Application of Alternative Covering Neural Networks in Data Classification Based on Rough Set

摘要

Based on discussing in the alternative covering neural networks (ACNN), the integrated algorithm are proposed based on rough set (RS) theory and ACNN. RS is applied to reduce and process the original data. While ensuring the integrity of information, the data dimension is reduced. ACNN is used to design multi-layer forward network. Through using RS to reduce data dimension, the calculation of ACNN is decreased to lower the complexity of network computing. The experimental results prove that the integrated approach is effective. Comparing with the results by K-W method, it is concluded that the importance of the data classification with RS is analyzed and the results are in keeping with the practical data operation, which directly approves better validity of RS in data classification.