Event(s) on February 2005

• 1/2/2005

  題目:	Lagrangian-type functions with applications
  講員:	Prof. Yang Xiaoqi, Department of Applied Mathematics, Hong Kong Polytechnic University, Hong Kong
  時間/地點:	11:30  -  12:30 FSC1217
  摘要:	Lagrangian-type functions play an important role in unconstrained optimization methods for solving constrained optimization problems. Augmented Lagrangian, nonlinear Lagrangian and lower order penalty functions are xamples of Lagrangian-type functions. In this talk, we will in particular discuss lower order penalty functions from the following aspects: exact penalization representation, formula of computing the least exact penalty arameter, convergence analysis of optimality conditions. Finally we will present an application of lower order penalty functions to the evaluation of American option pricing.

  
  

• 2/2/2005

  題目:	Dimension-reducing Methods in Regressions: In Action
  講員:	Prof. Lixing Zhu, Department of Statistics and Actuarial Science, The University of Hong Kong, Hong Kong
  時間/地點:	14:30  -  15:30 FSC1217
  摘要:	Regression analysis applied to many practical fields is an �oldest�, as well as a �youngest� member in the statistics family. It is never less active. Much special attention on it has been received in recent years due to the process of attacking �curse of dimensionality�. Often, every observation consists of response and many covariables. When the number of covariables is too large, how to efficiently estimate regression function through the collected observations is very challenging. This is because of the data sparseness in high-dimensional space. This challenge excites the enthusiasm from statisticians to attack this problem. For this, reducing dimension is a natural idea and plays a crucial role. In this seminar, some conventional methodologies will be reviewed and some newly developed methodologies will be discussed. Some interesting problems will be addressed. This seminar is going to draw an outline of this research area, and to show how long the way has to go to reach a destination that regression analysis works well in these difficult settings.

  
  

• 15/2/2005

  題目:	Categorical Data, Markov Models and Applications
  講員:	Prof. Michael K. Ng, Department of Mathematics, The University of Hong Kong, Hong Kong
  時間/地點:	11:30  -  12:30 FSC1217
  摘要:	Categorical data sequences occur frequently in many real world applications. The most important step in analyzing a data sequence is the selection of an appropriate mathematical model for the data. Because it helps in predictions, hypothesis testing and rule discovery. In this talk, we survey some of the latest developments on using Markov models for categorical data analysis. Applications to inventory control, web mining,credit risk, genetic network are given.