Federal Project Grant Award Summary

Purdue University received a $357,921 CAREER (FacultyearlyCareerDevelopment) Project Grant from the National Science Foundation's (NSF) Division of Computing and Communication Foundations (Computer and Information Science and Engineering program, CFDA 47.070) effective July 1, 2026 through June 30, 2031. The award funds research and algorithmic development focused on structure-aware machine learning, with the goal of creating more reliable, data-efficient, and updatable artificial intelligence systems. The research addresses fundamental computational challenges in modern machine learning by investigating how training data structure influences loss function geometry in deep neural networks, with three primary research objectives: characterizing and improving optimization stability through curvature-based diagnostics, designing data summaries and synthetic training sets that reduce computational costs while preserving learning integrity, and developing efficient methods for selectively removing training example influence with minimal model degradation.

Beyond the core research activities, the grant encompasses substantial educational and workforce development components. Deliverables include undergraduate and graduate research training, course-integrated projects, open-source software and educational materials, and hands-on outreach programming for secondary school students and educators focused on data science, algorithms, and responsible artificial intelligence. Research validation will be conducted across multiple machine learning domains including image classification, natural language processing, continual learning, and generative models, with publicly released benchmarks and instructional resources to support broader adoption and advancement of the field.