Abstract—To improve the performance of neural network (NN), a new approach based on input space partitioning is introduced, i.e. partitioning according to the correlation between input attributes. As a result, the effect of weak correlation and non-correlation is excluded from the crucial stage of training. After partitioning, CBP network is introduced to train different sub-groups. The results from different networks are then integrated. According to the experimental results, improved performance is attained.
Index Terms—Correlation, input attributes, neural network, partitioning.
Shujuan Guo, Sheng-Uei Guan and Shang Yang are with the School of Electronic & Information Engineering, Xi’an Jiaotong University, Xi’an, China (e-mail: Steven.Guan@xjtlu.edu.cn)
Weifan Li, Linfan Zhao and Jinghao Song are with the Dept. of Computer Science and Software Engineering, Xi’an Jiaotong-Liverpool University, Suzhou, China.
[PDF]
Cite:Shu Juan Guo, Sheng-Uei Guan, Shang Yang, Wei Fan Li, Lin Fan Zhao, and Jing Hao Song"Input Partitioning Based on Correlation for Neural Network Learning," Journal of Clean Energy Technologies vol. 1, no. 4, pp. 335-338, 2013.