Comparison between evidence (Dempster-Shafer) theory and Bayesian theory

Both Bayesian theory and Dempster-Shafer (evidence) theory assign non-negative weights to set of events. In Bayesian theory, the finite set of possible events denoted by , each individual event, is assigned a non-negative weight called probability denoted by . The probabilities satisfy the following properties. for all In Dempster-Shafer theory, the finite set called frame…

Uncertainty in Ontology Approach

Ontology is the most widely used tool to integrate semantic into activity recognition. However, besides lack of temporal reasoning, ontology cannot deal with uncertainty [1]. Diaz et al. proposed a fuzzy ontology to solve this limitation. But, fuzzy approach allows more precise knowledge representation rather than deciding among multiple hypotheses and having a more coherent…

Why dynamic sliding window technique is necessary in activity recognition?

Window size in (acceleration) signal segmentation for activity classification is important because it needs to capture necessary characteristics of the signals. Various sizes of window have been used in previous works without stating a specific reason. A window size is selected based on past experiments and hardware limitations. Majority of approaches used window size in…

Sensor constraints and issues

Wilson and Atkeson [1] discussed several constraints and issues of sensors used in activity recognition system. The two main factors are sensor cost and sensor acceptance. Cameras and microphone are not suitable because inhabitant would feel uncomfortable living in the house. RF tags, badge, beacon or marker type of sensors including any wearable sensors potentially…

Evaluation of accelerometer based multi-sensor versus single-sensor activity recognition systems

The paper describes the comparison of activity recognition system consisting of a single accelerometer sensor and multiple accelerometer sensors. The paper also presented the influence of sampling rate towards recognition accuracies, most widely used features in this research area and the time take to calculate them, and comparison between classifiers in activity recognition. The result…

Features Extraction

Ravi et al. extracted features from raw data using a window size of 256 samples with 128 overlapping between consecutive windows. Sampling rate used is 50Hz which equivalent to 5.12s for each window. The calculated features are mean, standard deviation, energy and correlation. A single triaxial accelerometer was worn near the pelvic region in the…

Window Segmentation Techniques

Preece et al. explained windowing techniques are used in most activity classification researches. In real-time applications, windows are defined concurrently with data collection. There are three different windowing techniques which are sliding windows, event-defined windows and activity-defined windows. The sliding window technique divided signals into fixed length (window) with no inter-window gap. The range of…

Body Area Sensor Networks: Requirements, Operations, and Challenges

B. Johny and A. Anpalagan suggested health-care monitoring challenges can be tackled by interfacing sensors and actuators which form body area networks (BANs) with the human body together with the support of wireless technology and mobile and cloud computing [1]. Three stages of health-care monitoring system have been defined, which are sensors, data hub and…

Context Acquisition

Acquiring context is based on five factors which are responsibility, frequency, context source, sensor type and process of acquisition [1]. Based on responsibility Responsibility factor is about who is making the decision on sensing and communication, either the sensor hardware or user software. This method of context acquisition is called push and pull. It is…