基于聚类算法的二分类问题研究

Research on binary classification problems based on clustering algorithms

分类预测是机器学习的基础任务,在机器视觉、文本分析、在线广告等领域均有广泛的应用,对行业发展具有极大的促进作用。随着信息技术的发展,数据规模不断扩大,复杂的高维数据使得传统的分析方法变得困难,以至于现有的深度学习模型在对复杂数据集进行分类预测时,常常出现预测性能不够理想的情况。在广告点击预测领域,通过引入聚类方法,充分利用数据内在的隐式关系,有助于构建更准确、鲁棒性更好的分类模型。

Classification prediction is a basic task of machine learning,which has a wide range of applications in machine vision,text analysis and online advertising,etc.,and has a great promoting effect on the development of the industry.With the evolution of information technology and the increasing size of data,complex and high-dimensional data make traditional analytical methods difficult,such that existing deep learning models often have less than ideal prediction performance when classifying and predicting complex datasets.In the field of ad click prediction,the introduction of clustering methods,which make full use of the implicit relationships inherent in the data,helps to construct more accurate and robust classification models.