Optimized FPGA-based Deep Learning Accelerator for Sparse CNN using High Bandwidth Memory | IEEE Conference Publication | IEEE Xplore