Understanding AI - eXplainable AI (XAI) techniques in practice
The IFoA eXplainable AI (XAI) Working Party would like to show you how AI and machine learning models are not in-fact so called “black boxes” and can be used with confidence in actuarial work. We will dispel the myth that Neural Networks, Random Forests etc. cannot be explained, by demonstrating various XAI techniques including ICE, PDP, M-plots, ALE and SHAP with examples in R, building on our article published in The Actuary