A Feature Structure Based Interpretability Evaluation Approach for Deep Learning | IEEE Conference Publication | IEEE Xplore