ICERs Supercomputer and Super Staff Contribute to XPRIZE Competition
Sponsors of the XPRIZE Rainforest competition put $10 million in prize money on the line for whoever can develop novel technologies to rapidly and comprehensively survey rainforest biodiversity and use that data to improve our understanding of complex rainforest ecosystems.
A group of experts from Michigan State University (MSU) were part of the XPRIZE team known as ACT NOW – Amazonas Action Alliance (ANAAA) to make a difference in our world’s rainforests. The hard-working team included field experts, indigenous Amazonian rainforest experts, and logistics, science, and data experts. The task was to survey rainforest biodiversity in a 100-hectare plot of dense jungle, which required more data analysis capacity than a typical computer can handle. So, the scientists turned to MSU’s supercomputer for help.
MSU’s Institute for Cyber-Enabled Research (ICER) provided the supercomputer resources and support staff for machine learning and artificial intelligence, data transfer and storage, and data analysis and integration. Read more about XPRIZE and MSU’s team at the following link from the College of Natural Science: Spartans join indigenous forest protectors on XPRIZE Rainforest biodiversity quest. Read on to learn about ICER’s unique contributionhttp:// to this massive effort.
While some XPRIZE contributors had years to prepare for the competition, ICER joined just months before. ICER staff rose to the challenge to provide computational resources and human expertise to the team. Pat Bills, ICER software engineer and data scientist, admitted there were moments of doubt.
“There were several moments when we wondered if we could pull it all together in time, as data was coming in late, with issues formatting things, but in the end, the ICER teams came through and submitted a report on time,” said Bills.
Before analyzing the massive amount of data, it had to be transferred from the competition site in Singapore to MSU. This was made possible by the high-speed MSU Research Network, which helps MSU researchers collaborate more efficiently with colleagues locally and globally.
The MSU/ANAAA team used machine learning programs to identify organisms that appeared in thousands of photos and sound recordings taken during the sampling phase of the semifinal competition. ICER provided computing power and disk space for this immense quantity of data.
Two ICER Research Consultants went above and beyond to ensure data analysis was accurate and on time. Dr. Nicholas Panchy, a bioinformatics expert, created a pipeline to analyze environmental DNA despite having no prior experience with that technique. Dr. Claire Kopenhafer designed a pipeline to ingest data and perform artificial intelligence work designed by MSU faculty.
“Since this competition required our team to identify as many species as possible in a short window, we relied heavily on machine learning to identify the animals and plants present in our images, audio recordings, and DNA samples,” recounted Kopenhafer. “ICER's computing resources - particularly those that are part of our MSU Data Machine - allowed us to both train these models and apply them to field data well within the 48-hour analysis time period.”
This part of the analysis was largely orchestrated by ICER’s research consultants, who were responsible for directing data from seven potential data sources through five different machine learning models. The output of these models, as well as analysis from human experts, was then compiled into one report.
“We were able to contribute in this way because we knew ICER's system administrators had our backs - keeping the system up and running while answering our late-night requests,” said Kopenhafer.
The system administrators who are responsible for keeping ICER’s research supercomputer running also created the robust machine for XPRIZE that the MSU/ANAAA team knew they could rely on to handle the necessary data analysis within the short timeframe of the competition.
A significant facet of the XPRIZE Rainforest competition involves the speed of scientific assessment and discovery. Given the size and density of the Earth’s rainforests, a successful biodiversity assessment procedure needs to collect, analyze, and interpret data quickly. Bills recounted one such example of the team’s speedy scientific endeavors in discovering an invasive mollusk from the DNA left behind by a snail track across a leaf. The sample was collected halfway around the world by the team in Singapore less than three days prior.
The mission of the MSU/ANAAA team kept them going through the long hours. They never lost sight of their goal to help indigenous and marginalized populations living in the rainforest. The team had the unique distinction of being the only group collaborating extensively with the local communities in the regions they studied.
“I am in awe of the amazing people from ICER that I worked side-by-side with on this project,” lauded Bills. “Witnessing how ICER stepped up and went above and beyond for MSU to be able to participate in this competition just confirmed that I’m in the right place to help researchers at MSU.”