深度信念网络的等效模型及权值扩展算法研究

Research on equivalent model and weight extension algorithm of deep belief network

针对深度信念网络(DBN)中小样本情况下,训练模型泛化性较差,分类识别率不够理想,系统性能有待提高等问题,研究了DBN的等效模型,分析了小样本情况下识别率差的问题;并提出一种区间化权值扩展方法,扩大了样本和权值的匹配空间,使判决更有利于正确分类,提高了小样本情况下的图像分类准确性;用检测与估值理论给出了算法能提高系统检测性能的依据,并在不同的数据库上进行了实验测试,进一步证明了小样本情况下图像分类准确率的提高。最后,将该方法应用到了小样本绝缘子故障识别中。

Aiming at the problem of low recognition accuracy of the training model in the case of less data samples in deep belief network(DBN),which led to the classification recognition rate is not ideal,so the system performance needs to be improved.This paper researches the equivalent model of DBN,analyzes the problem of poor recognition rate in the case of small samples;Then,a weight expansion algorithm is proposed to enlarge the matching space between the sample and weight,so that the decision is more conducive to correct classification,which improves the accuracy of image classification under the condition of small sample size;The algorithm is proved that can promote the performance of the system by using the detection and estimation theory,the test of different sample banks is further proof of this completion.Finally,the method is applied to small sample faulted insulator recognition.