Right for the Right Reasons: Training Differentiable Models by Constraining their Explanations.

Andrew Slavin Ross, Michael C. Hughes, Finale Doshi-Velez

Browse the full IJCAI paper archive.