用阶梯样本提取受训MLP或SVM的输入输出关系  

Extracting correlations between inputs and outputs of trained MLP or SVM using ladder-samples

作　　者：刘挺[1] 丁香乾[2] 冯天瑾[2] 杨宁[2] 

机构地区：[1]中国海洋大学电子工程系,山东青岛266071 [2]中国海洋大学信息工程中心,山东青岛266071

出　　处：《哈尔滨工程大学学报》2006年第B07期356-359,共4页Journal of Harbin Engineering University

摘　　要：神经网络具有强大的函数拟合能力，但是，训练的结果是‘黑盒子’．采用构造阶梯样本、计算陡峭指数的方法，将受训多层感知器或支撑向量机的输入输出变量关系以直观的形式（阶梯图）反映出来．与传统方法比较，这种新的相关性分析方法能更充分地揭示样本中输入与输出变量间的非线性关系．在烟草行业单料烟模拟感觉质量评估的应用中，仿真试验验证了该方法的有效性与实用性．Neural networks have powerful ability of approximating functions, but the training results are black boxes. The correlations between inputs and outputs of trained MLPs or SVMs might be extracted with the method of constructing ladder-samples, visually reflected in the form of ladder-image, and calculating cragged-index. In contrast with traditional method, this new method unfolds the linear and nonlinear correlation between inputs and outputs more adequately. While the method is applied in forecasting the sensory-quality evaluation for tobaccos, the validity and practicality of the method has been testified by the simulative experiments.

关 键 词：相关性 BP网络 样本阶梯法 非线性相关 陡峭指数 

分 类 号：P18[天文地球—天文学]