基于PCA降维的MNIST手写数字识别优化
Optimization of MNIST Handwritten Digit Recognition Based on PCA Dimensionality Reduction
田春婷1
作者信息
- 1. 兰州石化职业技术大学 信息工程学院,甘肃 兰州 730207
- 折叠
摘要
PCA数据降维技术广泛应用于数据降维和数据的特征提取,可以很大程度上降低算法的计算复杂度,提升程序运行效率.文章将MNIST原始数据集和对原始数据集进行PCA降维处理之后的数据集作为样本,分别采用K-邻近算法、决策树ID3 算法、SVC分类模型,以及选取不同分类算法作为基础分类器的集成学习方法,实现手写数字识别.在对MNIST数据集进行PCA降维前后,以及不同分类算法和模型执行结果的时间复杂度与预测准确率进行比对与分析,进一步强化与优化手写数字识别准确率等各项指标.
Abstract
PCA data dimensionality reduction technology is widely used in data dimensionality reduction and feature extraction,which can greatly reduce the computational complexity of algorithms and improve program efficiency.This paper takes the MNIST original dataset and the dataset after PCA dimensionality reduction as samples,and uses K-Nearest Neighbor algorithm,Decision Tree ID3 algorithm,SVC classification model,as well as Ensemble Learning methods that select different classification algorithms as basic classifiers to achieve handwritten digit recognition.It compares and analyzes the time complexity and prediction accuracy of different classification algorithms and models before and after PCA dimensionality reduction on the MNIST dataset,further enhances and optimizes various indicators such as handwritten digit recognition accuracy.
关键词
PCA降维/MNIST手写数字识别/K-邻近算法/决策树/SVC分类模型/集成学习Key words
PCA dimensionality reduction/MNIST handwritten digit recognition/K-Nearest Neighbor algorithm/Decision Tree/SVC classification model/Ensemble Learning引用本文复制引用
基金项目
甘肃省教育厅高校教师创新项目(2023A-205)
出版年
2024
国家科技期刊平台
NSTL
万方数据