An Energy-Efficient Architecture for Accelerating Inference of Memory-Augmented Neural Networks | IEEE Conference Publication | IEEE Xplore