Logic Programming
Machine Learning
A performance measurement for classification models that outputs probabilistic values between 0 and 1.
Detailed Explanation
Log Loss, also known as Cross-Entropy Loss, evaluates the accuracy of a classification model by penalizing incorrect probabilistic predictions. It measures how well the predicted probabilities align with true labels, with lower values indicating better performance. Log Loss is particularly useful for models that output probabilities, as it considers the confidence level of each prediction, promoting more reliable probabilistic models.
Use Cases
- •Compare the predicted probabilities of a spam classifier against actual labels to optimize model calibration.
Related Terms
Other terms in the Machine Learning category