Analyzing Big Remote Sensing Data via Symbolic Machine Learning

This paper discusses the applicability of a recently introduced method for Earth Observation data and justifies eligibility in the framework of Big Data processing and. The main novelty that this classifier brings forward the Remote Sensing domain is the automated learning from representations as opposed to the traditional quantitative. The method has been designed to work as a pot at which, data arriving from diverse sources and heterogeneous characteristics (sensors, scales, time periods) fused and refined to valuable information. A description the classification method is presented briefly. In, results from the production of the multi-temporal resolution Global Human Settlement Layer that derived Landsat satellite imagery of the last 40 years are and assessed.