STORAGE CAPACITY FOR HIERARCHICALLY CORRELATED PATTERNS

被引:5
作者
ENGEL, A
机构
[1] Sektion Phys., Humboldt-Univ., Berlin
来源
JOURNAL OF PHYSICS A-MATHEMATICAL AND GENERAL | 1990年 / 23卷 / 06期
关键词
D O I
10.1088/0305-4470/23/6/009
中图分类号
O4 [物理学];
学科分类号
0702 ;
摘要
The storage capacity of a Hopfield model with patterns forming a two-level hierarchy is calculated without reference to a special learning rule. If the number of different classes becomes very large strong correlations within the classes decrease the storage capacity in striking contrast to what is known about one-level hierarchies, i.e. patterns with magnetisation.
引用
收藏
页码:L285 / L289
页数:5
相关论文
共 23 条
[21]  
VANHEMMEN JL, 1987, LECTURE NOTES PHYSIC, V275
[22]  
VANHEMMEN JL, 1987, PHYS REV A, V36, P1953
[23]   THE EFFECT OF SYNAPSES DESTRUCTION ON CATEGORIZATION BY NEURAL NETWORKS [J].
VIRASORO, MA .
EUROPHYSICS LETTERS, 1988, 7 (04) :293-298