PEARC24 ACCESS Support Hackathon Winner: Maryam Berijanian
ICER is excited to announce that Maryam Berijanian, Michigan State University doctoral student in Computational Math, Science, and Engineering and ICER user, was among the PEARC24 ACCESS Support Intro to AI and Mini-Hack participants and won Best Originality/Creativity in the advanced category as well as a $3,000 travel award.
“Participating in the PEARC24 ACCESS Support AI Hackathon was both challenging and rewarding. I'm thrilled to have received the 'Best Originality/Creativity' award, recognizing my efforts to exceed task requirements and put my natural language processing and artificial intelligence knowledge to the test,” Berijanian reflected.
PEARC24 is a conference for practice and experience in advanced research computing. The conference provides an opportunity for industry professionals to share challenges, opportunities, and solutions in the research computing industry. This year’s conference offered a student program for Intro to AI and a Mini-Hack that provided participants with an opportunity to explore AI techniques and IMDB movie reviews to predict sentiment by extracting and evaluating the effectiveness of machine learning algorithms.
These AI techniques included understanding how computer systems learn material, how they can predict various things such as a movie you would like to watch, and understanding how a digital fingerprint, or digital tracking, can make computer learning possible.
“I appreciate that the competition valued not just the best performance results but also creativity, encouraging us to think outside the box,” said Berijanian. “I'm grateful for the opportunity—not only was it fun, but I also learned a lot from many talented participants.”
Berijanian’s research at Michigan State focuses on using large language models to facilitate scientific research, such as evaluation of measures for information extraction tasks. As an ICER user, Berijanian uses ICER resources to help assist them during their research.
“I use ICER's high-performance computing (HPC) resources, especially the GPU clusters, which are essential for running my computationally-intensive experiments—something that wouldn't be feasible on a personal computer,” shared Berijanian.
Berijanian shared that their experience as an ICER user assisted them in using ACCESS during the Hackathon. ACCESS is a similar, national-level HPC resource.
“My familiarity with HPC and guidance from my advisor and Director of User Support at ICER, Dr. Dirk Colbry, helped me to quickly adapt to the platform under competition pressure. Without prior HPC experience, it would have been challenging to navigate the technical aspects during the Hackathon,” shared Berijanian.
ICER cannot wait to see the amazing things Maryam Berijanian accomplishes in the future!
Written by ICER Intern Kylie McClung