discussion / Acoustics  / 7 September 2023

unsupervised machine learning to infer syntax and temporal organisations of animal vocalizations

I wish to share an update on my MSc thesis project, that contributes to the field of decoding animal communication. 
In my work I conducted a factorial experiment  to improve machine learning techniques to infer syntax and temporal organizations in animal communication, in this study focusing on fruit bats.

By using an dataset of vocalizations only annotated with behavioural context, I researched to improve unsupervised machine learning approach to quantify the animal repertoire (the types of "phonemes") , which was then used to explore syntactical rules.

I found combinatorics patterns in context-dependent vocalisations of fruit-bats, with an approach that can help in exploring their properties (e.g.  distribution of vocal patterns, or properties of compositions).

I am happy to have learnt skills that can be applied to different species.

I am indeed exploring for PhDs in decoding animal communication by means of machine learning.

I am also looking at on-field opportunities - examples are on-field data-collection and evaluation - because on-field experience is important for the type of research I'd like to carry out.

Always interested in  collaboration opportunities or working opportunities where I can contribute with this new skillset (which I see just a starting point) and with my background and experience in digital innovation (where I hold more "seniority" and can coach, contribute as design/developer/data-scientists - I hold qualification in all these areas - or as project manager - I hold experience in tech startups).

As soon as the graduating process will be finished, I will post the link to my work.
Meanwhile, happy to network!