Model Interpretability and Explainability

88 lessons in AI Fundamentals

• Partial Dependence PlotsFlashcard
• Partial Dependence Mutates the DatasetCode Quiz
• Training a Global SurrogateCode Quiz
• Limitations of Explanation MethodsSlides / Video
• Stakeholders and Audiences for ExplanationsSlides / Video
• Interpretability Tools and LibrariesSlides / Video
• Evaluating the Quality of ExplanationsSlides / Video
• Attention as an Explanation and Its CaveatsSlides / Video
• Surrogate Models for Explaining Black BoxesSlides / Video
• Counterfactual ExplanationsSlides / Video
• Saliency Maps for Deep ModelsSlides / Video
• Individual Conditional Expectation PlotsSlides / Video
• Partial Dependence Plots (PDP)Slides / Video
• LIME: Local Model ExplanationsSlides / Video
• SHAP Values and Shapley ExplanationsSlides / Video
• Permutation Feature ImportanceSlides / Video
• Feature Importance MethodsSlides / Video
• The Accuracy-Interpretability TradeoffSlides / Video
• White-Box Models: Linear & TreesSlides / Video
• Model-Agnostic vs Model-Specific MethodsSlides / Video
• Global vs Local ExplanationsSlides / Video
• Intrinsic vs Post-Hoc InterpretabilitySlides / Video
• Why Model Interpretability MattersSlides / Video
• Anchor Rule PrecisionCode Quiz
• Normalizing Attention WeightsCode Quiz
• Choosing an Attribution BaselineCode Quiz
• Faithfulness by Feature RemovalCode Quiz
• Correlation as AttributionCode Quiz
• Counterfactual Search LoopCode Quiz
• Example-Based ExplanationsQuiz
• LIME ExplanationsQuiz
• Feature Importance ConceptQuiz
• Surrogate ModelsQuiz
• Interpreting Decision TreesQuiz
• Right to ExplanationQuiz
• Faithfulness of ExplanationsQuiz
• ICE PlotsQuiz
• Attention as ExplanationQuiz
• Interpretability vs ExplainabilityQuiz
• Counterfactual ExplanationsQuiz
• Saliency MapsQuiz
• Global vs Local ExplanationsQuiz
• Linear Model CoefficientsQuiz
• SHAP and Shapley ValuesQuiz
• What Interpretability MeansQuiz
• Model-Agnostic vs Model-SpecificQuiz
• Interpretable vs Black-Box ModelsQuiz
• Permutation Feature ImportanceQuiz
• Accuracy-Interpretability Trade-offQuiz
• Stakeholder-Appropriate ExplanationsQuiz
• Partial Dependence PlotsQuiz
• ICE Curves Collapsed Into a PDPCode Quiz
• Integrated Gradients Missing Input DifferenceCode Quiz
• Grad-CAM Averages Over Wrong AxisCode Quiz
• LIME Surrogate Ignores Proximity WeightsCode Quiz
• Perturbation Explanation Predicts Wrong InputCode Quiz
• Ranking SHAP Contributions by ImpactCode Quiz
• Saliency Map Keeps Signed GradientsCode Quiz
• Most Important Linear CoefficientCode Quiz
• Permutation Feature Importance SignCode Quiz
• SHAP Additivity CheckCode Quiz
• Global Importance Over Whole DatasetCode Quiz
• Averaging Shapley Marginal ContributionsCode Quiz
• Reconstructing a Prediction from SHAPCode Quiz
• Classifying Intrinsic vs Post-Hoc MethodsCode Quiz
• Odds Ratio from Logistic CoefficientCode Quiz
• Linear Model CoefficientsFlashcard
• Evaluating Explanation QualityFlashcard
• Chain-of-Thought as ExplanationFlashcard
• Grad-CAM for CNNsFlashcard
• Faithfulness vs PlausibilityFlashcard
• Attention as ExplanationFlashcard
• Pitfalls of Explanation MethodsFlashcard
• Counterfactual ExplanationsFlashcard
• ICE PlotsFlashcard
• Saliency MapsFlashcard
• Surrogate ModelsFlashcard
• LIMEFlashcard
• Interpretability vs ExplainabilityFlashcard
• White-Box ModelsFlashcard
• Permutation Feature ImportanceFlashcard
• Model-Specific vs Model-AgnosticFlashcard
• Global vs Local ExplanationsFlashcard
• Why Interpretability MattersFlashcard
• Accuracy-Interpretability TradeoffFlashcard
• SHAPFlashcard
• Intrinsic vs Post-hocFlashcard
• Feature ImportanceFlashcard