Highlight

• Participated Netflix competition and is currently ranked at 3rd place amont more than 16000 teams.

• Designed fully automatic algorithms to estimate human gender and age from images taken under complex real life environment. The gender classification accuracy is more than 90 percent and the average age estimation error is around 5 years.

• Designed core algorithms for video surveillance system: foreground detection, human/vehicle detection&classification, multiple object tracking. The system has been running in San Francisco Aiport for years and helped the airport identifying real security violation many time. The developed system is the basis of successful start-up company focusing on video surveillance. Received NEC Technology Impact Award because of this.

• Designed a computer automated stock trading platform. Designed a simple statitical arbitrage trading strategy. The platform and stragegy have been used by Wei Xu himself in REAL UNATTENDED stock trading and so far have profited him a small amount of money. (This effort is temporarily discontinued because of participating Netflix competition)
 
• Designed algorithms to automatically identity different speakers from audio recording, which can be used to organize and browse audio archive.

• Designed a new dimension reduction algorithm called concept factorization and applied it to document clustering. The clustering accuracy compares favorably with other major clustering algorithms

• Researched and developed natural language based (in contrast to menu-based) human computer telephone conversational system for domain specific tasks such as travel reservation for flight, hotel and rental car.

Education

M.S.(2000) Language Technologies Institute, School of Computer Science, Carnegie Mellon University

B.S.(1998) Department of Computer Science, Tsinghua University

Publication

Mei Han, Wei Xu, Yihong Gong, Video Object Segmentation by Motion-based Sequential Feature Clustering, ACM Multimedia 2006

Dan Kong, Mei Han, Wei Xu , Hai Tao, and Yihong Gong, “A conditional random field model for video super-resolution,” in Proc. International Conference on Pattern Recognition, vol. 3, pp. 619-622, 2006

Huazhong Ning, Wei Xu, Yihong Gong, Thomas S. Huang Improving Speaker Diarization by Cross EM Refinement, IEEE International Conference on Multimedia & Expo (ICME) 2006

Xiang Ji, Wei Xu, Shenghuo Zhu, "Document Clustering with Prior Knowledge", in the proceedings of29th Annual International ACM SIGIR Conference on Research & Development on Information Retrieval, Seattle 2006

Shenghuo Zhu, Xiang Ji, Wei Xu and Yihong Gong, "Multi-lablelled Classification Using Maximum Entropy Method" In Proceedings of Annual ACM Conference on Research and Development in Information Retrieval (SIGIR'05) , 2005

Hui Han, Wei Xu, Hongyuan Zha, Lee Giles, “A Hierarchical Naive Bayes Model for Name Disambiguation in Author Citations”, to appear in the proceedings of the 20th Annual ACM Symposium on Applied Computing Special Track on Information Access and Retrieval (SAC-IAR 2005)

Yue Zhou, Wei Xu, Hai Tao, Yihong Gong, “Background Segmentation Using Spatial-Temporal Multi-Resolution MRF ”  in the proceedings of IEEE Workshop on Motion & Video Computing, 2005

Yihong Gong, Mei Han, Wei Hua, Wei Xu, “Maximum Entropy Model-Based Baseball Highlight Detection and Classification”, International Journal of Computer Vision and Image Understanding, 96 (2004) 181-199

Wei Xu, Yihong Gong, “Document Clustering by Concept Factorization”, in the proceedings of the 27th ACM SIGIR Conference on Research and Development in Information Retrieval, Sheffield, UK (SIGIR 2004)

Mei Han, Wei Xu, Hai Tao, Yihong Gong, “An Algorithm for Multiple Object Trajectory Tracking”, International Conference on Computer Vision and Pattern Recognition (CVPR 2004)

Wei Xu, Xin Liu, Yihong Gong, "Document Clustering Based on Non-negative Matrix Factorization", in the proceedings of 26th ACM SIGIR Conference on Research and Development in Information Retrieval, Toronto (SIGIR 2003)

Yi Zhang, Wei Xu, Jamie Callan, “Exploration and Exploitation in Adaptive Filtering Based on Bayesian Active Learning” In the proceedings of 2003 International Conference on Machine Learning, Washington (ICML 2003)

Xin Liu, Yihong Gong, Wei Xu, Shenghuo Zhu, "Document Clustering with Cluster Refinement and Model Selection Capabilities", in the proceedings of 25th ACM SIGIR Conference of Research and Development in Information Retrieval, Tempare, Finland (SIGIR 2002)

Yi Zhang, Wei Xu, Jamie Callan, “Exact Maximum Likelihood Estimation for Word Mixtures” Text Learning Workshop in International Conference on Machine Learning (ICML), Sydney, Australia, 2002

Xiaoyan Zhu, Yu Wang, Wei Xu, "Speech Recognition Model Based on Recurrent Neural Networks", Chinese Journal of Computers, Vol. 24, No 2: 213-218, Feb. 2001

Wei Xu, Alex Rudnicky, “Language Modeling for Dialog System”, 6th International Conference on Spoken Language Processing, Beijing, China (ICSLP2000)

Wei Xu, Alex Rudnicky, “Can Artificial Neural Networks Learn Language Models?” 6th International Conference on Spoken Language Processing, Beijing, China (ICSLP2000)

Alex Rudnicky, Eric Thayer, Paul Constantinides, Chris Tchou, Rande Shern, Kevin Lenzo, Wei Xu, Alice Oh, “Creating Natural Dialogs in the Carnegie Mellon Communicator System” Proceedings of 6th European Conference on Speech Communication and Technology, Budapest, Hungary (Eurospeech99)

Wei Xu, Alex Rudnicky, “Task-based dialog management using an agenda” ANLP/NAACL 2000 Workshop on Conversational Systems, May 2000

Alex Rudnicky, Wei Xu, “An agenda-based dialog management architecture for spoken language system” The 1999 international workshop on Automatic Speech Recognition and Understanding (ASRU’99)

Wei Xu , Xiaoyan Zhu “A New Model of Recurrent Neural Networks for Speech Recognition” 5th International Conference on Neural Information Processing, Kitakyushu, Japan (ICONIP'98)