A Novel Visual Interpretability for Deep Neural Networks by Optimizing Activation Maps with Perturbation.

Qing-Long Zhang, Lu Rao, Yubin Yang

Browse the full AAAI paper archive.