A holographic memory approach for pollution forecasting in a high-density urban environment

Abstract

In this work, the Holographic Associative Memory (HAM) paradigm was used as the core of a forecasting software tool for benzopyrene estimations near a highly populated zone. The presented tool was trained with data coming from a monitoring station near a steel plant in Genova, Italy. The decoding of test stimuli was performed with two different methods, the holographic complex number technique (HCD) and the closest holographic neighbor decoding (CHN). The cost–performance relation of both methods is outlined and compared. The atmospheric scenarios used for modeling benzopyrene behavior contained meteorological and chemical variables correlated to the formation and dispersion of such contaminant. The obtained results show an accurate performance of the HAM method either for identifying the main features involved in benzopyrene estimation and for the forecasting itself. Finally, some concluding remarks regarding the performance of both decoding methods are presented.