I mainly tackle regression problems by CNNs, and want to find a reliable method to calculate the heatmaps for NN's results. However, I find almost all interpretation methods including CAM is used for classification NNs but not for regression NNs.

Is there any interpretation method suitable for CNNs which do regression tasks?


1 Answer 1


Yes you are correct that GradCam does not support regression in its formulation, but most gradient-based saliency map methods do support regression, for example:

  1. Guided Backpropagation
  2. Integrated Gradients
  3. SmoothGrad

Also in non-gradient methods, LIME supports regression as well.

  • $\begingroup$ Thanks for you answering! Whether these methods are also useful in relatively simple Fully connected neural network or is only used for CNNs. And how can i choose a suitable one among them? Are there a document about limitations or comparison about these methods. $\endgroup$
    – minghuisvn
    Dec 24, 2022 at 1:47

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .