Integrated Gradients
A gradient-based attribution method that computes each input feature's contribution by integrating gradients along the path from a baseline to the input. It satisfies axioms such as sensitivity and implementation invariance.
Timeline (last 90 days)
Gradient-based saliency maps (such as Integrated Gradients) can provide local explanations, but attention must be paid to the gradient saturation problem.
Common variants of gradient attribution include Vanilla Gradient, Integrated Gradients, Grad-CAM, and SmoothGrad, among others.
Integrated gradients is an attribution method that assigns importance by integrating gradients along the path from a baseline to the input.
The nine evaluated XAI methods cover gradient-based methods (Vanilla Gradient, Integrated Gradients), perturbation-based methods (LIME, KernelSHAP), and propagation-rule-based methods (LRP).
All Facts (4)
Gradient-based saliency maps (such as Integrated Gradients) can provide local explanations, but attention must be paid to the gradient saturation problem.
50%UnverifiedCommon variants of gradient attribution include Vanilla Gradient, Integrated Gradients, Grad-CAM, and SmoothGrad, among others.
50%UnverifiedIntegrated gradients is an attribution method that assigns importance by integrating gradients along the path from a baseline to the input.
50%UnverifiedThe nine evaluated XAI methods cover gradient-based methods (Vanilla Gradient, Integrated Gradients), perturbation-based methods (LIME, KernelSHAP), and propagation-rule-based methods (LRP).
50%