|
Tim Paek
Researcher Machine Learning and Applied Statistics |
Research Interests
My research passion is to foster and improve human interaction with automated systems, and in particular, those capable of engaging in conversational dialogue.
Systems that use speech as the primary means of communications go by various names, such as spoken dialogue systems (SDS), interactive voice response systems (IVR), voice user interfaces (VUI), and speech applications. All of these systems depend on the fundamental task of dialogue management, which concerns what actions or responses the system should take next. Because systems may be uncertain about what they heard, believe, and even what users might believe, I have focused my research on incorporating uncertainty into dialogue management. Furthermore, because human beings, without necessarily having perfect speech recognition, seem to be able to manage uncertainty in a conversation, I have focused on endowing systems with better error handling, both before and after they happen. Finally, because systems, like human beings, should be continually learning from their experiences and interactions, I have also focused on applying various kinds of machine learning methods, but only when it makes sense to do so. In many cases, I have found that good design can overcome tough engineering problems. As such, I have a strong interest in human-computer interaction as well as empirical and experimental validation of interaction techniques.
For all of the research I do, I aspire to have a practical impact on real-time, deployable systems. Because I strongly believe that mobile speech is where dialogue technology is likely to impact our everyday lives the most, I helped ship the latest Voice Command 1.6 product, a command-and-control speech application for mobile devices. Many of my papers in fact focus on mobile interaction. The following is a list of topics I have written papers on with respect to automated systems: spoken dialogue management, dialogue evaluation, user modeling, language modeling, personalization, mobile speech, voice user interface, reinforcement learning, decision-theory, and bayesian networks.
Background
I received a Ph.D. in Cognitive Psychology and a M.S. in Statistics from Stanford University in 2001 under Herbert H. Clark. Prior to that, I received a B.A. in Philosophy from the University of Chicago.
A complete C.V. is available upon request.
Publications
Journal Papers
- T. Paek & R. Pieraccini. Automating spoken dialogue management design using machine learning: An industry perspective. Speech Communication, Special Issue on Evaluating New Methods and Models for Advanced Speech-Based Interactive Systems , 2008, 50(8-9): 716-729.
- T. Paek & D. Chickering. Improving command and control speech recognition on mobile devices: Using predictive user models for language modeling. User Modeling and User-Adapted Interaction, Special Issue on Statistical and Probabilistic Methods for User Modeling, 2007, 17(1-2): 93-117.
- D. Chickering & T. Paek. Personalizing influence diagrams: Applying online learning strategies to dialogue management. User Modeling and User-Adapted Interaction, Special Issue on Statistical and Probabilistic Methods for User Modeling, 2007, 17(1-2): 71-91.
- E. Horvitz & T. Paek. Complementary computing: Policies for transferring callers from dialog systems to human receptionists. User Modeling and User-Adapted Interaction, Special Issue on Statistical and Probabilistic Methods for User Modeling, 2007, 17(1-2): 159-182.
- T. Paek & D. Chickering. Evaluating the Markov assumption in Markov decision processes for spoken dialogue management. Language Resources and Evaluation, Special Issue on Data Resources, Evaluation and Dialogue Interaction, 2006, 40(1): 47-66.
- E. Horvitz, C. Kadie, T. Paek, D. Hovel. Models of attention in computing and communication: From principles to applications. Communications of the ACM, 2003, 46(3): 52-59.
Refereed Conference, Workshop & Symposium Papers
- G. Shani, C. Meek, T. Paek, B. Thiesson & G. Venolia. Searching large indexes on tiny devices: Optimizing binary search with character pinning. Proceedings of Intelligent User Interfaces (IUI), 2009, pp. 257-266.
- T. Paek, B. Thiesson, Y. Ju & B. Lee.
Search Vox: Leveraging multimodal refinement and partial knowledge for mobile voice search.
Proceedings of User Interface Software and Technology (UIST), 2008, pp. 141-150.
Demo (2MB)
- Presentation given at Voice Search 2009.
- T. Paek & Y. Ju. Accommodating explicit user expressions of uncertainty in voice search or something like that. Proceedings of Interspeech, 2008.
- T. Paek, S. Gandhe & D. Chickering Rapidly deploying grammar-based speech applications with active learning and back-off grammars . Proceedings of the 9th SIGDIAL Workshop on Discourse and Dialogue, 2008.
- T. Paek, Y. Ju & C. Meek. People Watcher: A game for eliciting human-transcribed data for automated directory assistance. Proceedings of Interspeech, 2007. ELRA Best Paper Award.
- A. Kun, T. Paek & Z. Medenica. The effect of speech interface accuracy on driving performance. Proceedings of Interspeech, 2007.
- J. Sherwani, D. Yu, T. Paek, M. Czerwinski & A. Acero. VoicePedia: Towards speech-based access to unstructured information. Proceedings of Interspeech, 2007.
- T. Paek, S. Gandhe, D. Chickering & Y. Ju. Handling out-of-grammar commands in mobile speech interaction using backoff filler models. Proceedings of the ACL Workshop on Grammar-Based Approaches to Spoken Language Processing (SPEECHGRAM), 2007, pp. 33-40.
