In this article, we present a hardware architecture optimized for sparse and dense matrix processing in TensorFlow Lite and compatible with embedded-heterogeneous devices that integrate central ...