基于量子竞争决策算法的电力营销信息多标签分类方法

Multi-label Classification Method of Power Marketing Information Based on Quantum Competitive Decision Algorithm

为了提升分类稳定性,利用量子竞争决策算法求解分类问题优势,研究了基于量子竞争决策算法的电力营销信息多标签分类方法。通过轨迹计算求解节点结构相似性,将节点相似性作为输入更新隐藏层向量,确保全局结构稳定,构建深度自动编码器的多标签分类模型;采用节点邻接数据获取网络的高度非线性结构得出节点向量;通过量子竞争决策算法对该模型目标函数进行寻优,获取目标的最优解,并通过量子位简化群体规模、增强优化效率;采用进化博弈论理论,提高寻优能力,实现电力营销信息多标签快速分类。实验结果表明,所提出的方法寻优能力、收敛率、稳定性优良,分类效果较好,且平均排序损失值低、平均精度值高,并且可以充分利用节点的结构相似性,得出最佳宏平均和微平均值,电力营销信息多标签分类具备有效性。

In order to improve the classification stability,a multi label classification method of power marketing information based on quantum competitive decision algorithm is studied.The node structure similarity is solved by trajectory calculation,and is used as the input to update the hidden layer vector to ensure the global structure stability.A multi label classification model of depth automatic encoder is constructed,and the node vector is obtained by using the node adjacency data to form the highly nonlinear structure of the network.The objective function of the model is optimized by quantum competitive decision algorithm.The optimal solution of the target is obtained,and the qubit is used to simplify the group size and enhance the optimization efficiency.The evolutionary game theory is used to improve the optimization ability and realize the multi label fast classification of power marketing information.The experimental results show that the method has good optimization ability,convergence rate,stability,good classification effect,low average sorting loss and high average precision,and can make full use of the structural similarity of nodes to obtain the best macro average and micro average,which has the effectiveness of multi label classification of power marketing information.