In today's rapidly evolving artificial intelligence landscape, engineers and risk managers play pivotal roles in ensuring innovation and accountability. Our course on explainable AI (XAI) equips professionals with essential skills to navigate this rapidly expanding complex terrain. Students will gain hands-on experience exploring compliance considerations derived from real-world applications across diverse sectors. By merging technical expertise with compliance insights, this course empowers individuals to drive impactful, transparent AI solutions in their respective domains, fostering innovation and integrity.

This course explores XAI, covering essential background definitions and concepts, explainable feature engineering, the diverse ecosystem of XAI models, post-hoc explanation methods, and the latest developments in audit and transparency laws and regulations.

Some featured topics in this course on explainable AI (XAI):

• The fANOVA framework

• From penalized GLM to EBM and GAMinet

• Monotonic GBMs

• Surrogate model approaches

• Explainable feature engineering

• LOCO, LOFO, and pertubation

• SHAP: The good, bad, and ugly

• Security and bias considerations

Background information and definitions are sourced from the National Institute for Standards and Technology (NIST), the National Academies, and other authoritative sources. The instructor teaches technical subjects with Python examples and compliance lessons based on his real-world experience in consumer finance, employment, and other regulated applications of ML. (Note: The instructor is not an attorney.)

This course is designed primarily for data scientists and ML engineers. Technical risk executives or risk managers may find this course helpful in providing an updated overview of newer ML approaches suited for high-stakes applications. This course may interest others too. Materials may help regulators or policy professionals gain insights into the current state of ML technologies that can be used to comply with laws, regulations, or standards. If you're coming to ML from physics, econometrics, or psychometrics, this course can help you learn how to blend newer ML techniques with established domain expertise and notions of validity or causality.