专题链接

学术看板 当前位置:网站首页 > 专题链接 > 学术看板

机器学习与人工智能讲坛:Data61/NICTA屈立真博士报告通知

作者:统计机器智能与学习实验室 来源:统计机器智能与学习实验室 阅读次数:973日期:2018/07/02

一、主题:Two Tales of Language : Semantics and Diversity
二、主讲人:屈立真博士  Data61/NICTA

三、时间:2018年7月4日下午3:00
四、地点:
主楼B1-104

五、主持人:徐增林 计算机科学与工程学院教授 国家“青年千人”计划入选者

六、内容简介:

    In this talk, I will present our recent work on named entity recognition with few examples and generation of diverse language outputs.

    Most knowledge bases grow by extending the existing type hierarchy with new types. Usually such a knowledge base contains very few entities for those new types. Instead of having a large number of entities for each type, it is easy to manually construct a handful of example entity mentions for each novel type for model training. In this challenging setting, I will talk out two of our recent models, achieving at least doubled performance than the competitive baselines.

    In the second part, we investigate the diversity aspect of language generation. A natural characteristic of language is to express the same meanings with diverse expressions. We have looked at this model in two tasks: visual question generation and paraphrase generation. In the former task, the model we designed not only significant improve the diversity of generated questions, but also the semantic appropriateness compared to baselines.  In the latter task, we propose a simple method Diverse Paraphrase Generation (D-PAGE), which extends neural machine translation (NMT) models to support the generation of diverse paraphrases with implicit rewriting patterns. Our experimental results on two real-world benchmark datasets demonstrate that our model generates at least one order of magnitude more diverse outputs than the baselines in terms of a new evaluation metric Jeffrey's Divergence. We have also conducted extensive experiments to understand various properties of our model with a focus on diversity.

七、主讲人简介:

    屈博士是Data61/CSIRO(Commonwealth Scientific and Industrial Research Organization)的Research Scientist,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20多篇,H指数10,google引用406次。曾获得3rd Place in Data Mining Cup 2007和3rd Place in Data Mining Cup 2006。。

八、主办单位:计算机科学与工程学院

    承办单位:统计机器智能与学习实验室(SMILE Lab)

   2018年7月2日