Now in his sixth and remaining yr of the Ph.D. program with Illinois Computer Science, Adam Stewart knew he was not the traditional pupil coming into this stage in his tutorial profession.
Relatively than having a background in CS, Stewart beforehand studied Earth Science at Cornell College. His main motivation for analysis right here throughout his Ph.D. facilities on the influence synthetic intelligence (AI) can have on environmental points.
“We’ve seen major advancements in many areas of computing, but I don’t want to stop at technological advancements,” Stewart mentioned. “I need to see what sort of influence they’ll make in different domains, which implies I’m centered on a lot of various questions. Can we use strategies from laptop imaginative and prescient and deep studying to sort out local weather change and air air pollution? Can we enhance digital agriculture and precision farming utilizing satellite tv for pc imagery? Can we predict earthquakes, tsunamis, and volcanic eruptions earlier than they occur and save numerous lives?
“These are very different applications involving prediction and forecasting. It takes a ton of data, very complex data, which can be challenging to work with.”
The years he has spent researching objects in accordance with these overarching targets produced a paper titled, “TorchGeo: Deep Learning With Geospatial Data.”
Stewart paired along with his Ph.D. advisor – Illinois CS professor, Arindam Banerjee – and 4 different investigators from the AI for Good Analysis Lab at Microsoft and the College of Texas at San Antonio. Collectively, they addressed how the variance in information assortment strategies and dealing with of geospatial metadata make the applying of deep studying methodology to remotely sensed information nontrivial.
The group developed TorchGeo – a Python library for integrating geospatial information into the PyTorch deep studying ecosystem – to clear up this downside, and their paper outlined the design, implementation, and use of TorchGeo.
The influence this may have on the intense issues of Earth Science that encourage Stewart had been significant sufficient for the group to earn a Greatest Paper Runner Up Award on the thirtieth ACM SIGSPATIAL Convention in November.
“The inspiration for TorchGeo was based on a notion I’ve gathered given my experience working with this data and machine learning. I have found that it is extremely difficult to work in this field unless you have a Ph.D. in both remote sensing and computer science,” Stewart mentioned. “There are only a few researchers who’ve experience in each of those very totally different fields. So, the aim was, can we make this less complicated?
“We know that small datasets are not going anywhere. They’re not going to get bigger; we can’t just throw money at the problem. Instead, we need to be able to train models on small data sets. Accordingly, we’ve focused on model pretraining and transfer learning.”
To get to this level – each along with his paper, and larger image, with the arrogance to tackle such a matter – Stewart mentioned that his advisor performed a key function.
That is primarily as a result of Banerjee’s analysis pursuits coincide so properly along with his personal.
With out a pure CS background, Stewart mentioned there have been moments he succumbed to imposter syndrome. However discovering a dwelling for his distinctive analysis pursuits alongside Banerjee proved to be a second that altered his path.
“It’s been great to work with someone who is passionate about the same things,” Stewart mentioned. “Arindam has done a lot to champion my research, tell people about it, brag about it, and help connect me with interdisciplinary collaborators.”
His advisor, nonetheless, was fast to credit score the coed, whom he lauded for his efforts within the Ph.D. program and with this paper.
“The transition Adam has undertaken is quite remarkable. He has gone far beyond just transitioning to CS; he is among the best in software engineering and building large-scale, production-quality systems. That, combined with his expertise in AI and Geosciences, make him an asset to any team interested in research and system building at this intersection,” Banerjee mentioned. “A few of us within the AI/ML neighborhood have been making a case for extra deal with scientific and societal functions for over a decade.
“Adam gets full credit for the work he is doing, as I would have hesitated to pursue this direction because of how ambitious the TorchGeo project is. Credit goes also to his Microsoft collaborators and mentors, especially Caleb Robinson, for helping Adam bring this amazing project to fruition.”
The outcome proved each rewarding to Stewart whereas concurrently serving as inspiration for what comes subsequent in his tutorial efforts.
“Winning a runner-up award like this, and even just getting this publication out, was very validating for me,” Stewart mentioned. “I think it really goes to show that, yes, this field is moving in a direction where people really do value the application of cutting-edge technology to real world phenomena that we desperately need to solve.”