Abstract:The paper is on the basis of the principle of structure-risk-minimum in statistical learning theory,theoretically gives structural design method for neural networks and practice.The algorthm can expand self-adaptively the volume of necral networks,so can finish structural design of network and improve training precision and general ability in a condition of limited samples,so thatimproves reliability of prediction results of networks.The application cases that use the method to predict oil and gas reservoir in a development block.Jilin Oilfield were given in the paper and showed a good results.
张向君, 李幼铭, 刘洪. 神经网络结构风险最小油气预测[J]. 石油地球物理勘探, 2002, 37(1): 73-76.
Zhang Xiangjun, Li Youming, Liu Hong. Oil and gas prediction by neural networks with stricture-risk-minimum. OGP, 2002, 37(1): 73-76.