Speech and Language Technologies for Disaster Management and Emergency Response-Silicon Valley Campus - Carnegie Mellon University

Speech and Language Technologies for Disaster Management and Emergency Response

Monday May 23, 3:00 pm, Bldg 23, Rm 109

Ian Lane, Assistant Research Professor, CMU

Summary:Technologies for speech and language processing offer a unique set of tools to support users in Disaster Management and Emergency Response teams. From data-mining of large text data sets, such as live twitter feeds, to information access in hands-and-eyes busy environments, the possibilities for supporting and enhancing current operational procedures are numerous. However, there are a number of critical limitations that remain before such systems can be effectively deployed in the field. One critical issue is system adaptability. Typical language understanding systems are built for well-known domains, with development processes that are performed over many months of iterative data collection and system testing. While, effective for long standing tasks, they are unsuitable for situations such as disaster response where the language used in the field varies significantly between events. In this talk I will present our initial work on rapid system adaption using intelligent crowd-sourcing technologies and field adaptive spoken language understanding systems. I will complete the talk by introducing a number of research prototypes developed by our team, including intelligent crowd-sourcing tools for rapid data mining of real-time observations, speech-based interactive control of robots, and speech-to-speech translation systems for first responders.

Presentation slides (.pdf)

About the speaker:Ian Lane is a Research Assistant Professor at Carnegie Mellon University. His research interests include spoken language understanding, speech recognition, machine learning and applications of these technologies. He has published extensively in these fields and has received several patents for his work. At Carnegie Mellon University his research efforts have focused on robust integration of speech recognition and machine translation, unsupervised topic adaptation and optimization of speech-to-speech translation systems for mobile devices. During his time at CMU he led the development of numerous world-leading speech translation systems including submissions to IWSLT, TransTAC and GALE. Before joining Carnegie Mellon University, Ian was an intern researcher at ATR Spoken Language Communication Laboratories in Kyoto, Japan. He obtained a Ph.D. degree from Kyoto University in 2006 and a B.Tech. degree in 2000 from Massey University, New Zealand.