FINDING STRUCTURE IN SIGNALS, IMAGES, AND DATA
The talk will be a tutorial survey, concentrating on the main principles and categories of unsupervised neural learning in the problem of data mining for signals, images, and data. In neural computation, there are two classical categories for unsupervised learning methods and models: first, extensions of Principal Component Analysis and Factor Analysis, and second, learning vector coding or clustering methods that are based on competitive learning. The talk concentrates on two of these extensions: for the first category, the novel technique of Independent Component Analysis, and for the second category, the Kohonen Self-Organizing Map. The more recent trend in unsupervised learning is to consider this problem in the framework of probabilistic generative models. If it is possible to build and estimate a model that explains the data in terms of some latent variables, key insights may be obtained into the true nature and structure of the data. This approach is also briefly reviewed. After a brief introduction to the underlying theoretical foundations of these ideas, unsupervised neural learning will be illustrated by several applications in data mining ranging from document and pictorial databases to blind signal separation.
unsupervised learning data mining independent component analysis self-organizing map
Erkki Oja
Helsinki University of Technology Neural Networks Research Centre P.O. Box 5400, FIN-02015 HUT, Finland
国际会议
8th International Conference on Neural Information Processing(ICONIP 2001)(第八届国际神经信息处理大会)
上海
英文
9-13
2001-11-14（万方平台首次上网日期，不代表论文的发表时间）