|
Abstract:
|
Recent years, with the increasing development of sensor and communication technologies for vehicles, multi-channel sensor signals for vehicles can be obtained in real time with a high rate (e.g., per second or millisecond). Hence, data generated by these sensors are increasing enormously with high volume. From the online signals, certain events can be detected, and it is often of interest to categorize those events into different types. Thus, different actions can be taken for different types of events. In this paper, we treat the sensory data as multivariate functional data. Because not all variables are useful for clustering, it is important to include only those relevant variables in the clustering procedure. We propose a clustering algorithm for multivariate functional data with automatic variable selections. We investigate the performance of the developed procedure via simulations. The online vehicle sensory data are used for illustrations.
|
