Accurate machine translation of human languages (e.g., Chinese to English) is a longstanding challenge for computer science. Recently, deep learning approaches have made a significant impact on this field, greatly simplifying systems, but also yielding opaque, hard-to-analyze solutions. This talk will investigate what linguistic information neural networks choose to represent when they learn from large quantities of human translation data. It will also speculate about the power of recurrent networks as use in practice, both in general and with respect to how they are currently trained.
Prof. Kevin Knight is Dean's Professor of Computer Science at the University of Southern California (USC) and a Research Director and Fellow at USC's Information Sciences Institute (ISI). He received a PhD in computer science from Carnegie Mellon University and a bachelor's degree from Harvard University. Prof. Knight’s research interests include machine translation, natural language generation, automata theory, and decipherment. He has taught computer science at USC for more than 25 years, authored over 150 research papers on natural language processing, and received several best paper awards. Prof. Knight also co-authored the widely-adopted textbook "Artificial Intelligence" (McGraw-Hill). In 2001, he co-founded Language Weaver, Inc., which provides machine translation software and solutions, and in 2011, he served as President of the Association for Computational Linguistics (ACL). He served as General Chair for the Annual Conference of the ACL in 2005, and General Chair for the North American ACL conference in 2016. He is a Fellow of the ACL, a Fellow of ISI, and a Fellow of the Association for the Advancement of Artificial Intelligence (AAAI).