摘要
针对框架小波网络存在的网络结构不易确定的问题．提出一种基于TBIC(Testing Bayesian information criterion)信息准则的网络结构设计方法。该方法权衡网络规模和测试精度两方面因素,得到了在TBIC信息准则下的最优小波网络,达到了优化网络结构的目的。仿真结果表明,用该方法设计得到的小波网络具有较小的网络结构和良好的测试精度。相比于BIC(Bayesian information criterion)准则函数,该准则能更准确地反映出测试精度随网络结构变化的趋势。
Based on the testing Bayesian information criterion （TBIC）, a method is presented to solve the indeterminacy of the network structure existed in frame wavelet networks. Both the scale and the precision of the network are considered, thus obtaining an optimal network structure under the TBIC rule. Simulation results show that the wavelet network has a reasonable scale and good testing performance by using the method. Compared with the Bayesian information criterion （BIC）, TBIC can accurately reflect the precision along with the changing trend of the network structure.
出处
《南京航空航天大学学报》
EI
CAS
CSCD
北大核心
2006年第B07期124-127,共4页
Journal of Nanjing University of Aeronautics & Astronautics
