【摘要】 针对传统KNN算法手写数字识别速度和精度都不高的问题,提出了一种改进的KNN手写数字识别方案。改进的KNN算法利用PCA方法对手写数字进行降维,降低传统KNN算法在距离计算过程中的时间和空间复杂度;并采用距离权重统计改善手写数字识别错误问题。在手写数据集MNIST上验证改进的KNN算法,结果显示识别速度有显著提升,识别准确度有一定提高。更多还原

【Abstract】 Using traditional KNN algorithm to recognize handwritten numerals,the speed and accuracy are not high. So an improved KNN handwritten numeral recognition scheme is proposed. In order to improve the recognition speed,the improved KNN algorithm uses PCA to reduce the dimension of handwritten number which reduce the time and space complexity in the similarity calculation process. And it uses distance weight statistics to improve the accuracy of handwritten numeral recognition. Experiments show that,the recognition speed of the improved KNN algorithm is much higher than traditional KNN,and the recognition accuracy is better.更多还原