Approximation by Feed-forward Neural Networks
Speaker:Prof. Feilong Cao
College of Science
China Jiliang University
Date & Time:4 May 2012 (Friday) 11:00
Organized by:Department of Mathematics


As we all know, many problems in applications can be converted into problems of approximating unknown functions with some sample points. Since feed-forward neural network (FNN) has a simple structure and a universal approximation property, it is very popular in applications. There are two basic problems in the researches of approximation by FNNs, including density and complexity. This lecture will introduce the two problems.


Prof. Cao is currently Dean and Professor of College of Science of China Jiliang University. He obtained his Ph.D. in Applied Mathematics in Xi’an Jiaotong University. Prof. Cao is studying on Error Controlled Computation and reliability algorithm, Learning Theory on Sphere, Complexity of Structure and Essential Approximation Order of Neural Networks etc. His publications include Multivariate weighted Bernstein-type inequality and its applications, Learning rates of support vector machine, Estimates of learning rates of regularized regression via polyline functions etc.