2021/05/01 by Michelangelo Acconcjaioco, Acconcjaioco, Michelangelo, Stavros Ntalampiras +1 · 1 citation
Biochemistry, Genetics and Molecular Biology · Environmental Science · #Animal Vocal Communication and Behavior #Artificial Intelligence (cs.AI) #Audio and Speech Processing (eess.AS) #FOS: Computer and information sciences #FOS: Electrical engineering #Machine Learning (cs.LG) #Rangeland and Wildlife Management #Sound (cs.SD) #Spider Taxonomy and Behavior Studies #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2105.00202
openalex publication_date 2021/05/01 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
This work introduces the one-shot learning paradigm in the computational\nbioacoustics domain. Even though, most of the related literature assumes\navailability of data characterizing the entire class dictionary of the problem\nat hand, that is rarely true as a habitat's species composition is only known\nup to a certain extent. Thus, the problem needs to be addressed by\nmethodologies able to cope with non-stationarity. To this end, we propose a\nframework able to detect changes in the class dictionary and incorporate new\nclasses on the fly. We design an one-shot learning architecture composed of a\nSiamese Neural Network operating in the logMel spectrogram space. We\nextensively examine the proposed approach on two datasets of various bird\nspecies using suitable figures of merit. Interestingly, such a learning scheme\nexhibits state of the art performance, while taking into account extreme\nnon-stationarity cases.\n