一种带有缺失标记的不完备数据的多标记分类方法及装置

• Format: Patent
• Language: Chinese
• Published: 15.10.2024
• Subjects: CALCULATING; COMPUTING; COUNTING; DATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FORADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORYOR FORECASTING PURPOSES; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE,COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTINGPURPOSES, NOT OTHERWISE PROVIDED FOR

本发明涉及一种带有缺失标记的不完备数据的多标记分类方法及装置，属于数据分类技术领域。本发明首先基于邻域粗糙集理论，通过样本之间的差异性和相似性，构造了邻域可辨识和不可辨识矩阵，以此恢复不完备的信息，并得到所恢复信息的特征权重矩阵；然后基样本之间的模糊相似关系，结合模糊相似关系、回归模型以及特征权重矩阵建立考虑特征之间非线性关系的新的目标函数，并通过梯度下降方法对其进行优化求解，从而实现对带有缺失标记的不完备数据的多标记分类。本发明充分考虑了特征之间的非线性关系，大大提高了带有缺失标记的不完备数据的多标记分类的精度和效率。 The invention relates to a multi-label classification method and device for incomplete data with missing labels, and belongs to the technical field of data classification. According to the invention, firstly, on the basis of a neighborhood rough set theory, neighborhood recognizable and unrecognizable matrixes are constructed through differences and similarities among samples, so that incomplete information is recovered, and a feature weight matrix of the recovered information is obtained; then, based on the fuzzy similarity relation between the samples, a new objective function considering the nonlinear relation between the features is established in combination with the fuzzy similarity relation, a regression model and a feature weight matrix, opti