Learning the Influence Structure between Partially Observed Stochastic Processes Using IoT Sensor Data

The recent widespread of availability of sensors, as part of the IoT, presents the opportunity to learn the properties of compound distributions in practical applications. Understanding temporal distributions by observations collected from the IoT can advance many intelligent applications. In this paper we develop an algorithm to learn influence between stochastic processes using observations obtained from the IoT. The proposed method learns these processes using temporal models independently, and then attempts to recover the underlying distribution of influence between them. Experimental results are provided which demonstrate the effectiveness of our method. This approach is useful in applications that require an understanding of how partially observed high-level processes can influence each other given a set of observations at different times.