基于差异性聚类的选择性集成人体行为识别模型

doi:10.11896/j.issn.1002-137X.2018.01.053

摘要/Abstract

摘要： 为了提高基于智能手机的人体行为识别率,优化多分类器集成系统的泛化性能及个体分类器的差异性,提出了基于差异性增量聚类(Diversity Measure Increment-Affinity Propagation clustering,DMI-AP)的选择性集成人体行为识别模型。首先对训练集的所有样本进行bootstrap抽样并训练基分类器,选出大于平均识别率的基分类器构成分类器集合；然后将集合的基分类器作为聚类对象进行分组,通过计算基分类器间的双误差异性值求出表征个体分类器特征的双误差异性增量值,输入近邻传播聚类算法得到k个类簇,选取每簇的中心分类器构成多分类器集成系统；最后使用等概率均值法融合k个分类器的输出结果。实验表明,该模型算法使个体分类器的差异性增大、分类器搜索空间缩小；与传统的Bagging,Adaboost以及RF方法相比,该模型的识别准确率平均提高了8.11%。

关键词: 选择性集成,差异性增量,近邻传播聚类,行为识别

Abstract: To improve the accuracy of human activity recognition based on mobile phone,and optimize the generalization performance of multiple classifiers ensemble system and the diversity of individual classifier,an activity recognition model based on selective ensemble learning of diversity measure increment-affinity propagation clustering(DMI-AP) was proposed.Firstly,all the samples are bootstrapped and base classifiers are trained in the training set. The mode selects the base classifiers whose accuracy is greater than the average accuracy. The classifier set consists of the selected classifiers,and then the base classifiers of the training set are chosen to cluster,the double default diversity increment value are got by calculating the double default diversity measure value between base classifiers.The value is clustered by the affinity propagation clustering algorithm and divided into k clusters.Each cluster’s center classifier forms multi-classifier systems.Finally,the outputs of classifiers are fused by calculating the average.The experimental results show that the diversity of individual classifier increases and the searching space of the classifier decreases by using the DMI-AP model.Compared with the traditional Bagging,Adaboost and RF methods,the recognition accuracy of the proposed model is improved by 8.11%.

Key words: Selective ensemble,Diversity measure increment,Affinity propagation clustering,Activity recognition

王忠民,张爽,贺炎. 基于差异性聚类的选择性集成人体行为识别模型[J]. 计算机科学, 2018, 45(1): 307-312. https://doi.org/10.11896/j.issn.1002-137X.2018.01.053

WANG Zhong-min, ZHANG Shuang and HE Yan. Selective Ensemble Learning Human Activity Recognition Model Based on Diversity Measurement Cluster[J]. Computer Science, 2018, 45(1): 307-312. https://doi.org/10.11896/j.issn.1002-137X.2018.01.053

参考文献

[1] CAGATAY C,SELIN T,ELIF P,et al.On the use of ensemble of classifiers for accelerometer based activity recognition[J].ELS-EVIER Applied Soft Computing,2015,7(C):1018-1022.
[2] CHETTY G,WHITE M,AKTHER F.Smart Phone Based Data Mining for Human Activity Recognition [J].Procedia Computer Science,2015,6:1181-1187.
[3] JUREK A,NUGENT C,BI Y,et al(1)Clustering-Based Ensemble Learning for Activity Recognition in Smart Homes[J].Sensors,2014,14(7):12285-12304.
[4] HENG X,WANG Z M.Human Activity Recognition Based on Accelerometer Data from a Mobile Phone[J].Journal of Xi’an University of Posts and Telecommunications,2014,19(6):76-79.(in Chinese) 衡霞,王忠民.基于手机加速度传感器的人体行为识别[J].西安邮电大学学报,2014,9(6):76-79.
[5] WANG Z M,CAO D.A Feature Selection Method for Behavior Recognition Based on Ant Colony Algorithm[J].Journal of Xi’an University of Posts and Telecommunications,2014,9(1):73-77.(in Chinese) 王忠民,曹栋.基于蚁群算法的行为识别特征优选方法[J].西安邮电大学学报,2014,9(1):73-77.
[6] TANG C,WANG W J,LI W,et al.Human Action Recognition Algorithm Based on Selective Ensemble Rotation Forest[J].Pattern Recognition and Artificial Intelligence,2016,9(4):313-321.(in Chinese) 唐超,王文剑,李伟,等.基于选择性集成旋转森林的人体行为识别算法[J].模式识别与人工智能,2016,9(4):313-321.
[7] WANG Z M,WANG B.Feature Selection Method for Moblie User Behavior Recognition Based on Multiband Time Domain Decomposition[J].Application Research of Computers,2015,32(7):1956-1958.(in Chinese) 王忠民,王斌.多频段时域分解的行为识别特征优选方法[J].计算机应用研究,2015,2(7):1956-1958.
[8] LI H,DING S F.AP Twice Clustering Based Neural NetworkEnsemble Algorithm[J].Computer Science,2015,42(2):224-227.(in Chinese) 李辉,丁世飞.基于AP二次聚类的神经网络集成算法研究[J].计算机科学,2015,2(2):224-227.
[9] YUAN Y,WANG C,ZHANG J Z,et al(1)An Ensemble Approach for Activity Recognition with Accelerometer in Mobile-phone[C]∥Proceedings of the 17th IEEE International Conference on Computational Science and Engineering (CSE).2014:1469-1474.
[10] ZHANG C X,ZHANG J S.A Review of the Selective Ensemble Learning Algorithm [J].Journal of Computer Science,2011,34 (8):1399-1410.(in Chinese) 张春霞,张讲社.选择性集成学习算法综述[J].计算机学报,2011,4(8):1399-1407.
[11] ROONEY N,PATTERSON D,NUGENT C.Non-strict heterogeneous Stacking[C]∥ International Workshop.2009:478-487.
[12] ZHANG C X,DUIN R P.An Empirical Study of a Linear Regression Combiner on Multi-class Data Sets[C]∥ International Workshop on Multiple Classifier Systems.2009:478-487.
[13] ZHOU Z H,CHEN S F.Neural Network Ensemble[J].Chinese Journal of Computers,2002,5(1):1-8.
[14] LAZAREVIC A,OBRADOVIC Z.Effective pruning of neural network classifier ensembles[J].International Joint Conference on Neural Networks,2001(2):796-801.
[15] FREY B J,DUECK D.Clustering by Passing Messages Between Data Points[J].Science,2007,5(5814):927-976.
[16] KUNCHEVA L I,WHITAKER C J.Measures of Diversity in Classifier Ensembles and Their Relationship with Ensemble Accuray [J].Machine Learning,2003,1(2):187-207.
[17] DENG W Y,ZHENG Q H,WANG Z M.Cross-person activity recognition using reduced kernel extreme learning machine[J].ELSEVIER Neural Networks,2014,3(5):1-7.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed