Many Facets of Energy Disaggregation

In this week's Ebiquity meeting, PhD student Nilavra Phatak from the UMBC Information Systems Department will discuss his work on "Many Facets of Energy Disaggregation".

The objective of energy disaggregation is to get the appliance-wise energy consumption from the whole home energy signal. The disaggregated energy consumption provides a better insight into the electrical usage and helps the consumers to modify usage in order to save money and energy. The application of energy disaggregation is much more useful for utility providing companies the disaggregated information for energy forecasting, energy pricing, setting up competitive power saving scenarios etc. The key challenge lies for a big data scenario where data is collected continuously from millions of houses by the utility companies. The greatest obstacle is coming across with labels for the data. The discussion will be regarding modelling problem as supervised, unsupervised and semi-supervised learning which requires time-series learning and also the application of some of the new trending topics like active learning.