江西科学
江西科學
강서과학
JIANGXI SCIENCE
2015年
4期
449-452,457
,共5页
詹棠森%付长春%卢金珠%丁巍
詹棠森%付長春%盧金珠%丁巍
첨당삼%부장춘%로금주%정외
关联变权%小波函数%艺术陶瓷定价
關聯變權%小波函數%藝術陶瓷定價
관련변권%소파함수%예술도자정개
correlation transfered weight%the wavelet function%the art ceramic price
通过对小波函数的紧支撑性及光滑性的分析,结合神经网络算法,提出一种基于关联变权的小波神经网络算法。这种算法对于权值变小情况下比较适用,预测的误差几乎处处为0,并具有较好泛化能力。
通過對小波函數的緊支撐性及光滑性的分析,結閤神經網絡算法,提齣一種基于關聯變權的小波神經網絡算法。這種算法對于權值變小情況下比較適用,預測的誤差幾乎處處為0,併具有較好汎化能力。
통과대소파함수적긴지탱성급광활성적분석,결합신경망락산법,제출일충기우관련변권적소파신경망락산법。저충산법대우권치변소정황하비교괄용,예측적오차궤호처처위0,병구유교호범화능력。
By means of the analysis of compactly supported properties and the smoothness to wavelet function,combined with neural network algorithm, a algorithm of wavelet neural network based on correlation transferedweight is put forward. The algorithm is applicable for cases of smaller weights, its prediction errors are almost-everywhere zeros,and the algorithm has better generalization capabili-ty.
