I use Captum's Integrated Gradient to interprete my PyTorch's neural network. I know that from github and original paper mentioned that ...
Positive attribution score means that the input in that particular position positively contributed to the final prediction and negative means the opposite. The magnitude of the attribution score signifies the strength of the contribution. Zero attribution score means no contribution from that particular feature.
But what exactly positively/negatively contribution mean in human language ? I have to explain these to my co-worker.
Does negative mean that value of feature have less informative to prediction of that class ? When I average score of all values as graph below. What exactly the most negative(or most positive) score mean in human language ?