Conference: Perceptrons and Syntactic Structures at 60: Computational Modeling of Language

会议：60 岁的感知器和句法结构：语言的计算建模

• 批准号: 1651142
• 负责人: Joseph Pater
• 金额: $ 2.42万
• 依托单位: University of Massachusetts Amherst
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-08-01 至 2019-01-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1651142&HistoricalAwards=false
• 关键词: Conference; Perceptrons; Syntactic; Structures; 60

This workshop will bring together leading researchers in cognitive science and artificial intelligence who specialize in the integration of linguistic theory with statistical approaches, especially neural networks. Neural networks have been important in many of the recent advances in language technologies (in what's called "deep learning"), and the greater integration of linguistic structure into these models promises to lead to further breakthroughs. The main focus of this meeting will be on how integration of models of linguistic structure with probabilistic learning theories may lead to a deeper understanding of the way that humans process and represent language. This sort of integration has been difficult to achieve in the past in part because of the separation of researchers in each tradition into different disciplines interacting in different conferences. In-depth analysis of the structure of human languages is conducted in mostly in Linguistics, while neural network modeling and other statistical learning research is conducted mostly in Psychology and Computer Science. The workshop will be held as part of the inaugural meeting of the Society for Computation in Linguistics, taking place concurrently with the meeting of the Linguistic Society of America. It will thus bring researchers from other disciplines into contact with linguists, and will stimulate productive intellectual exchange.

本次研讨会将汇集认知科学和人工智能领域的领先研究人员，他们专门研究语言学理论与统计方法的整合，特别是神经网络。神经网络在语言技术（所谓的“深度学习”）的许多最新进展中发挥了重要作用，将语言结构更好地整合到这些模型中有望带来进一步的突破。本次会议的主要重点将是如何整合语言结构模型与概率学习理论可能会导致人类处理和表示语言的方式有更深的理解。这种整合在过去很难实现，部分原因是每个传统的研究人员都被分成不同的学科，在不同的会议上进行互动。对人类语言结构的深入分析主要在语言学中进行，而神经网络建模和其他统计学习研究主要在心理学和计算机科学中进行。该研讨会将作为语言学计算学会成立大会的一部分，与美国语言学会会议同时举行。因此，它将使其他学科的研究人员与语言学家接触，并将促进富有成效的智力交流。