Explanations of black-box models based on directional feature interactions

Published in International Conference on Learning Representation (ICLR), 2022