Visualize: Figure1.
Description: "We focus on explanations based on
saliency scores on the movie dataset."
" we focus on local explanation approaches that identify the most influential parts of the input for a particular prediction. "