Building Trust in Deep Learning Models via a Self- Interpretable Visual Architecture | IEEE Conference Publication | IEEE Xplore