2008 China-US-France Summer Courses on

Machine Learning, Statistics and Computer Vision

Jun.30 2008--Jul.11 2008

Lotus Hill Research Institute for Computer vision and Information Science

The lectures is open to all graduates, postgraduates, post doctors and youth professors from home and abroad who engage in the field of computer vision, pattern recognition, machine learning and other related subjects (applied mathematics, statistics and computer ). In this summer school, the lectures involving several themes will last about two weeks, which will systematically introduce the foundation, advanced theories, principles, methods, and the latest progress of computer vision, pattern recognition and machine learning.

Organizoring Committee:

   Prof. Song-Chun zhu, University of California, Los Angeles.
   Prof. Veronique Prinet, Institute of Automation, CAS.
   Prof. Xuming He, University of Illinois, Urbana Champion.
   Prof. Jian-Feng Yao, IRMAR, Université de Rennes 1.

Lecturers from the US:

   Prof. Bin Yu,Department of Statistics, University of California, Berkeley,
   Prof. Ji Zhu, Department of Statistics, University of Michigan,
   Prof. Chengxiang Zhai, Department of CS, University of Illinois, Urbana Champaign,
   Prof. Yuguo Chen, Department of Statistics, University of Illinois, Urbana Champaign,
   Prof. Song-Chun Zhu, Departments of Statistics and CS, University of California, Los Angeles.

Lecturers from France:

   Prof. Bill Triggs, Laboratoire Jean Kuntzmann, Grenoble,
   Prof. Ivan Laptev, Inria, Rennes,
   Prof. Cordelia Schmid, Inria, LEAR Project, Alpes.

Topics for short courses: ( new)

   (Week 1: June 30-July 4)

   Short course 1:
   Machine Learning: Kernels, Boosting, Random Forest, Variable Selection
   Topics: 1.Kernel Methods I
                 2. Kernel Methods II
                 3. Random Forest and Boosting
                 4. Clustering
                 5. Variable Selection
   Professor Ji Zhu, University of Michigan

   Short course 2:
   Machine Learning: Statistical Models for Information Retrieval and Text Mining
   Topics: 1. Overview of Information Retrieval Models
                 2. Statistical Language Models for Information Retrieval
                 3. Feedback in Information Retrieval
                 4. Contextual Probabilistic Latent Semantic Analysis
                 5. Applications of Topics Models in Text Mining
                 Professor Chengxiang Zhai, University of Illinois, Urbana Champaign

   Short course 3:
   Statistics: Topics on sequential Monte Carlo and Markov Chain Monte Carlo
   Topics: 1. Introduction to Monte Carlo Methods
                 2. Sequential Importance Sampling
                 3. Sequential Monte Carlo for Hidden Markov Models
                 4. Markov Chain Monte Carlo - I
                 5. Markov Chain Monte Carlo - II
   Professor Yuguo Chen, University of Illinois, Urbana Champaign

   Short course 4:
   Vision: visual recognition and human motion reconstruction
   Topics: to be advised
   Professor Bill Triggs, Laboratoire Jean Kuntzmann (LJK) in Grenoble

Weekend Vacation plan:

   We are planning a weekend July 5-6 at Lu Shan (庐山) for a vacation which is 3.5 hours away from Lotus Hill. Lu Shan is known for its cool weather in the summer.

   (Week 2: July 7-July 11)

   Short course 5:
   Machine Learning: Topics on Sparsity
   Topics: 1. Sparsity through model selection
                 2. Sparsity through L1 constrained L2 penalization (Lasso)
                 3. Theoretical results on Lasso
                 4. Group and hierarchical sparsity
                 5. Understanding the visual pathway through sparsity
   Professor Bin Yu, University of California, Berkeley

   Short course 6:
   Vision: Image features and Object recognition (Tentative)
   Topics: 1. Local features I
                 2. Local feautres II
                 3. Efficient image search
                 4. Category recognition I
                 5. Category recognition II
   Professor Cordelia Schmid, Inria, LEAR Project, Alpes.

   Short course 7:
   Vision: human action recognition
   Topics: 1. Human actions in computer vision
                      Why is it important, applications, ...
                      Why is it hard
                      Overview of existing methods and their focus
                 2. 2D, 3D structure models, motion priors,
                      tracking, applications in Graphics
                 3."Silhouette world", Active Shape Models
                      Motion History
                 4. Space-time volume representations
                      Action similarity, matching, alignment
                 5. Local methods, space-time interest points
                     action recognition "in the wild"
   Professor Ivan Laptev, Inria, Alpes.

   Short course 8:
   Vision and Learning: Manifold Learning and Image Parsing
   Topics: 1. Pursuing Manifolds in the Universe of Image Patches by Information Projection
                 2. Information Scaling and Regimes of Statistical Models of Image
                 3. Primal sketch: Integrating Textures and Structures
                 4. Active Basis Model for Object Recognition and Learning
                 5. Image Parsing with Data Driven Markov Chain Monte Carlo
                 6. Swendsen-Wang Cuts for fast MCMC computing
                 7. Stochastic Image Grammar and top-down/bottom-up inference
   Professor Song-Chun Zhu, UCLA

Co-sponsors:

   National Science Foundation, USA,
   PRA(Sino-French Advanced Research Project),
   LIAMA (Sino-french lab, Inst. Of Automation), CAS,
   Lotus Hill Institute,
   Center for Statistics Science, AMSS, CAS of China,
   The Virtual Center for U-China Collaboration among Statisticians.

Pictures of the Lotus Hill Resorts:

How to go to EZhou:

1. By train

   According to your starting station, you can choose the destination （Wuchang or Hankou railway station）.The former is preferable. Then you can go to FuJiaPo bus station and take a bus to Ezhou from there.

2. By bus

    According to your starting station, you can choose to go to Ezhou directly. If a nonstop is not available, you can first go to FuJiaPo bus station, then take a bus to Ezhou from there.

3. By airplane

   After arriving at Tianhe airport , you can go to FuJiaPo bus station in WuHan city and take a bus there to Ezhou . Also, you can take a taxi to Ezhou directly from the airport. It costs about 300RMB.