Keynote Speakers
The following speakers have graciously accepted to give keynotes at CCMT-2020.
Graham Neubig
Title: Beyond Likelihood: New Training Objectives for Neural Machine Translation
Abstract: Maximum likelihood estimation (MLE) is the workhorse of training NMT models, but has a number of issues such as disregard for the actual test-time decoding algorithm, lack of consideration of the inherent ambiguity in the generation of translation references, and inability to deal with training/test-time data distribution mismatch. In this talk I will discuss some new developments in training objectives for machine translation that move beyond the standard paradigm of training with maximum likelihood estimation. Specifically I will first discuss a method that explicitly trains models to maximize the semantic similarity between MT outputs and the human-provided references. Second, I will discuss methods that automatically learn to appropriately weight training data to maximize test-time performance, including in multilingual learning settings.
Bio: Graham Neubig is an associate professor at the Language Technologies Institute of Carnegie Mellon University. His work focuses on natural language processing, specifically multi-lingual models that work in many different languages, and natural language interfaces that allow humans to communicate with computers in their own language. Much of this work relies on machine learning, and he is also active in developing methods and algorithms for machine learning over natural language data. He publishes regularly in the top venues in natural language processing, machine learning, and speech, and his work has won awards at EMNLP 2016, EACL 2017, and NAACL 2019.
韦福如
Title: 自然语言预训练模型进展
Abstract: 大规模预训练语言模型很大程度上改变了自然语言处理模型的研究和开发范式，在工业界和学术界都引起了广泛的关注。本报告将对现有的语言模型预训练工作进行总结和比较，然后介绍面向自然语言理解和生成任务的统一预训练语言模型UniLM以及多语言预训练模型InfoXLM，并就未来面临的挑战和进一步的研究方向进行讨论和展望。
Bio: 韦福如博士，微软亚洲研究院自然语言计算组首席研究员，长期从事自然语言处理的基础研究和技术创新。在自然语言处理领域重要会议和期刊发表论文100余篇，被引用9000余次，多项研究成果转化到微软重要产品中。入选2017年《麻省理工科技评论》中国区“35岁以下科技创新35人”榜单，2019年第六届世界互联网大会“领先科技成果”奖。近年来，团队开发的预训练模型（UniLM, InfoXLM, LayoutLM, MiniLM等）被广泛应用于微软的产品中。