曾光辉
（广州工程技术职业学院，广东 广州 510900）
摘要：为提升网络信息的识别与分类准确率，针对海量网络信息的高维、高噪等特点，提出基于匹配自主学习的网络信息识别与分类算法。首先，利用支持向量机对网络信息进行识别；然后，通过奇异值分解算法构建检索矩阵进行奇异值分解、相关性查询；接着，计算网络信息的相似性匹配度，并将匹配度较高的网络信息输入到卷积神经网络中进行学习、训练；最后，输出网络信息分类结果。实验结果显示，该算法网络信息识别准确率达到97.90%以上，针对不同类别网络信息的平均分类准确率为98.04%，证明了该算法在实际应用中的有效性。
关键词：匹配自主学习；网络信息；支持向量机；奇异值分解；卷积神经网络；识别与分类
中图分类号：TP309 文献标志码：A 文章编号：1674-2605(2024)03-0007-06
DOI：10.3969/j.issn.1674-2605.2024.03.007
Network Information Recognition and Classification Algorithm Based on Matching Autonomous Learning
ZENG Guanghui
(Guangzhou Institute of Technology, Guangzhou 510900, China)
Abstract: To improve the accuracy of network information recognition and classification, a network information recognition and classification algorithm based on matching autonomous learning is proposed to address the high dimensionality, high noise and other characteristics of massive network information. Firstly, using support vector machine to recognize network information; Then, a retrieval matrix is constructed using singular value decomposition algorithm for singular value decomposition and correlation queries; Finally, calculate the similarity matching degree of network information, and input the network information with higher matching degree into the convolutional neural network for learning and training, outputting the network information classification results. The experimental results show that the network information recognition accuracy of the algorithm reaches over 97.90%, and the average classification accuracy for different types of network information is 98.04%, which has certain practical application effectiveness.
Keywords: matching autonomous learning; network information; support vector machine; singular value decomposition; convolutional neural network; recognition and classification
