Methods for Explainable Artificial Intelligence

Gerald Friedland | 18-ERD-021

Executive Summary

We plan to develop simple yet powerful tools to inspect machine learning models and evaluate their alignment with the user's understanding of the data. This project will fundamentally advance the field of artificial intelligence, on which national security missions increasingly rely, by introducing tools for accountability, explainability, and debugging.

Publications and Presentations

Friedland, G. and A. Metere. 2018. "Machine Learning for Science: Occam's Razor, Accountability and Explainability." SIAM-USC UQ SciML Workshop. Los Angeles, CA, June 2018. LLNL-PROP-744327.

Friedland, G., M. Krell, and A. Metere. 2018. "A Practical Approach to Sizing Neural Networks." arXiv . Oct. 4, 2018. LLNL-TR-758456.