自然界生物系统中的神经网络不是设计出来的,而是进化产生的,这种神经网络具有模块结构且协同工作。利用进化手段模拟生物系统中神经系统的产生和工作过程对神经网络研究意义重大,但目前这种研究处于相当初级的阶段。香港大学的学者提出了使用遗传算法决定优化的神经网络拓扑结构和参数,并用自适应BP算法加速训练过程,用于咖啡分类问题。遗传算法用于决定最有效的隐层数、每层隐 The neural network in the natural biological system is not designed but evolved. This neural network has the module structure and works together. The use of evolutionary means to simulate the production and working process of the nervous system in biological systems is of great importance to the research of neural networks. However, at present, this kind of research is in a quite primitive stage. University of Hong Kong scholars proposed the use of genetic algorithms to optimize neural network topology and parameters, and use adaptive BP algorithm to speed up the training process for coffee classification. Genetic algorithms are used to determine the most effective hidden layers, hidden at each layer