Abstract: In this work, we present an energy-efficient posit processing element (PE) for utilization in array-based deep neural network (DNN) accelerators along with an approximation method for ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results