论文部分内容阅读
提出一种面向对象的神经网络类库的构造方法。该类库采用面向对象的方法 ,集成了当前比较成熟的人工神经网络模型 ,充分利用了OOP方法的继承性、封装性和多态性[1] ,便于在现有基础上派生出新的神经网络模型。该类库将被用于实现一种最新的同步振荡神经网络———LEGION网络。由于系统的结构性和代码的可重用性 ,既有利于科研工作者进行神经网络的研究 ,也便于工程人员的神经网络应用
An object-oriented neural network library construction method is proposed. The library uses an object-oriented approach, integrates the current relatively mature artificial neural network model, take full advantage of OOP inheritance, encapsulation and polymorphism [1], to facilitate the derivation of a new nerve on the existing basis Network model. This library will be used to implement a new type of synchronous oscillating neural network --- the LEGION network. Due to the structural and code reusability of the system, it is not only beneficial for researchers to carry out the research of neural network, but also to the neural network application of engineering staff