Unsupervised learning for spectral data analysis as a novel sensor for identifying rodent infestation in urban environments

Abstract

Rodent urine is known to fluoresce. This research aims to use spectral imaging data to detect rodent activity via chromophores. We introduce unsupervised learning techniques for classification and clustering of rodent urine samples from the spectral data directly. We classify and compare the rodent urine against additional chemical compounds such as human urine and coffee to validate our analysis and models. In order to facilitate the visualisation of the chemical compound's spectral data, we use manifold techniques for spectral clustering visualisation.