HUA YUANZHI

Data compaction and memory bandwidth reduction for sparse neural networks