In the randomization test, we need to randomize the model and look at the correlation between randomized and unrandomized explanations. We have a correlation value for each test sample. We take the average, but do we randomize independently for each sample, or do we randomize once at the beginning? Does it matter? The source paper is not clear.
In the randomization test, we need to randomize the model and look at the correlation between randomized and unrandomized explanations. We have a correlation value for each test sample. We take the average, but do we randomize independently for each sample, or do we randomize once at the beginning? Does it matter? The source paper is not clear.