- Issue:
- 2001年01期

- Page:
- 108-112

- Research Field:

- Publishing date:

- Title:
- Small-sample Machine Learning Theory: Statistical Learning Theory

- Author(s):
- TanDongnin TanDonghan①
- Electronic 55 th Institude,Ministry of Information Industry,Nanjing 210016)

- Keywords:
- sample trees; stat ist ical est imat ion; pattern recognit ion; statist ical learning theory; machine learning

- PACS:
- TP18

- DOI:
- -

- Abstract:
- Stat ist ical Learning T heory, a recent ly developed new theory for pattern recogn-i t ion, is a smal-l sample stat ist ics proposed by Vapnik et al, w hich deals mainly w ith the statist ic principles w hen samples are limited, especially to describe the propert ies of learning procedure in such cases. It provides us a new framework for the smal-l sample learning problem, and also a novel powerful learning method called Support Vector Machine, w hich can solve smal-l sample learning problem better. T his paper w ill introduce the basic ideas of the theory, its major characterist ics, some current research t rends of it and some thinking from us about it .

1 Vapnik V N. Estimation of dependencies based o n empir ical data. Berlin: Spr ing er-Verlag , 1982

2 Vapnik V N. The nature of statistical learning theory. New York: Spr inger-Verlag, 1995

3 Cherkassky V, Mulier F. Learning from data: concepts, theory and met hods. New York: Jo hn Viley& Sons, 1997

4 加肇祺1 模式识别1 北京: 清华大学出版社, 1988

5 Vapnik V N, Lev in E, Le Cun Y. Measuring the VC-dimensio n of a learning machine. Neural Computation, M IT Press, 1994( 6) : 851~ 876

- Memo:
- -

Last Update: 2013-03-25