Explaining Text Classifiers with Counterfactual Representations
The paper presents a method for explaining text classifiers using counterfactual representations. Counterfactuals are hypothetical events identical to real observations except for one categorical feature. Constructing counterfactuals for texts poses…
Continue reading