论文部分内容阅读
本文证明了示例学习中的最大复合问题（ＭＧＣ）是ＮＰ难题，给出了求解最大复合问题的近似算法，并将此示例学习算法应用于手写数字识别．实验证明，基于最大复合的学习算法和ＡＱ１５相比，速度快、得到的公式少、匹配精度高．
This paper proves that the largest compound problem (MGC) in the example learning is an NP difficult problem. An approximate algorithm for solving the maximum compound problem is given. The example learning algorithm is applied to handwritten digital recognition. Experiments show that compared with AQ15, the learning algorithm based on the maximal compound is fast, less formulas are obtained, and the matching accuracy is high.