透镜成像对立学习的SMA算法及舆情预测应用

Slime mould algorithm based on lens imaging and opposite-learning and network public opinion prediction application

网络舆情具有小样本特征,而传统方法预测准确率低,容易得到局部最优。为此,提出一种改进黏菌算法优化支持向量机的网络舆情预测模型ISMA-SVM。引入混沌Circle映射机制提高初始种群的多样性;利用对数非线性调节反馈因子均衡算法全局搜索与局部开发;设计透镜成像对立学习机制对最优个体变异,扩展搜索空间并避免算法陷入局部最优。利用改进黏菌优化算法优化支持向量机模型,构建网络舆情预测模型。以若干舆情热点百度指数作为样本进行实证研究,结果表明,改进模型具有更高的数据拟合度,预测准确度更高。

The network public opinion has small sample characteristics,while the traditional method has lower prediction precision,leading to local optimum.A network public opinion prediction model ISMA-SVM based on improved slime mold algorithm with optimizing support vector machine was proposed.The chaotic circle mapping was introduced to improve the diversity of the initial population.The feedback factor was adjusted in a nonlinear way based on logarithm for balancing global search and local development of the algorithm.The lens imaging opposition learning mechanism was designed to disturb the optimal individual variation,expand the search space and avoid the algorithm falling into local optimization.The support vector machine model was optimized using the improved slime mold optimization algorithm,and the network public opinion prediction model was constructed.Some public opinion hot Baidu indexes were used as the sample to conduct an empirical study.The results show that the improved model has higher data fitting degree and prediction accuracy.