Out-of-distribution Detection on CIFAR10 (0-4) ID vs {CIFAR10 (5-9), LSUN, SVHN, Fashion-MNIST, MNIST} (test) benchmark leaderboard · SOTA2 Research

Out-of-distribution Detection on CIFAR10 (0-4) ID vs {CIFAR10 (5-9), LSUN, SVHN, Fashion-MNIST, MNIST} (test)

0.9544AUROCEntropy Oracle

Evaluation Results

Method	Links
Entropy Oracle2022.01		0.9544	0.6851	0.9957	0.1727
One-vs-All Oracle2022.01		0.9138	0.5375	0.9915	0.3594
UQGAN (Ours)2022.01		0.8964	0.5315	0.9901	0.4354
UQGAN (Ours)2022.01		0.8649	0.4908	0.9872	0.4578
GEN2022.01		0.8601	0.4232	0.9866	0.4539
MC-Dropout2022.01		0.7756	0.3875	0.9685	0.8235
Deep-Ensembles2022.01		0.7424	0.3281	0.9643	0.8507
Bayes-by-Backprop2022.01		0.7423	0.2991	0.9648	0.8397
Confident Classifier2022.01		0.7333	0.3232	0.9629	0.8504
Entropy2022.01		0.7285	0.3043	0.9621	0.8541
One-vs-All Baseline2022.01		0.7252	0.3224	0.9601	0.8874
Max. Softmax2022.01		0.7252	0.3052	0.961	0.8768