Automatic construction of accurate bioacoustics workflows under time constraints using a surrogate model

Automated bioacoustics analysis is being increasingly used to describe environmental phenomena such as species abundance and biodiversity. Within this research area, many algorithms have been proposed. These achieve different sub-objectives within processes can be combined form workflows. However, these are typically evaluated in a limited number of scenarios rarely with combinations other tasks. This result workflows that not well optimised serve given scenario, particularly under resource time constraints, which ultimately leads inaccurate analyses. work examines the problem workflow construction by searching ordering tasks determine produce most accurate output while remaining user-defined constraints. Workflow investigated scenario where need classified synthetically generated soundscapes numbers species, noise levels, densities species. A search algorithm created applies Particle Swarm Optimisation (PSO) neural network-based surrogate model. efficiently for candidate structures. compared random search, genetic algorithm, PSO without model, existing based on previous research. It found all scenarios, model-based method quickly find effective low searches. Furthermore, it effectiveness varies depending recordings used. • Bioacoustics picking task build approach quickly. builds good quicker than genetic, PSO, algorithms.