Body Sensor Networking, Design and Algorithms

9Machine Learning for Sensor Networks

9.1 Introduction

‘Machine learning’ refers to a combination of learning from data and their clustering or classification. It is often combined with some signal processing techniques to extract the best discriminating features of the data prior to taking learning and classification steps. The emerging machine learning techniques, such as deep neural networks (DNNs), are also capable of feature learning.

Often, in body sensor networks (BSNs), multimodal data are used, and thus features from various measurement modalities should be combined or exploited. This leads to another area in machine learning called sensor (data) fusion.

Sensors are used almost everywhere. The newly emerging sensor technology is beginning to closely mimic the ultimate sensing machine, i.e. the human being. The sensor fusion technology leverages a microcontroller (which mimics the human brain) to fuse individual data collected by multiple sensors. This allows for a more accurate and reliable realisation of the data than one would acquire from an individual sensor.

Sensor fusion enables context awareness, which has significant potential for the Internet of Things (IoT). Advances in sensor fusion for remote emotive computing (emotion sensing and processing) could also lead to exciting new applications in the future, including smart healthcare. This approach motivates personalised healthcare providers to fine tune and customise systems which best suit individuals' needs. ...