ABSTRACT
The application of artificial intelligence enhances the ability of sensor and networking technologies to realize smart systems that sense, monitor and automatically control our everyday environments. Intelligent systems and applications often automate decisions based on the outcome of certain machine learning models. They collaborate at an ever increasing scale, ranging from smart homes and smart factories to smart cities. The best performing machine learning model, its architecture and parameters for a given task are ideally automatically determined through a hyperparameter tuning process. At the same time, edge computing is an emerging distributed computing paradigm that aims to bring computation and data storage closer to the location where they are needed to save network bandwidth or reduce the latency of requests. The challenge we address in this work is that hyperparameter tuning does not take into consideration resource trade-offs when selecting the best model for deployment in smart environments. The most accurate model might be prohibitively expensive to computationally evaluate on a resource constrained node at the edge of the network. We propose a multi-objective optimization solution to find acceptable trade-offs between model accuracy and resource consumption to enable the deployment of machine learning models in resource constrained smart environments. We demonstrate the feasibility of our approach by means of an anomaly detection use case. Additionally, we evaluate the extent that transfer learning techniques can be applied to reduce the amount of training required by reusing previous models, parameters and trade-off points from similar settings.
ABSTRACT
Sensors provide the foundation of many smart applications and cyber-physical systems by measuring and processing information upon which applications can make intelligent decisions or inform their users. Inertial measurement unit (IMU) sensors-and accelerometers and gyroscopes in particular-are readily available on contemporary smartphones and wearable devices. They have been widely adopted in the area of activity recognition, with fall detection and step counting applications being prominent examples in this field. However, these sensors may also incidentally reveal sensitive information in a way that is not easily envisioned upfront by developers. Far worse, the leakage of sensitive information to third parties, such as recommender systems or targeted advertising applications, may cause privacy concerns for unsuspecting end-users. In this paper, we explore the elicitation of age and gender information from gait traces obtained from IMU sensors, and systematically compare different feature engineering and machine learning algorithms, including both traditional and deep learning methods. We describe in detail the prediction methods that our team used in the OU-ISIR Wearable Sensor-based Gait Challenge: Age and Gender (GAG 2019) at the 12th IAPR International Conference on Biometrics. In these two competitions, our team obtained the best solutions amongst all international participants, and this for both the age and gender predictions. Our research shows that it is feasible to predict age and gender with a reasonable accuracy on gait traces of just a few seconds. Furthermore, it illustrates the need to put in place adequate measures in order to mitigate unintended information leakage by abusing sensors as an unanticipated side channel for sensitive information or private traits.
