Undergraduate Research

Research experiences for undergraduates on Jetstream2

A group of Jetstream REU students and Jetstream staff stand outside the Cyberinfrastructure Building at IU Bloomington.

Jetstream is the National Science Foundation’s first production cloud computing system. The goal of Jetstream2 is to increase access to cloud computing resources to researchers in the long-tail-of-science (including social science and humanities research).

Students will work with Indiana University researchers and staff to do mentored research projects about Jetstream2. Potential research topics include efficient utilization of Jetstream2 for data analytics tasks, minimizing energy consumption via dynamic load balancing, best practices with big data analytics using Jetstream2, best practices with compute-intensive humanities projects.

Students will receive a generous stipend and, if needed, accommodations in one of IU’s dorms for the duration of the experience. Most program activities will be conducted on the IU campus (Bloomington, IN); however, we anticipate travel to a conference (travel expenses provided).

Jetstream REU student Tenecious Underwood presenting at the Supercomputing 2019 conference

Finding frogs

REU student Tenecious Underwood

In summer 2019, Tenecious Underwood, an REU participant, used machine learning and high-performance computing to identify automatic recognition of frog calls.

Learn about Underwood’s REU experience

Description of the video:

[MUSIC]

Video: Closeup of a frog with its throat bellowing in and out.

Text appears: The world's frogs are in a crisis.

Video: Tenecious Underwood sits in front of a window.

Words appear: Tenecious Underwood, Livingstone College

Tenecious speaks: Eighty-one percent of all frog species are in a total decline, and you have 18% that are evasive species that are decreasing, and with that it is imperative that national frog surveyors be able to accurately identify where they're at, where they're moving to, and not delegate a lot of manpower towards it.

Video: Two young children stand in front of a waterfall.

Words appear: But human frog counters have their limits.

Video: Tenecious Underwood sits in front of a window.

Tenecious speaks: So, basically you have 24 hours and they train you very minimally

on local frog calls, and that's really all you're supposed to be able to identify. So like I said, if those other 18% of evasive species come into your community and you're actually trying to identify who they are. you might not be able to do it because you haven't been accurately trained.

Video: A computer screen

Words appear: Automating frog call recognition with machine learning can help.

Video: Tenecious Underwood sits in front of a window.

Words appear: Human surveyors accurately predict 80% of frog call species. Underwood's neural networks produced over 97% accuracy.

Tenecious speaks: The human national frog surveyors accurately predict 80% on average accurately—we do 80% of the frog call species that they have—but my CNN image base produced over 97% accuracy with the four frog species. The CNN image, the CNN audio, and RNN audio both produced over 88.9% accuracy.

[Music]

Video: Jet contrails in a blue sky.

Words appear: Underwood was part of the Jetstream Research Experience for Undergraduates at Indiana University.

Video: Green words and numbers are visualized on a black background.

Words appear: Jetstream supplies cloud computing resources to researchers across the country.

Video: Tenecious Underwood sits in front of a window.

Tenecious speaks: Like, we couldn't have accomplished anything without Jetstream as far as using supercomputers. And being able to use that Jetstream platform and all of the GPUs that it has instead of using our own CPUs—it made the process more efficient, more effective, and allowed us to be able to do a lot more.

[Music]

Video: People sitting around a table look at some graphs.

Words appear: Jetstream is designed to bring access to those who have not previously used high-performance computing.

Video: Tenecious Underwood sits in front of a window.

Tenecious speaks: Because I feel like there needs to be more women, there needs to be more cultures, and there definitely needs to be an increase in African Americans in this field.

Words appear: This field is predominantly dominated by middle-aged white men. We must go past that.

Tenecious speaks: Not only do most products that come out from companies are targeted

specifically to us with no idea what we have in mind—which is very, very disturbing—but it also needs to happen because it is really what the American public needs to see in a large field that is predominantly dominated by middle-aged white men, which is the baby boomers. We must go past that.

[Music]

Video: Two people walk down an all-white hallway.

Words appear: Underwood wants to encourage others to join him in STEM.

Video: Tenecious Underwood sits in front of a window.

Tenecious speaks: Don't let them tell you you can't do it. Despite what they say, despite what statistics may tell you, don't ever let them tell you that you can't be the successful person that you've envisioned your entire life. No matter your faults, no matter your failures, use them as the brink to pull apart your success.

Video: The screen turns to white.

Words appear: Science Node, sciencenode.org.