What are the approaches to Explainability?Review LaterGloballyLocallyManuallyLogically

The approaches to Explainability in AI and Machine Learning can be broadly categorized into four types:

1. Global Explainability: This approach aims to provide a comprehensive explanation of the model's behavior. It tries to explain the overall logic of the model, how it makes decisions based on different features. Techniques like Feature Importance, Partial Dependence Plots are used in this approach.

2. Local Explainability: This approach focuses on explaining individual predictions. It tries to explain why the model made a specific decision for a particular instance. Techniques like LIME (Local Interpretable Model-agnostic Explanations) and SHAP (SHapley Additive exPlanations) are used in this approach.

3. Manual Explainability: This approach involves manually inspecting the model to understand its behavior. This is often done by looking at the weights and structure of the model. This approach is more applicable to simpler models like linear regression or decision trees, which are inherently interpretable.

4. Logical Explainability: This approach involves using logical reasoning to explain the model's decisions. It often involves creating rules or decision trees that mimic the model's behavior. Techniques like Decision Tree Surrogate or RuleFit are used in this approach.

Each of these approaches has its strengths and weaknesses, and the choice of approach depends on the specific requirements of the task at hand.