Skip navigation
PDF 986 KB
DOI: http://dx.doi.org/10.7551/978-0-262-33936-0-ch086
Page 536
First published July 1 2016

Learning Cassins Vireo (Vireo cassinii) syntax through grammatical inference

Julio G. Arriaga, Richard Hedley, Edgar Vallejo, and Charles Taylor

Abstract (Excerpt)

Birdsong may be regarded as a complex adaptive system. In this paper we study the relationship between complexity and consistency of an evolving model of Cassins Vireo syntax, using a genetic algorithm to approximate a Minimal Consistent Deterministic Finite-state Automata (MCDFA) capable of accepting vocal sequences produced by birds of the study species. Our results imply that, despite the complex vocal behaviour of this species, the complexity of the model can be reduced considerably to encompass all of the positive samples while retaining the ability to exclude similar negative samples. These results suggest the existence of important regularities in the song sequences of this species.