多层感知器的局部化设计
A Local Design for Multi-Layer Perceptron
-
摘要: 结合前馈网络的全局化设计和局部化设计的各自特点,提出一种由多个既独立又关 联的同构子网络构成的局部化多层感知器.仿真结果表明,该网络对复杂非线性系统具有良 好的学习性能,并适用于学习控制的直接逆模型法和远程学习法.Abstract: By syncretizing the global design and local design of feedforward neural networks, this paper proposes a localized multi-layer Perceptron (LMLP), which is constructed by a set of distributively associated subnets with the same structure. Simulation results show that this network has strong learning capacity for complex nonlinear systems, and is applicable for real-time learning control, by using either direct inverse modeling or distal learning strategy.
-
Key words:
- Multilayer perceptron /
- local design /
- CMAC network
计量
- 文章访问数: 2733
- HTML全文浏览量: 95
- PDF下载量: 924
- 被引次数: 0