Unsupervised clustering of IoT signals through feature extraction and self organizing maps

This thesis scope is to build a clustering model to inspect the structural properties of a dataset composed of IoT signals and to classify these through unsupervised clustering algorithms. To this end, a feature-based representation of the signals is used. Different feature selection algorithms are then used to obtain reduced feature spaces, so as to decrease the computational cost and the memory demand. Thus, the IoT signals are clustered using Self-Organizing Maps (SOM) and then evaluated