发帖
雷达卡
 | MATH 00112230 高等统计选讲 (Selected Topics in Advanced Statistics) Fall 2009 |
| Course Information |
答疑 (office hours):Wed 2:00PM-4:00PM. 理科一号楼 1283E.
Email address:Personal: yuany@math (DOT pku DOT edu DOT cn).
Group: pkumath112230 AT googlegroups DOT com.
课程内容:This course will cover some of my personal research fields as well as selected topics in statistical machine learning.
Lecture 1. 9/16. A Personal Overview on Statistical Machine Learning.
Lecture 2. 9/18. Clustering and Biomolecular Dynamics I. [slides]
Lecture 3. 9/23. Clustering and Biomolecular Dynamics II. [slides]
Lecture 4. 9/30. From PageRank to HodgeRank: A Differential Geometric Perspective on Statistical Ranking. [slides]
Lecture 5. 10/14. Unfolding Manfolds: ISOMAP vs. LLE, An Introduction to Geometric Data Analysis. [slides]
Lecture 6. 10/16. A Challenge Problem Set in Causal Network Learning. Presenter: WANG, Changzhang. [note][reference slides]
Lecture 7. 10/21. Yeast Cell-Cycle Regulation Network Inference. Presenter: WANG, Lin. [slides]
Lecture 8. 10/28. Invited Speaker: Dr. WANG Li, Stem Cell Research Center, Peking Univeristy.
Title: Gene Expression Studies on the effect of superovulation in Embryo Development. [slides]
Abstract. The effect of superovulation on preimplantation embryo quality and subsequent post-implantation embryo development remains controversial. In the present study we analyzed global gene expression profile of embryos generated from superovulated and natural cycles and compared their developmental capacity.
Lecture 9. 10/30. Theory and Application of Copula. Presenter: MAO, Guangyu. China Center for Economic Research, Peking University. [slides]
Lecture 10. 11/4. Causal Inference, I. Presenter: DING, Peng. Department of Probability and Statistics, Peking University. [slides] [reference]
Lecture 11. 11/11. Causal Inference and Causal Network Learning, II. Presenter: DING, Peng and WANG, Changzhang. Department of Probability and Statistics, Peking University. [Changzhang's matlab codes]
Lecture 12. 11/13. Compressive Sensing and Clique Identification in Social Networks. [slides]
Lecture 13. 11/18. Confidence Intervals without Pivots in Structural Equation Modeling (SEM) Reconstruction. [notes][R-package SEM Manual][R-package SEM 0.9] which requires [R-package NLME 3.1]
Lecture 14. 11/25. Invited Speaker: ZHAO Xin, Department of Computer Science, Peking University.
Title: An Introduction to Text Mining and Related Research Topics: Statistical Language Modeling Approach. [slides]
Abstract. The talk mainly introduced topic models in statistical language modeling, such as pLSA (Probabilistic Latent Semantic Analysis) and LDA (Latent Dirichlet Allocation), as well as the applications in web page retrieval, text with networks, and opinion mining, etc. A survey is also given on the main projects carried in Li Xiaoming's Lab in Peking University.
Lecture 15. 11/27. The combinatorial regulation of motif modules and microRNAs, I. Presenter: WANG, Yufu. Department of Probability and Statistics, Peking University.[reference]
Lecture 16. 12/2.
Presenter I: WANG, Yufu. The combinatorial regulation of motif modules and microRNAs, II.
Presenter I: ZOU, Chenchen. Graph Embedding and Extensions: A General Framework for Dimensionality Reduction. [slides][reference]
Lecture 17. 12/9. Experimental Design. Presenter: LIU, Senmao. Department of Probability and Statistics, Peking University. [slides][reference 1 (Chen-Sun-Wu'1993)][reference 2 (Box'1961)] [reference 3 (Li-Zhao-Zhang'08)][reference 4 (Zhang-Mukerjee'07)]
Lecture 18. 12/16. Invited Speaker: JIANG Xiaoye, ICME, Stanford University.
Talk I: Identity Management Problem: Reasoning and Inference over Permutations [slides]
Talk II: Identity Management on Homogeneous Spaces [slides]
Lecture 19. 12/23. Invited Speaker: CHAI Anwei, ICME, Stanford University.
Title: Imaging Localized Scatterers Using the Singular Value Decomposition and l_1 Optimization. [slides]
Abstract. Motivated by theory in Compressed Sensing, we consider narrow band array imaging of localized scatterers using the singular value decomposition (SVD) and $l_1$ minimization and compare the results with other imaging methods such as multiple signal classification (MUSIC). We show that well-separated point scatterers can be recovered exactly with $\ell_1$ optimization. Moreover, using the SVD we determine optimally subsampled array data for the $\ell_1$ optimization. Numerical simulations indicate that this imaging approach is robust with respect to additive noise. A comparative study of different imaging methods in random medium will also be shown. This is a joint work with Laurent Demanet and George Papanicolau.
Lecture 20. 12/25. Invited Speaker: Professor YUAN Xiaoru, School of EECS, Peking University.
Title: Multi-Dimensional Data Visualization: Parallel Coordinates. [slides]
Abstract. This talk will introduce multi-dimensional data visualization conducted in Yuan's Visualization Lab at Peking University, in particular the new development on Parallel Coordinates approach together with softwares.
Lecture 21. 12/25. Invited Speaker: LI Youquan, Peking University Health Science Center.
Title: New Ideas in Radiology for Lung Cancer. [slides]
Abstract. This talk will introduce a new methodology in radiology for lung cancer treatment and a derived statistical prediction problem in such a scheme. The prediction problem is to predict the residual volume and position of mass center of the target after the treatment. Previous work are presented [Zheng's Master Thesis] and new challenge still remains (data is available).
Lecture 22. 12/30. Invited Speaker: WANG Wei, School of EECS, Peking University.
Title: Visual Attention: What Attract Us? [slides]
Abstract. This talk will give a survey on visual attention and several computational models, which in particular includes recent research in WANG Yizhou's Lab.
by YAO, Yuan. http://www.math.pku.edu.cn/teachers/yaoy/math112230/
京公网安备 11010802022788号
