|
|
|
|
|
|
| PROGRAM |
|
|
- Feb 12 (Tue):
- 10:30-10:50 Opening Remarks and Project Introduction
- Tsujii, Jun'ichi
- 10:50-12:10 Session I: New Models for NLP
- Haghighi, Aria, University of California at Berkeley, USA Slides
- "Latent Variable Models in NLP"
- Okanohara, Daisuke, University of Tokyo, Slides
- "Dualized L1-regularized Log-Linear Models and Its Application in NLP"
- 12:10-14:00 Lunch Break
- 14:00-16:00 Session II: Information Extraction
- Sagae, Kenji, University of Tokyo, Japan
- "Predicate-argument analysis with DAG parsing"
- McDonald, Ryan, Google, USA, Slides
- "Leveraging User Annotations in Sentiment Summarization"
- Liu, Tie-Yan, Microsoft Research Asia, China, Slides
- "Learning to Rank: From Pairwise Approach to Listwise Approach"
- 16:00-16:20 Break
- 16:20-17:40 Session III: Structured Output Learning for NLP
- Tsuboi, Yuta, IBM Tokyo Research Laboratory, Japan, Slides
- "Conditional Random Fields Incorporating
Incomplete Annotations"
- Suzuki, Jun, NTT Communication Science Laboratories, Japan, Slides
- "Semi-supervised Structured Output Learning"
- Feb 13 (Wed):
- 10:00-12:00 Session IV: Synctactic Parsing
- Matsuzaki, Takuya, University of Tokyo, Japan, Slides
- "Comparative parsing evaluation across different grammar frameworks"
- Miyao, Yusuke, University of Tokyo, Japan, Slides
- "Towards Framework-Independent Evaluation of Syntactic Parsing"
- Smith, Noah, Carnegie Mellon University, USA, Slides
- "Two Topics in Statistical NLP: the Jeopardy Model and an M-Estimator"
- 12:00-14:00 Lunch Break
- 14:00-16:00 Session V: Bayesian Models
- Kurihara, Kenichi, Tokyo Institute of Technology, Japan, Slides
- "Collapsed Variational Inference for Hierarchical Dirichlet Process"
- Sato, Issei, University of Tokyo, Japan, Slides
- "Nonparametric Bayesian Approach for The Distributional Hypothesis"
- Mochihashi, Daichi, NTT Communication Science Laboratories, Japan, Slides
- "Present and Future of a Text Modeling"
- 16:00-16:20 Closing Remarks
- Each invited speaker will have 30 minutes for her/his talk, which is
followed by 10 minutes Q/A.
|
|
