CHE-5414: Explainable AI for Domain Insights

Description: Interpretability and its importance for machine learning in chemical sciences and engineering. Integration of scientific principles into machine learning including Bayesian inference of physical models from data and physics-informed neural networks. Feature importance analysis for local and global interpretations. Performance metrics for evaluating interpretation algorithms. Open-source libraries for interpretation. Applications of explainable artificial intelligence (AI) for materials design and process engineering. General tradeoffs between model performance and interpretability. Common pitfalls of explainable AI and their solutions.

Pathways: N/A

Course Hours: 3 credits

Corequisites: N/A

Crosslist: N/A

Repeatability: N/A