Maximum softmax probability acts as a baseline for detecting misclassified and out-of-distribution examples in neural networks.
The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.NE 1years
2016 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks
Maximum softmax probability acts as a baseline for detecting misclassified and out-of-distribution examples in neural networks.