Casasent D, Botha E
Appl Opt. 1992 Mar 10;31(8):1030-40. doi: 10.1364/AO.31.001030.
A new neural net is described that can easily and cost-effectively accommodate multiple objects in the field of view in parallel. The use of a correlator achieves shift invariance and accommodates multiple objects in parallel. Distortion-invariant filters provide aspect-invariant distortion. Symbolic encoding, the use of generic object parts, and a production system neural net allow large class problems to be addressed. Optical laboratory data on the production system inputs are provided and emphasized. Test data assume binary inputs, although analog (probability) input neurons are possible.
本文描述了一种新型神经网络，它能够轻松且经济高效地并行处理视野中的多个物体。相关器的使用实现了平移不变性，并能并行处理多个物体。失真不变滤波器提供了长宽比不变的失真处理。符号编码、通用物体部件的使用以及产生式系统神经网络使得大规模分类问题得以解决。文中提供并强调了关于产生式系统输入的光学实验室数据。测试数据假定为二进制输入，不过模拟（概率）输入神经元也是可行的。
