Cynthia Diane Rudin

Cynthia Diane Rudin is a Professor of Computer Science, Electrical and Computer Engineering, Statistical Science, and Biostatistics & Bioinformatics at Duke University, where she directs the Interpretable Machine Learning Lab. Previously, she held faculty positions at MIT Sloan School of Management and research roles at Columbia University and NYU.

• PhD in Applied and Computational Mathematics, Princeton University
• BS in Mathematical Physics and Music Theory, University of Chicago

Her research pioneers interpretable machine learning for high-stakes domains where transparency is non-negotiable. She challenges the accuracy-interpretability tradeoff myth, proving that transparent models can match black-box performance in healthcare, criminal justice, and power grid reliability applications. Her work has produced deployable systems like the 2HELPS2B ICU seizure predictor and NYPD's Patternizr crime detection tool.

Rudin's publication record shows consistent focus on interpretable algorithms since 2007, with recent work exploring the Rashomon set of equally accurate models and causal inference frameworks. Her highly cited 2018 Nature Machine Intelligence paper fundamentally shifted industry practices toward transparent AI.

• Squirrel AI Award for AI Benefiting Humanity (2022)
• Guggenheim Fellowship (2022)
• Triple INFORMS Innovative Applications Award winner (2013, 2016, 2019)
• Fellow of AAAI, ASA, and IMS

She actively mentors students through Duke's Data+ program and has advised teams winning international competitions. Her lab maintains strong industry partnerships focused on ethical AI deployment, with current grants from NSF, NIH, and DARPA supporting healthcare and criminal justice applications. Rudin serves on National Academies committees shaping federal AI policy and has testified before Congress on algorithmic transparency.

The Interpretable Machine Learning Lab maintains an open-science ethos with all code publicly available. Current projects focus on sparse decision trees for medical diagnostics, causal inference frameworks for high-stakes decisions, and extending the Rashomon effect theory to new application domains.