Decision-making based on a series of diagnoses

The main objective of the research was to investigate the possibility of comprehensively identifying and categorizing bee behaviors via the analysis of their produced sound. This article discusses the extraction of information from bee sounds to establish criteria for facilitating decision-making in apiary management. The subject of the study was the actual sounds made by a colony of honey bees. The sounds recorded during various insect behaviors were studied. The process of asynchronous averaging of sound samples in order to reduce the variability of their values is described. Diagnostic vectors were created from raw averaged sound samples and dimensionality reduction by various methods. The results of the study were presented as images of the reduced diagnostic observation space, illustrating the states and selected diagnostic sequences. The impact of data dimensionality reduction on the classification efficiency of the studied bee behaviors was evaluated. Attempts to classify unknown sounds were conducted using expert-selected recordings of real sounds.