Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead - Cynthia Rudin

Explore the critical importance of using interpretable machine learning models for high-stakes decisions in this thought-provoking lecture by Cynthia Rudin from Duke University. Delve into the potential societal consequences of relying on black box models and their unreliable explanations. Discover the advantages of interpretable models, which provide faithful explanations of their computations. Examine real-world examples in seizure prediction for ICU patients and digital mammography. Learn about different types of machine learning problems, optimization techniques, and the balance between accuracy and interpretability. Gain insights into case-based reasoning, prototype layers, and the application of interpretable AI in medicine. Engage with interactive tools and demos that showcase the power of interpretable machine learning in addressing critical healthcare challenges.

Introduction
Bad decisions
Definitions
Why
Article
Crossvalidation
Accuracy interpretability
Two types of machine learning problems
Critically ill patients
Two helps to be
Optimization problem
Saliency maps
Casebased reasoning
My network
Prototype layer
Redbellied woodpecker
Wilsons warbler
Accuracy vs interpretability
Computeraided mammography
Interpretable AI
Case Study
Results
Two Layer Additive Risk Model
Submission to Special Issue
Paper
Problems
Most powerful argument
Papers
Interactive tool
Demo
Optimization
Machine Learning in Medicine