Fusion of Unobtrusive Sensing Solutions for Home-Based Activity Recognition and Classification Using Data Mining Models and Methods

Abstract

This paper proposes the fusion of Unobtrusive Sensing Solutions (USSs) for human Activity Recognition and Classification (ARC) in home environments. It also considers the use of data mining models and methods for cluster-based analysis of datasets obtained from the USSs. The ability to recognise and classify activities performed in home environments can help monitor health parameters in vulnerable individuals. This study addresses five principal concerns in ARC: (i) users’ privacy, (ii) wearability, (iii) data acquisition in a home environment, (iv) actual recognition of activities, and (v) classification of activities from single to multiple users. Timestamp information from contact sensors mounted at strategic locations in a kitchen environment helped obtain the time, location, and activity of 10 participants during the experiments. A total of 11,980 thermal blobs gleaned from privacy-friendly USSs such as ceiling and lateral thermal sensors were fused using data mining models and methods. Experimental results demonstrated cluster-based activity recognition, classification, and fusion of the datasets with an average regression coefficient of 0.95 for tested features and clusters. In addition, a pooled Mean accuracy of 96.5% was obtained using classification-by-clustering and statistical methods for models such as Neural Network, Support Vector Machine, K-Nearest Neighbour, and Stochastic Gradient Descent on Evaluation Test.

Data Mining and Fusion Framework for In-Home Monitoring Applications

“…Whilst the conceptual methodology informed the initial selection of software packages for the research, the experimental methodology was adopted for testing with real data obtained during sprained ankle rehabilitation exercises. The basis for the preliminary consideration of the software packages included the ability to recognise and categorise binary images, unsupervised feature extraction [54], CbyC capabilities [22,24,26], and ease of data fusion. The experimental methodology involved data collection processes with the aid of single, homogeneous, and heterogeneous SSs.…”

“…Neural Network E.g., stock market prediction [13] Partition-based E.g., medical datasets analysis [14] Decision Tree E.g., Banking and finance [15] Model-based E.g., multivariate Gaussian mixture model [16] Support Vector Machine E.g., big data analysis [17] Grid-based E.g., large-scale computation [18] Association-based E.g., high dimensional problems [19] Density-based Applications with noise. E.g., DBSCAN [20] Bayesian E.g., retrosynthesis [21] Hierarchy-based E.g., Mood and abnormal activity prediction [22,23] Data clustering techniques such as partition-based, model-based, grid-based, densitybased and hierarchical clustering can be used for data grouping [8]. Whilst the densitybased approach is centred on the discovery of non-linear structures in datasets, model-and grid-based methods utilise neural networks and grids creation, respectively.…”