Empirical Optimal Risk to Quantify Model Trustworthiness for Failure Detection

Open in new window