Data Mining as a Service for Sensing Applications

Many applications, for example in mobile devices, collect and store different sensor data. They send the data to a cloud for analysis - either only for per-device feedback, or crowd sourcing, when the individual device will be compared to all the other devices. The aim of running data mining algorithms in the cloud is afterwards to send the results back to the devices. The results can be used for improving the operation of the sensing application. We call this phase in its entirety for Data Mining as a Service (DMaaS). Our research aims to find suitable data mining algorithms for multiple sensor data sources, their optimizations, streaming possibilities, and also separating tasks between the client and the cloud efficiently.