Our paper “Integrating Mental Health and Juvenile Justice Outcomes: A Case Study of Model-Agnostic Interpretable Machine Learning” is now LIVE (article in advance) at the INFORMS Journal on Computing.
Here we create the Marginal Contribution Weighting method (MCW), a model-agnostic interpretable Machine Learning approach for explainable AI.
This is an imminently useful method for any organization that engages in systematic AI governance — it enables a governance team to consistently track how models treat their inputs, to understand where risks or overreliance on certain features may appear, and similar.
It was a joy to partner with Erin Espinosa, Monica Chiarini Tremblay, Arturo Castellanos, Rajiv Kohli, and our dedicated reviewers and editors and to bring attention to both a critical need for juveniles and a methodological improvement for interpretable AI!