DLA: Compiler and FPGA Overlay for Neural Network Inference Acceleration | IEEE Conference Publication | IEEE Xplore