论文部分内容阅读
目前人工神经网络模型学习与训练时间长，收敛性难以保证，网络鲁棒性差，存在局部极小。尤其当网络规模增大时，上述缺点变得尤为严重。为克服上述缺陷，提出了一组新的方法，即利用训练样本集提供的全局知识，通过不同数学工具，设计网络的结构与参数。结果表明，这套方法可以达到很好的效果，是改进人工神经网络性能的一种有效途径。
At present, artificial neural network model has a long learning and training process, and its convergence is difficult to guarantee. Its robustness is poor and there is a local minimum. In particular, the above shortcomings have become even more serious as the size of the network increases. In order to overcome these shortcomings, a new set of methods is proposed, which uses the global knowledge provided by the training sample set to design the structure and parameters of the network through different mathematical tools. The results show that this method can achieve good results and is an effective way to improve the performance of artificial neural networks.