International Conference on Mechanical Engineering and Technology (ICMET-London 2011)
168 Sample Selection Method of Neural Network Based on FCM and NN Method
Download citation file:
- Ris (Zotero)
- Reference Manager
In the application of neural network, we often encounter in such problems as the sample set contains too many similar samples or sample features are not representative enough. All these problems may lead to 1Âover learning1Â phenomenon or the case that the results predicted by neural network model deviate largely from the actual results. The paper proposes a method of using Fuzzy C-means (FCM) clustering algorithm and the nearest neighbor (NN) method to establish a neural network sample set. This method can achieve the purpose of establishing the best neural network model with a small and representative sample set, which is a good guide for the application of neural network.