James Donlon
$382,000
Salil Desai
Robert H Newman
Kristen L Rhinehardt
Madhuri Siddula
North Carolina Agricultural & Technical State University
North Carolina
Computer and Information Science and Engineering (CISE)
This project is an ExpandAI Capacity building pilot (CAP), which focuses on establishment of a robust Artificial Intelligence (AI) infrastructure at North Carolina Agricultural and Technical State University (NCAT) thereby enhancing the research capacity of the institution and facilitating AI-focused educational curriculum development and training. Towards this goal, the project will address the challenges in AI in the development of robust, explainable, fair, and privacy-preserving models for sensitive COVID-19 data. Five research programs are pursued to develop new AI models and tools. The research activities aim to broaden the participation of faculty members at NCAT and especially graduate and undergraduate students from underrepresented groups to enroll and explore degrees in different departments at NCAT, completing AI related thesis or dissertation through ExpandAI team collaboration. The project will also build community and new centers of excellence in AI where such activities were not previously well developed. This includes faculty participation in training and workshops will increase the number of faculty members using AI in research, development of more AI student researchers, and hosting of AI workshops featuring hands-on experiments to teach AI models based on experiments derived from this project’s research.<br/><br/>This project expands the AI capacity at NCAT through interdisciplinary AI research, education, and workforce development. The interdisciplinary collaboration and cross-disciplinary AI research spans five use-inspired research thrusts centered on pandemic response patterned after the lessons learned in COVID-19 detection. Specifically, the project will (1) develop robust AI models based on federated knowledge distillation for COVID-19 detection with great generalization ability on new emergent dataset; (2) apply explainable AI (XAI) techniques based on SHAP and LIME to identify and visualize the important features of COVID-19 images that play a significant role in AI models for COVID-19 detection; (3) use the identified features of COVID-19 images from XAI techniques as an input to the proposed generative multi-modal language model to generate COVID-19 images to (4) address biased and fairness issues of AI models based on fairness regularization techniques for COVID-19 detection; and (5) apply differential privacy in federated learning frameworks to build secure AI models to protect the private information of individuals and local clients’ data. The proposed AI models are targeted for application in a broader range of biomedical image analysis research. The project’s educational capacity building focuses on new curricular materials and course modules, including undergraduate/graduate AI senior design classes to increase students’ understanding and use of AI, private AI, fairness of AI, explainable AI, and general interest in this important, emerging field. Faculty members will conduct and participate in training and workshops will increase the number of faculty members using AI in research, leading to more AI student researchers and increasing integration of AI research and education capacity at NCAT.<br/><br/>The ExpandAI Program supports AI-powered education and workforce development, infrastructure and research at Minority Serving Institutions to strengthen and diversify U.S. research and education pathways and provide historically marginalized communities with new opportunities in STEM careers.<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.