LP-BNN: Ultra-low-Latency BNN Inference with Layer Parallelism | IEEE Conference Publication | IEEE Xplore