Edge AI meeting

[mattjbr123]

Here are my unedited handwritten notes, apologies I didn't have time to write them up neatly. Hope they are somewhat useful, and others can add their notes too
edge_ai.pdf

[metazool]

Thank you! "Adaptive Monitoring" as a keyphrase caught my eye.

https://www.webpages.uidaho.edu/~beth/REM357_readings_files/adaptive%20monitoring.pdf - older paper on Adaptive Monitoring, no visible date but the references are all 1990s/2000s, not a bad thing! Very relevant to the topic of "are the research questions we are asking with our long-term monitoring projects still useful questions" within the NC-UK Integrated Environmental Monitoring theme... #28

For me this dovetails neatly with the question of assurance for bioacoustic sensors - practical example being the next-gen AMI trap with speculative BirdPi on it - how do you verify that you're only recording and analysing certain frequencies, or prove on request that you haven't recorded anything heard before a trigger event? One would hope the phone-based voice assistant people struggle with these questions all the time :D