Purdue University
I'm currently working on trying to get my lab group to start using Arbimon, since it has some pretty cool functionality. However, in the Arbimon terms of service, I ran across this text:
"If you post content to the Platform or otherwise submit material (including, but not limited to audio files, audio labels, audio metadata), you grant Rainforest Connection, its affiliates, and third parties that Rainforest Connection has elicited to developing and fine tuning machine learning based AI models, a nonexclusive, royalty-free, perpetual, irrevocable right to use, reproduce, modify, adapt, translate and create derivative works using your content or material."
While I think this is probably fine for a lot of folks doing a lot of projects, this is definitely something that I at least feel the need to find some way to reassure my colleagues/supervisors about, given the significant financial investment that many researchers put into collecting their data and the constant worry in academia about being "scooped".
Curious if any of the community has general thoughts about this.
8 October 2021 8:36pm
Hi David!
It sounds like Rainforest Connection and its affiliates reserve the right to use your acoustic data and its labels to train and/or test sound identification algorithms.
In my opinion, Rainforest Connection is unlikely to "scoop" the direct ecological or conservation research outputs of your data, which could be what your collaborators would be most concerned about.
However, if one of your research goals is methods development (e.g. publishing trained sound classifiers or experimenting with different machine learning approaches), you may want to directly ask Rainforest Connection for clarification on this part of their terms of service.
Sidenote: one the most significant bottlenecks for developing accurate machine learning classifiers is a lack of labeled data. The bioacoustics community would certainly benefit from the open publication of more labeled data!
Tessa
Tessa Rhinehart
University of Pittsburgh