) |
I am a PhD student at Language Technology Institute, Carnegie Mellon University, from fall 2012. Previously, I graduated with a bachelor's degree in computer science from Zhejiang University, China. During my undergraduate study, I mainly under the supervision of Prof. Deng Cai in the State Key Laboratory of CAD & CG. I was a summer intern at UCLA in 2011, mainly work with Prof. Jenn Wortman Vaughan. My Email address is: xinleic at cs dot cmu dot edu. I am interested in the exploration and analysis of intelligence with different tools, as well as the design and application of intelligent algorithms to solve problems in the real world. |
Education
2012.8-present-
Artificial Intelligence, Carnegie Mellon University
-
Computer Science & Advanced Engineering, Zhejiang University, China
Publications
Conference Papers
-
Jiajun Lv, Xinlei Chen, Jin Huang, Hujun Bao. Semi-supervised Mesh Segmentation and Labeling. The 20th Pacific Conference on Computer Graphics and Applications (PG), 2012. [PDF]
-
Xinlei Chen, Zifei Tong, Haifeng Liu, Deng Cai. Metric Learning with Two-Dimensional Smoothness for Visual Analysis. The 25th IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2012. [PDF]
-
Xinquan Qu, Xinlei Chen. Sparse Structured Probabilistic Projections for Factorized Latent Space. The 20th ACM Conference on Information and Knowledge Management (CIKM), 2011. [PDF]
-
Xinlei Chen, Deng Cai. Large Scale Spectral Clustering with Landmark-based Representation. The 26th AAAI Conference on Artificial Intelligence (AAAI), 2011. [PDF][Code]
-
Xinlei Chen, Xinquan Qu, Zijian Li. Image Analysis with Nonlinear Adaptive Dimension Reduction. The 3rd International Conference on Internet Multimedia Computing and Service (ACM ICIMCS), 2011. [PDF]
Research
Projects
- Penalized Nonnegative Matrix Factorization
December 2011 -- January 2012
- Metric Learning for Image Analysis
October 2011 -- November 2011
- Q & A with Topic Models
July 2011 -- November 2011
- Sparse Structured Probabilistic Projections
February 2011--May 2011
- Graph Optimized Spectral Clustering
February 2011--May 2011
- Scalable Spectral Clustering
August 2010--January 2011
Worked on approximating a matrix with low rank, especially the nonnegative matrix factorization technique. Inspired in part by the Fused Lasso, I proposed a new regularization framework to penalize the basis vectors, and developed an algorithm to solve the general optimization problem.
Worked on learning a distance metric with side information. As the project leader, I quickly learned most state-of-the-art metric learning algorithms and proposed a new algorithm specially designed for images, which achieves a significant improvement.
Working on a joint program of UCLA and UMass, which seeks to automatically forward a question to some appropriate scientists, and detect time variations of user interests. I take part in both the design and implementation of the proposed model. The project is on-going.
Worked on extending the probabilistic canonical correlation analysis (PCCA) to a multi-view learning framework. The generated representation has a property of structured sparsity which enables it to extract both shared and view-dependent information. I participated in both the modeling and the experimentation part.
Worked on incorporating dimension reduction and clustering into a coherent procedure based on graph Laplacian. I proposed an iterative algorithm which seeks an intrinsically similar, but more robust, low-dimensional representation than its counterpart in spectral clustering.
Worked on accelerating the spectral clustering algorithm while preserving its high accuracy. Adopting the idea of landmark representation and singular value decomposition (SVD), I proposed a scalable method which has linear time complexity while still preserves good performance.
Honors & Awards
Miscellaneous
-
Interested in playing words in a sentence, in both Chinese and English.
-
Color-blind, with no exsisting algorithm to enhance my ability to distinguish colors.