AutoMix: Unveiling the Power of Mixup for Stronger Classifiers

Open paper
Benchmark Model Rank Results
image-classification-on-cifar-10ResNeXt-50 (AutoMix)#68Percentage correct: 97.84
image-classification-on-cifar-100WRN-28-8 +AutoMix#64Percentage correct: 85.16
image-classification-on-cifar-100ResNeXt-50(32x4d) + AutoMix#81Percentage correct: 83.64
image-classification-on-imagenetResNet-101 (AutoMix)#644Top 1 Accuracy: 80.98%Number of params: 44.6M
image-classification-on-imagenetResNet-50 (AutoMix)#725Top 1 Accuracy: 79.25%Number of params: 25.6M
image-classification-on-imagenetResNet-34 (AutoMix)#866Top 1 Accuracy: 76.1%Number of params: 21.8M
image-classification-on-imagenetResNet-18 (AutoMix)#939Top 1 Accuracy: 72.05%Number of params: 11.7M
image-classification-on-inaturalist-2018ResNeXt-101 (AutoMix)#34Top-1 Accuracy: 70.49%
image-classification-on-inaturalist-2018ResNet-50 (AutoMix)#42Top-1 Accuracy: 64.73%
image-classification-on-places205AutoMix (ResNet-50 Supervised)#8Top 1 Accuracy: 64.1
image-classification-on-tiny-imagenet-1ResNeXt-50 (AutoMix)#13Validation Acc: 70.72%
image-classification-on-tiny-imagenet-1ResNet18 (AutoMix)#17Validation Acc: 67.33%

Rank counts only results with a code link.