降维映射分析法及其应用

Analytical Method of Reducing Dimensional and Mapping and Its Applications

用人工神经网络将多维空间的样本数据降维映射到二维平面上 ,并生成目标函数的等值线 ,可全景式地展现出样本数据集操作空间的面貌和特征 ,由此可直接看出最优操作点或最优操作区域。映射平面上的任意点可通过逆映射算法将其还原到多维空间。本文对这个方法的基本原理进行了简要的描述 ,并用于操作优化的实例计算。演示的实例说明 ,这个方法比模式识别方法优越。

The sample datum in multidimensinal space are mapped to the two dimensional plane by a neural network. The contours of objective fuction are produced automatically on the plane, which completely reveals the characteristic and the shape of the operational space which sample datum cover. The optimal operational point or region on this mapping plane can be determined intuitively. An arbitrary point on the mapping plane may be reverted to original multi dimensional space by the algorthm of reversal mapping. This paper described briefly the basic principle of the method. The availability of the method was illustrated with the examples of the operational optimization. The results show that this novel method is superior to pattern recognition.