YZ Sözlüğü
Yapay Zekanın tam sözlüğü
Shapley Values
Concept from game theory that quantifies the average marginal contribution of a feature across all possible feature coalitions, serving as the theoretical foundation for fair attribution methods like SHAP.
Permutation Feature Importance
Global importance evaluation method that measures the decrease in model performance when the values of a feature are randomly permuted, thus indicating its dependence on that feature for predictions.
Gradient Saliency Map
Visual attribution method for convolutional neural networks that uses the gradient of the output with respect to the input to highlight the most influential pixels or regions for a given prediction.
Break-down
Local attribution method that decomposes a model's prediction into contributions from each feature by calculating conditional prediction changes when features are sequentially added.