SHREVEPORT – The artificial intelligence and machine learning lab at LSUS is teaching students to work on some of the world’s most complex problems.

Associate professor Dr. Subhajit Chakrabarty is leading a cadre of students performing research in various fields from bioinformatics and environmental pollution to deepfake detection and language translation.

The latest example of research achievement came in May when eight students presented five different papers at the Intermountain Engineering, Techology and Computing Conference hosted by Utah Valley University.

LSUS graduate and undergraduate computer science students employed artificial intelligence and machine learning principles to tackle questions involving brain damage identified by speech patterns, predicting particulate matter pollution in the atmosphere, highlighting underserved areas with access to hospitals that accept Medicare, using lifestyle factors to predict chronic heart disease, and accurate language translation using optical character recognition.

“Students are using what they learned in machine learning and deep learning classes, technically and in the classroom, and extending that into real-life applications,” said Chakrabarty, who came to LSUS in 2020 and now directs the Master of Science in Computer Systems Technology program. “These are master’s level and sometimes undergraduate students who are learning how to write intensive research papers and compete with other researchers to be accepted into regional, national and international conferences.

“When it comes to applying for a job or a position, what sets them apart is that they’ve been published. They are becoming independent researchers who really are conducting research that’s competitive against doctoral and postdoctoral researchers.”

Graduate student Udaysinh Rathod co-authored papers using raw satellite data to predict the presence of fine inhalable particulate matter and how far individuals across the country travel to receive care at a hospital that accepts Medicare.

Rathod said the student-faculty research dynamic fosters an exciting environment in which students collaborate on research in a variety of fields.

“The best part about it is, as students, we have enough free time to do the rigorous work,” Rathod said. “But we might not always know which direction to work in – it’s (Chakrabarty) that steers us and alerts us to certain obstacles we’ll come across in our topic.

“We can have an open research discussion where we present and critique ideas, and we identify those shortcomings together.”

Rathod recalls a four-hour discussion between himself, Chakrabarty and undergraduate Devesh Sarda in which the trio extensively brainstormed ideas, one of which led to a research topic they hope will publish in the coming months.

“Before coming to LSUS, I wasn’t really considering a PhD at all, but now I’m pretty sure that I am going to pursue a PhD,” Rathod said. “We’re learning how to do every aspect of research from a proper literature review to making sure that your data is solid.

“We’re learning through (Chakrabarty’s guidance) and also collaboration with each other.”

Other student researchers that presented at Utah Valley include Mridula Mavuri, Sharmin Nahar, Sweta Singh, Md Tareq Mahmud, Visar Rraci and Landrum Anderson.

Participating faculty members include Drs. Xi Jin, Peter Siska, Qingsong Zhao and Zhonghui Wang.

Students aren’t just generating their own ideas, they are helping solve problems for local and regional entities as well.

LSU Health Shreveport is the lab’s primary partner as students take incredibly large medical datasets and use AI to analyze them.

One of many research projects this partnership produced analyzed MRI images of lumbar spine stenosis patients. Based on spinal discs that can be seen on an MRI, this model generates an image of what the subsequent spinal disc that can’t easily be seen on an MRI would look like.

Chakrabarty and Sarda published this research this past December in the Institute of Electrical and Electronics Engineers International Conference on Bioinformatics and Biomedicine.

It’s one of two LSUS research papers accepted into the IEEEBIBM conference, which Chakrabarty says is one of the “top two or three conferences” in the field.

The other involves the creation of a location health access metric that calculates road distance to the nearest hospital that offers cancer treatment for people in every Louisiana census tract.

Rathrod and fellow student Sweta Singh conducted this research with Chakrabarty and two Alcorn State University personnel. That metric evolved in a subsequent study presented at Utah Valley that covered the entire United States.

Chakrabarty explains the impact of AI and machine learning on all types of research through the most recent project on predicting particulate matter levels in the atmosphere.

“Currently the way the data is done, only aerosols are being predicted, and they exist only at a particular depth in the atmosphere,” Chakrabarty said. “We’ve proven that that method of prediction isn’t good enough.

“Therefore we proposed a method of predicting by looking at the reflectants in the top of the atmosphere using raw satellite data. Before machine learning and deep learning, there wasn’t enough manpower to do these kinds of calculations or measurements. But now we have the tools to work with enormous raw datasets like these.”

Chakrabarty added that only two known research papers explored this topic using images from NASA’s MODIS satellites, putting LSUS research on the cutting edge.

The quality of research and significance of research topics allow LSUS students a multitude of options in industry and academia.

“People with this type of research experience can get higher-paying jobs in industry,” Chakrabarty said. “Anybody that’s done research, it’s such a structured process.

“For people that have gone through this process, organizations will quickly snatch them up for a future in higher management.”

The 10-student lab is mostly funded through grants (seven students are fully or partially funded this way with three students who volunteer).

But the lab’s burgeoning reputation is attracting more attention from organizations with grants to award and entities who need this type of data analysis.

Chakrabarty hopes to increase the lab’s size, and hopefully one day, to add a computer science doctoral program where students can pursue their research interests over an extended period of time.