Evaluation of radial basis function neural network minimizing L-GEM for sensor-based activity recognition

Abstract

Sensor-based activity recognition involves the automatic recognition of a user's activity in a smart environment using computational methods. The use of wearable devices and video-based approaches have attracted considerable interest in ubiquitous computing. Nevertheless, these methods have limitations such as issues with privacy invasion, ethics, comfort and obtrusiveness. Environmental sensors are an increasingly promising consideration in the ubiquitous computing domain for long-term monitoring, as these devices are non-invasive to inhabitants, yet certain challenges remain with activity recognition in sensorised environments, for example, addressing the challenge of intraclass variation between activities and reasoning from low-level uncertain information. In an effort to address these challenges, this paper proposes and evaluates the performance of a Radial Basis Function Neural Network approach for activity recognition with environmental sensors. The model is trained using the Localized Generalization Error and focuses on the generalization ability by considering both the training error and stochastic sensitivity measure. This measures the network output fluctuation with respect to the minor perturbation of input, to address the tolerance of the low-level uncertain sensor data. This approach is compared with three benchmark Neural Network approaches, including a popular deep learning approach using an Autoencoder, and it is evaluated with a simulated dataset as well as a number of publicly available datasets. The proposed method has shown advantages over the other models for all four evaluated datasets. This paper provides insights into the importance of model generalization abilities and an initial analysis of the limitation of deep Neural Networks with respect to sensor-based activity recognition.

Radial Basis Function Neural Network with Localized Stochastic-Sensitive Autoencoder for Home-Based Activity Recognition

“…These methods all face the problem of not being able to fully accommodate instances of unreliable sensor data [14,15]. Hong et al [15] proposed a framework to handle this situation.…”

“…RBFNN is a classic type of ANN, which has efficient training speed and the capability of approximating a function with any precision rate given enough hidden neurons [14]. Therefore, RBFNN is widely used in various fields [41][42][43].…”

“…Therefore, RBFNN is widely used in various fields [41][42][43]. Zhang et al [14] focused on strengthening the capabilities of models, and proposed an RBFNN based on LGE to address the tolerance issues surrounding low-levels of uncertain sensor data. This method focused on improving the generalization ability of the model by minimizing LGE.…”

“…In addition, sensors only obtain binary information, like ON/OFF, OPEN/CLOSE etc., which would not cause privacy issues that could be raised when, for example, using a camera. Home-based activity recognition approaches use sensors to capture various interactive information generated by inhabitants within their home environment, and subsequently extract key information to identify user activities [14]. These methods recognize the activities of daily living, like watching TV, taking a shower, using the toilet and sleeping, rather than simple actions such as raising hands or sitting down as may be the case in image analysis.There are some challenges associated with the home-based activity recognition.…”

“…Firstly, by contrast with the continuous data collected from cameras or the triaxial acceleration sensor in smart wearable devices, home-based sensors data are mostly binary, which carry less information. Therefore, the collected sensors data may not be enough to provide a full representation of the activities [14]. Secondly, sensors data may be unreliable due to the transmission error or hardware problems [15].…”