Remember how I said we’ve been busy working on some new stuff? Part of it is our moving into machine learning (hence Sheri taking classes!). Machine learning is becoming a larger and larger part of biology – really any aspect that has seen a huge increase in data collection – namely, genomics and ecology. There’s so much machine learning going on in genomics that there is a new workshop at Plant and Animal Genome (PAG) this year on using machine learning in CyVerse. There may also be a similar one at the Ecological Society of America Conference in 2020 (we’ll keep you posted on that one!).
Our work in machine learning came out of our CEW&T REU frog call project from last year. Eliza Foran worked with us to build the initial field recorder for frog calls (an extension of our previous Jetstream REU work seen here). Her initial work is covered here. She won best visualization for her presentation of this initial work at the Research Services Expo (see coverage here). She will be doing more work applying this work to bat surveys in Central America this year.
This work then inspired another REU project – including our first work in machine learning. We mentored three IU Jetstream REU students last summer – Eliza Foran, Evan Suggs, and Tenecious Underwood. As the REU program was focused on machine learning, and we worked with these students to build an automated frog call identifier using neural networks. They presented a poster and a conference paper at the annual PEARC meeting in Chicago this year (conference paper and poster are here).
Tenecious was then able to come give a presentation at the SuperComputing conference in Denver this November. He got a lot of engagement during and after his talk!
- IT Communications coverage of IU at Supercomputing, with Tenecius presenting his work
- ScienceNode article on the work
- YouTube interview with the man himself!
If you are interested in machine learning, stay tuned! We have more stuff on the way (and not just in ecology data)!