Project Grant 2503074

The National Science Foundation (NSF) awarded a Project Grant under the Computer and Information Science and Engineering (CFDA 47.070) program to The Curators of the University of Missouri, doing business as the University of Missouri at Kansas City (UMKC). The $150,000 grant, awarded on May 1, 2025, supports the "Collaborative Research: EAGER: NAIRR Pilot: Demystifying the Black Box: Algorithm-Informed Neural Networks" project. This project aims to develop a new approach called Algorithm-Informed Neural Networks (AINNs) that integrates well-established algorithmic principles into neural network design to enhance the explainability, reliability, and efficiency of AI systems. The research focuses on two key tasks: 1) constructing neural networks by systematically integrating algorithmic logic to reduce training data requirements and improve generalization and interpretability, and 2) developing a novel debugging tool called Latent Behavior Analysis to detect and trace input-specific anomalies in neural subnetworks. The project will evaluate AINNs across diverse tasks and expects to produce AI systems with greater transparency, lower data dependency, and enhanced reliability for real-world applications. The research outputs, including datasets, models, and tools, will be publicly released to promote broader adoption of algorithm-informed AI across multiple domains.