A new approach could help users know whether to trust a models predictions in safety-critical applications like health care and autonomous driving.
A new approach could help users know whether to trust a models predictions in safety-critical applications like health care and autonomous driving.