- T. Paek. Toward evaluation that leads to best practices: Reconciling dialogue evaluation in research and industry. Proceedings of the NAACL-HLT Workshop on Bridging the Gap: Academic and Industrial Research in Dialog Technologies, 2007, pp. 40-47.
- T. Paek & D. Chickering. The Markov assumption in spoken dialogue management. Proceedings of the 6th SIGDIAL Workshop on Discourse and Dialogue, 2005, pp. 35-44.
- T. Paek & R. Chandrasekar. Windows as a second language: An overview of the jargon project. Proceedings of the First International Conference on Augmented Cognition, 2005.
- T. Paek & E. Horvitz. Optimizing automated call routing by integrating spoken dialog models with queuing models. Proceedings of HLT-NAACL, 2004, pp. 41-48.
- T. Paek, S. Dumais & R. Logan. WaveLens: A new view onto internet search results. Proceedings of Computer-Human Interaction (CHI), 2004, pp. 727-734.
- T. Paek, M. Agrawala, S. Basu, S. Drucker, T. Kristjansson, R. Logan, K. Toyama & A. Wilson. Toward universal mobile interaction for shared displays. Proceedings of Computer Supported Cooperative Work (CSCW), 2004, pp. 266-269.
- T. Paek. Toward a taxonomy of communication errors. ISCA Workshop on Error Handling in Spoken Dialogue Systems, 2003, pp. 53-58.
- T. Paek & E. Horvitz. On the utility of decision-theoretic hidden subdialog. ISCA Workshop on Error Handling in Spoken Dialogue Systems, 2003, pp. 95-100.
- T. Paek. Empirical methods for evaluating dialog systems. ACL 2001 Workshop on Evaluation Methodologies for Language and Dialogue Systems, 2001, pp. 3-10.
- E. Horvitz & T. Paek. Harnessing models of users' goals to mediate clarification dialog in spoken language systems. Proceedings of the 8th International Conference on User Modeling (UM), 2001, pp. 3-13.
- T. Paek & E. Horvitz. Conversation as action under uncertainty.
Proceedings of the 16th Conference on Uncertainty in Artificial Intelligence (UAI), 2000, pp. 455-464.
Demo (32MB)
- T. Paek, E. Horvitz & E. Ringger. Continuous listening for unconstrained spoken dialog. Proceedings of the 6th International Conference on Spoken Language Processing (ICSLP), 2000.
- E. Horvitz & T. Paek. DeepListener: Harnessing expected utility to guide clarification dialog in spoken language systems. Proceedings of the 6th International Conference on Spoken Language Processing (ICSLP), 2000.
- T. Paek & E. Horvitz. Uncertainty, utility, and misunderstanding: A decision-theoretic perspective on grounding in conversational systems. AAAI Fall Symposium on Psychological Models of Communication in Collaborative Systems, 1999.
- E. Horvitz & T. Paek. A computational architecture for conversation. Proceedings of the 7th International Conference on User Modeling (UM), 1999, pp. 201-210.
Miscellaneous
- T. Paek. Introduction to Dialogue Management and Evaluation. Lecture for the Johns Hopkins University Summer School on Human Language Technology, July 10. 2007.
- T. Paek. Reinforcement learning for spoken dialogue systems: Comparing strengths and weaknesses for practical deployment. Interspeech-06 Workshop on "Dialogue on Dialogues - Multidisciplinary Evaluation of Advanced Speech-based Interacive Systems", 2006.
- T. Paek & E. Horvitz. Grounding criterion: Toward a formal theory of grounding. Microsoft Technical Report, MSR-TR-2000-40, April 2000.
Professional
I am currently serving as the President of (SIGDIAL), the Special Interest Group on Discourse and Dialogue for ACL and ISCA. Even though I work in industry, I enjoy interacting with Ph.D. students and am on the Advisory Committee for the Young Researchers' Roundtable on Spoken Dialogue Systems. If you are a motivated, creative, and hard-working Ph.D. student looking for an internship, please feel free to contact me.
In the past, I co-organized the ACL 2008 Workshop on Mobile Language Processing and the NAACL 2001 Workshop on Adaptation in Dialogue Systems.
Professional activities for 2009:
- Co-Organizer, SiMPE (Workshop on Speech in Mobile and Pervasive Environments) 2009.
- Reviewer, EACL 2009.
- Reviewer, ACL-IJCNLP 2009.
- Reviewer, NAACL-HLT 2009.
- Program Committee, IJCAI 2009 Workshop on Knowledge and Reasoning in Practical Dialogue Systems.
- Program Committee, Automotive User Interfaces 2009.
- Program Committee, SIGDIAL 2009.
- Program Committee, Intelligent User Interfaces 2009.
- Scientific Review Committee, Interspeech 2009.
Last updated: January, 2009
E-mail: timpaek at microsoft dot com
Directions: Microsoft Research
U.S.Mail: One Microsoft Way, Redmond WA,
98052-6399, USA
Tel: 425.703.8647
Fax: 425.706.7329 (This is the main MS FAX number so please use "Attention: Tim Paek")
