Self-Challenging Improves Cross-Domain Generalization

Open paper
Benchmark Model Rank Results
domain-generalization-on-office-homeRSC (ResNet18)#38Average Accuracy: 63.12
domain-generalization-on-pacs-2RSC (ResNet-50)#32Average Accuracy: 87.83
domain-generalization-on-pacs-2RSC (Resnet-50, DomainBed)#46Average Accuracy: 85.2
domain-generalization-on-pacs-2RSC (Resnet-18)#47Average Accuracy: 85.15
domain-generalization-on-pacs-2RSC (AlexNet)#76Average Accuracy: 76.05
domain-generalization-on-vlcsRSC (AlexNet)#32Average Accuracy: 75.43

Rank counts only results with a code link.