Internet of Things - Automatic Classification of Sensor Data
Written by: Dewire Labs Team
October 10, 2016
Currently, large amounts of data from numerous sensors stream rapidly into interconnected IT systems for processing and storage. Gartner project that the 0.9 billion sensors and 1.6 billion personal devices that were connected via the Internet in 2009 will become up to 75 billion devices in 2020, constituting the Internet of Things.
One approach to handle steams of data is to use a machine learning application that that classifies unlabeled sensor data based on previous knowledge. For example, measure the temperature in an office and automatically classify the measured temperatures as hot, cold or pleasant based on previously identified employee preferences.
- How can relations among attributes be revealed on live sensor data and the causal factors are reduced to ensure a correct and feasible live data analysis?
- How is it possible to leverage distribution techniques in order to decentralize the analysis on large scale industry networks?