Explainable Artificial Intelligence with Python

In today's modern landscape, eXplainable Artificial Intelligence (XAI) has risen to paramount importance. This course delves deep into the world of AI transparency, enabling learners to distinguish between "glass box," "white box," and "black box" machine learning models. It categorizes XAI based on scope, agnosticity, data types, and explanation techniques, while emphasizing the delicate balance between accuracy and interpretability. Practical insights are provided through the application of Microsoft's InterpretML package, facilitating the generation of explanations for machine learning models. The course also underscores the necessity of counterfactual and contrastive explanations. It covers the working principles and mathematical underpinnings of leading XAI techniques, including LIME, SHAP, DiCE, LRP, and counterfactual and contrastive explanations. Participants will apply these techniques to elucidate black-box models across diverse datasets, spanning tabular, textual, and image domains, ensuring they are equipped to navigate and enhance AI interpretability in the modern era.