Genetic Generalized Additive Models
AI automatically builds simpler, more accurate interpretable models using genetic algorithms on datasets like California Housing.
Researchers Kaaustaaub Shankar and Kelly Cohen developed Genetic Generalized Additive Models (GAMs), using the NSGA-II multi-objective genetic algorithm to automatically optimize model structure. The system jointly minimizes prediction error (RMSE) and a Complexity Penalty measuring sparsity, smoothness, and uncertainty. On the California Housing dataset, it discovered GAMs that outperform baseline LinearGAMs in accuracy or match performance with substantially lower complexity, producing simpler models with narrower confidence intervals.
Why It Matters
Automates creation of transparent, high-performing models where interpretability matters—critical for finance, healthcare, and regulated industries.