An Efficient and Low-Power MLP Accelerator Architecture Supporting Structured Pruning, Sparse Activations and Asymmetric Quantization for Edge Computing | IEEE Conference Publication | IEEE Xplore