At NeurIPS 2020, researchers proposed faster, more efficient alternatives to backpropagation

Backpropagation has led to impressive gains in AI and machine learning. But it has limitations researchers are attempting to move beyond.