一种基于自适应支持向量机的异常点检测方法

An Outliers Detection Method Based on Adaptive Support Vector Machine

为了从海量的移动目标轨迹数据中识别异常轨迹点数据,提出了一种基于自适应单类支持向量机的轨迹异常点检测方法。首先,提取轨迹点的运动特征构造特征向量作为模型输入;其次,基于粒子群算法构造最优单类支持向量机模型;最后,利用最优单类支持向量机模型识别异常轨迹点。实验结果表明,新提出的方法能够自适应地构造最优单类支持向量机模型,并有效识别轨迹数据中的异常点,具有很好的自适应性与准确性。

In order to recognize abnormal trajectory pointdata from massive moving target trajectory data,a trajectory outlier detection method based on adaptive one-class support vector machine is proposed.Firstly,the motion features of the trajectory points are extracted to construct the feature vector as model input.Secondly,particle swarm optimization is used to construct the optimal one-class support vector machine model.Finally,the optimal one-class Support Vector Machine model is used to detect outliers.The application results show the newly proposed method can adaptively construct the optimal one-class support vector Machine model and effectively detect outlier from mass moving target trajectory data,and it has with good adaptability and accuracy.