I've always made the explain based on the top label found by LIME, but i found that sometimes it is different by the one predicted by my model. In this case, what should i do? Should i let LIME doing prediction for the label he found or should i do it for the label actually predicted by my model? I tried the second option but obviouslly it is counterintuitive, indeed it generates red super pixels in the hot spot of my image even if the label predicted by the model was the right one.
I've always made the explain based on the top label found by LIME, but i found that sometimes it is different by the one predicted by my model. In this case, what should i do? Should i let LIME doing prediction for the label he found or should i do it for the label actually predicted by my model? I tried the second option but obviouslly it is counterintuitive, indeed it generates red super pixels in the hot spot of my image even if the label predicted by the model was the right one.
Thank you