DocumentCode
2017661
Title
On the overfitting of the five-layered bottleneck network
Author
Hiraoka, K. ; Shigehara, T. ; Mizoguchi, Hiroshi ; Mishima, T. ; Yoshizawa, S.
Author_Institution
Dept. of Inf. & Comput. Sci., Saitama Univ., Urawa, Japan
Volume
1
fYear
1999
fDate
1999
Firstpage
234
Abstract
In autoassociative learning for the bottleneck neural network, the problem of overfitting is pointed out. This overfitting is pathological in the sense that it does not disappear even if the sample size goes to infinity. However, it is not observed in the real learning process. Thus we study the basin of the overfitting solution. First, the existence of overfitting is confirmed. Then it is shown that the basin of the overfitting solution is small compared with the normal solution
Keywords
associative processing; learning (artificial intelligence); multilayer perceptrons; autoassociative learning; autoassociative neural network; five-layered bottleneck network overfitting; multilayered neural network; Computer science; Data engineering; Fault detection; H infinity control; Neural networks; Noise reduction; Pathology; Principal component analysis; Surface fitting; Unsupervised learning;
fLanguage
English
Publisher
ieee
Conference_Titel
Neural Information Processing, 1999. Proceedings. ICONIP '99. 6th International Conference on
Conference_Location
Perth, WA
Print_ISBN
0-7803-5871-6
Type
conf
DOI
10.1109/ICONIP.1999.843992
Filename
843992
Link To Document