自澳大利亚Data61/CSIRO的屈立真博士,与梭哈游戏师生分享他在深度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7月10日(周一)10:00

  二、地 点:清水河校区经管楼宾诺咖啡

  三、主 题:Deep Learning with Limited Training Data

  四、主讲人:屈立真博士(Data61/CSIRO Research Scientist)

  五、主持人:国家“青年千人”计划入选者 徐增林教授

  六、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)

  七、交流内容:

  Deep learning models often require large amount of manually labelled noise-free training data. However, due to the expensive and time-consuming process of building training datasets, those datasets in many businesses are in fact small or noisy.  In my talk, I’ll discuss three of our recent work, which focus on training deep neural networks with small or noisy training datasets. I will cover our novel transfer learning method for named entity recognition. For novel types in new domains, conventional supervised NER models often require substantial labelled data. Our method is able to leverage existing labelled data for known types to significantly reduce the amount of training data for new types. The resulted model achieved up to 160% improvement over the strongest baseline based on merely 125 target-domain training data.

  八、主讲人简介:

  屈立真博士是Data61/CSIRO(Commonwealth Scientific and Industrial Research Organization)的Research Scientist,博士毕业于德国马克斯-普朗克研究所,是Program Committee of Annual Meeting of the Association for Computational Linguistics (ACL)、ACM Conference of Information and Knowledge Management (CIKM)、Conference on Empirical Methods in Natural Language Processing (EMNLP)。在包括NIPS16、ACL17、IJCAI17、CVPR17、CIKM11等机器学习领域顶级会议上发表论文19篇,H指数8,google引用282次。特邀报告8次,包括2015年在悉尼大学the Machine Learning Summer School 2015的教学报告;曾获得3rd Place in Data Mining Cup 2007和3rd Place in Data Mining Cup 2006。


                    人力资源部教师发展中心

                      2017年7月3日


编辑:罗莎  / 审核:林坤  / 发布者:林坤

梭哈游戏 梭哈 梭哈怎么玩