An Incremental Approach of Clustering for Human Activity Discovery

One of the challenges in human activity recognition is the ability for an intelligent system to discover the activity models by itself. In this paper, we propose an incremental approach to discover human activities from unlabeled data using K-means. The approach does not require prior specification of the number of clusters, or k-value, and has the ability to reject random movements or noise. Simple algorithm is used making the approach easy to implement without requiring any prior knowledge in the data. We evaluated the effectiveness of the approach and the results show more than 30% improvement in precision and 19% improvement in recall when compared to the results obtained using a non-incremental approach with cluster validity index. The achievement in human activity discovery will enable the wide adoption of human activity recognition technologies in the natural human living environment where labeled data are not available.

[1]  Bernt Schiele,et al.  Weakly Supervised Recognition of Daily Life Activities with Wearable Sensors , 2011, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[2]  J.K. Aggarwal,et al.  Human activity analysis , 2011, ACM Comput. Surv..

[3]  J. MacQueen.  Some methods for classification and analysis of multivariate observations , 1967 .

[4]  T. Caliński,et al.  A dendrite method for cluster analysis , 1974 .

[5]  John A. Hartigan,et al.  Clustering Algorithms , 1975 .

[6]  Bernt Schiele,et al.  Discovery of activity patterns using topic models , 2008 .

[7]  W. Krzanowski,et al.  A Criterion for Determining the Number of Groups in a Data Set Using Sum-of-Squares Clustering , 1988 .

[8]  Zhengyou Zhang,et al.  Microsoft Kinect Sensor and Its Effect , 2012, IEEE Multim..

[9]  Takafumi Koseki,et al.  Unsupervised Human Activity Detection with Skeleton Data from RGB-D Sensor , 2013, 2013 Fifth International Conference on Computational Intelligence, Communication Systems and Networks.

[10]  Ronald Poppe,et al.  A survey on vision-based human action recognition , 2010, Image Vis. Comput..

[11]  Shengrui Wang,et al.  ADR-SPLDA: Activity discovery and recognition by combining sequential patterns and latent Dirichlet allocation , 2012, Pervasive Mob. Comput..

[12]  Anil K. Jain.  Data clustering: 50 years beyond K-means , 2008, Pattern Recognit. Lett..

[13]  Lifeng Sun,et al.  A Matrix-Based Approach to Unsupervised Human Action Categorization , 2012, IEEE Transactions on Multimedia.

[14]  Juan Carlos Niebles,et al.  Unsupervised Learning of Human Action Categories , 2006 .

[15]  P. Rousseeuw.  Silhouettes: a graphical aid to the interpretation and validation of cluster analysis , 1987 .

[16]  Donald W. Bouldin,et al.  A Cluster Separation Measure , 1979, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[17]  Ujjwal Maulik,et al.  Performance Evaluation of Some Clustering Algorithms and Validity Indices , 2002, IEEE Trans. Pattern Anal. Mach. Intell..

[18]  Irfan A. Essa,et al.  Unsupervised Activity Discovery and Characterization From Event-Streams , 2005, UAI.

[19]  Rama Chellappa,et al.  Machine Recognition of Human Activities: A Survey , 2008, IEEE Transactions on Circuits and Systems for Video Technology.