Author:
Chen Haibo,Zhang Chenyu,Hu Gao
Abstract
Abstract
We presents an adaptive learning algorithm of RBF neural network. This algorithm uses an adaptive splitting operation based on network sensitivity and sample density to dynamically change the number of nodes in the hidden layer of RBF network. At the same time, a refactoring operation based on energy consumption is proposed, and the connection weights of the hidden layer and the output layer are obtained by using the least square method. In the experiment of iris sample classification, the recognition rate of the model is 95%.
Reference6 articles.
1. Universal Approximation Using Radial-Basis-Function Networks;Park;Neural Computation
2. Radial Basis Function Neural Classifier using a Novel Kernel Density Algorithm for Automobile Sales Data Classification;Mary;International Journal of Computer Applications,2011
3. Automatic construction of the complete architecture of a radial basis function network using differential evolution;Tovias-Alanis,2017
4. A combined SMOTE and PSO based RBF classifier for two- class imbalanced problems;Gao;Neurocomputing,2011
5. A new optimized GA-RBF neural network algorithm.[J];Jia;Computational Intelligence & Neuroscience,2014