INVARIANCE AND NEURAL NETS

被引:40
|
作者
BARNARD, E [1 ]
CASASENT, D [1 ]
机构
[1] CARNEGIE MELLON UNIV,DEPT QUIM ANALIT,PITTSBURGH,PA 15213
来源
关键词
D O I
10.1109/72.134287
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Invariance with respect to certain transformations is one of the main tasks of pattern-recognition systems. We study various techniques for obtaining this invariance with neural net classifiers and identify the invariant-feature technique as the most suitable for current neural classifiers. A new formulation of invariance in terms of constraints on the feature values leads to a general method for transforming any given feature space so that it becomes invariant to specified transformations. A case study using range imagery is used to exemplify these ideas, and good performance is obtained.
引用
收藏
页码:498 / 508
页数:11
相关论文
共 50 条