摘要
针对现实中交通正常运行状态远多于事件状态这一事实,提出了面向不平衡数据集的交通事件检测算法。运用SMOTE(Synthetic Minority Over-sampling Technique)算法重构训练集,使之平衡,以支持向量机(Support VectorMachine,SVM)作为分类器,对交通事件进行检测。使用美国I-880高速公路获取的交通数据进行算法的训练和性能测试。结果表明,基于SMOTE-SVM的交通事件自动检测(Automatic Incident Detection,AID)算法可以提高检测率,减少平均检测时间。
Considering the fact that the traffic condition of normal operation is far more than that of event, we pro posed an AID （Automatic Incident Detection） algorithm based on imbalanced datasets. The SMOTE （Synthetic Minority Over-sampling Technique） algorithm was used to reconstruct and balance the training set, the SVM （Support Vector Ma chine） was adopted as the classifier to detect traffic incidents. Using actual traffic data from the America I-880 freeway, the proposed AID algorithm was trained and the performance of this algorithm was tested. The experimen results show that the AID algorithm based on SMOTE-SVM can improve the detection rate and reduce the testing time.
出处
《武汉理工大学学报》
CAS
CSCD
北大核心
2012年第11期58-62,123,共6页
Journal of Wuhan University of Technology
基金
国家自然科学基金(61074141)
国家"863"计划(2011AA110302)
