Authors: Manassas, A.; Samaras, T.;
ΒioEM 2024, Chania Crete, 16-21 June, 2024
Abstract
The scope of this ongoing work is to build Artificial Intelligence (AI) models using either Machine Learning (ML) or Deep Learning (DL) techniques to predict electromagnetic field values at ground level. The target values are collected either by spot measurements using a spectrum analyzer or by drive test measurements using an electric bike equipped with an exposimeter and GPS. So far, the focus of the work has been on collecting publicly available data, known as ‘features’, that can contribute to predicting the target values. These features include the distance between the points of interest (POI) and the antennas, the transmitted power of the antennas, and information collected by GIS systems, including the built environment around the POI. Special attention was given to validating and correcting the accuracy of feature data, a process known as data cleaning in AI.