Informal Miniseminar Talk: all invited. 
Thursday, Nov 29, 2001, 1:00pm room 4310

                    Generalization Machine
                        Xiaotong Shen 
                   The Ohio State University 

  The concept of large margins have been recognized as an 
important principle in analyzing learning methodologies, 
including Boosting, Neural Networks, and Support Vector 
Machine (SVM). However, this concept alone
is not adequate for learning in nonseparable cases. In this talk,
I will present a new learning methodology, called Generalization
Machines (GM), that is derived from a direct consideration of
generalization errors.  In particular, I will discuss three
aspects: (1) the methodology development, (2) computational tools,
and (3) the generalization ability.  Numerical examples will be 
provided to examine GM's generalization ability. 
An application of GM to breast cancer classification 
will also be discussed. 

Based on the joint work with Xuegong Zhang, George Tseng and 
Wing Hung Wong.