UA Little Rock Professor Nitin Agarwal’s AI Research Published at World’s Top Artificial Intelligence Conference
Artificial intelligence research from Dr. Nitin Agarwal at the University of Arkansas at Little
Rock has earned international recognition after a collaborative paper was published at the 39th Annual Conference on Neural Information Processing Systems (NeurIPS). The conference is widely considered the world’s premier gathering for artificial intelligence and machine learning research.
The achievement places UA Little Rock alongside many of the world’s leading universities,
technology companies, and research laboratories, highlighting the university’s contributions to
cutting-edge AI research.
The paper was co-authored by Agarwal, director of the Collaboratorium for Social
Media and Online Behavioral Studies (COSMOS) Research Center, Maulden-Entergy Chair, and
Donaghey Distinguished Professor of Information Science at UA Little Rock, along with
collaborators from the University of Arkansas, Fayetteville, and the University of Florida.
COSMOS is an interdisciplinary research center that studies online behavior and develops data
analytics tools to better understand online communities and emerging cognitive threats in online information environments.
“NeurIPS is where many of the most influential advances in AI are first introduced,” said
Agarwal, whose research on cognitive security has been recognized by the U.S. Department of
War, NATO, the Five Eyes intelligence alliance (consisting of Australia, Canada, New Zealand, the United Kingdom, and the United States), the World Health Organization, and other
organizations. “Having a paper accepted means the research has undergone an extremely
rigorous peer-review process and has been recognized by leading experts in the field as making a meaningful contribution to advancing AI.”
As one of the world’s most selective AI conferences, the acceptance process is highly
competitive, drawing thousands of submissions from top universities, research labs, and
technology companies across the globe.
“Acceptance reflects both novelty and technical excellence,” said Agarwal, an Arkansas
Academy of Computing fellow, an Arkansas Research Alliance (ARA) fellow, and IEEE (Institute of Electrical and Electronics Engineers) Distinguished Visitor and Senior Member. “For COSMOS and UA Little Rock, it demonstrates that the research being conducted here is competing successfully on a global stage alongside some of the world’s leading AI institutions. It reinforces that impactful, internationally recognized AI research is being developed right here in Arkansas.”
Agarwal and his team’s paper introduces MANGO: Multimodal Attention-based Normalizing
Flow Approach to Fusion Learning, a new framework designed to improve how AI systems learn from and combine multiple types of information, including images, text, and other data sources.
The research centers on multimodal AI, a type of artificial intelligence that learns from multiple
forms of information — such as text, images, or audio — simultaneously. While people naturally combine different sources of information to understand the world around them, AI systems often struggle to do so effectively because the data doesn’t always match up in obvious ways. MANGO is designed to help AI better understand and combine different types of data. Unlike many existing AI models that simply merge information together, MANGO identifies which sources are most relevant and how they relate to one another, allowing AI systems to make more accurate and reliable decisions.
“The real world is inherently multimodal, so if we want AI to better understand people and
complex environments, it needs to learn from multiple sources simultaneously rather than
treating each independently,” Agarwal said.
Research like this helps AI evolve to provide more context-aware analysis, enabling systems to
make more informed and trustworthy decisions.
As AI becomes increasingly integrated into daily life and industry, MANGO could improve
applications across various fields. In healthcare, it could help physicians make more informed
decisions by analyzing medical images alongside patient records and clinical notes. In self-
driving vehicles, it could improve safety by allowing AI to better interpret data from cameras,
radar, and other sensors. The technology could also strengthen security by analyzing social
media to identify coordinated influence campaigns or emerging cognitive threats through the
combined analysis of text, images, videos, and patterns of online behavior.
“Ultimately, research like this helps build AI systems that are more reliable, more explainable,
and better suited for solving complex real-world problems,” Agarwal said.
The publication also reinforces the COSMOS Research Center’s role as a hub for collaborative,
interdisciplinary AI research. By partnering with researchers at UA Fayetteville and UF, the
project demonstrates how UA Little Rock is contributing to advances at the highest levels of the international AI community.
“This publication further demonstrates that UA Little Rock is producing AI research that is
recognized at the highest international level,” Agarwal said. “It enhances the university’s
visibility within the global AI community, strengthens our research collaborations, and creates
new opportunities for our students to work on cutting-edge problems.”
Looking ahead, Agarwal and his collaborators plan to expand MANGO to work with additional
data types, larger AI foundation models, and real-time decision-making environments while
continuing to improve the transparency and reliability of AI systems. Their goal is to develop AI
systems that not only achieve high performance, but also operate reliably in complex applications where accuracy and trust are essential.
Disclaimer: This research was funded in part by the U.S. National Science Foundation, U.S.
Army Research Office (awards W911NF23S0001, W911NF2410078, and W911NF2510147), U.S. Office of Naval Research, U.S. Air Force Office of Scientific Research, U.S. Air Force Research Laboratory, U.S. Defense Advanced Research Projects Agency, the Australian Department of Defense Strategic Policy Grants Program, Arkansas Research Alliance, the Jerry L. Maulden/Entergy Endowment, and the Donaghey Foundation at the University of Arkansas at Little Rock. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the funding organizations.