This $300,000 Project Grant awarded by the National Science Foundation's (NSF) Computer and Information Science and Engineering (CISE) program (CFDA 47.070) aims to develop advanced visualization and interpretability tools for understanding how artificial intelligence (AI) models like AlphaFold predict protein structures. The two-year "VIZFOLD: Visualizing & Interpretable Protein Folding in AlphaFold and Similar Protein Models" project will focus on three key objectives:
- Developing visualization tools to examine the internal architectural components and activation propagation within AlphaFold-like models. 2) Enhancing model interpretability through probing techniques, dimensionality reduction, and layerwise relevance propagation. 3) Building an interactive science gateway powered by CyberShuttle to enable real-time visualization and analysis of model outputs.
This work aims to illuminate the "black box" of how these AI models operate, foster a deeper understanding of protein folding processes, and democratize access to these capabilities for a wider range of scientists. The award to the Georgia Tech Research Corporation, doing business as the Office of Sponsored Programs, will run from February 2025 to January 2027.
Generated 3/4/25, 6:23 AM