Squeezing Backbone Feature Distributions to the Max for Efficient Few-Shot Learning

Benchmark Model Rank Results
few-shot-image-classification-on-mini-imagenet-cub-5-way-1-shotPEMnE-BMS*#2Accuracy: 63.90
few-shot-image-classification-on-mini-imagenet-cub-5-way-5-shotPEMnE-BMS#2Accuracy: 79.15
few-shot-image-classification-on-cifar-fs-5-way-1-shotPEMnE-BMS*#3Accuracy: 88.44
few-shot-image-classification-on-tiered-imagenet-5-way-1-shotPEMnE-BMS*#3Accuracy: 86.07
few-shot-image-classification-on-cub-200-5-way-5-shotPEMnE-BMS*#4Accuracy: 96.43
few-shot-image-classification-on-tiered-imagenet-5-way-5-shotPEMnE-BMS*#4Accuracy: 91.09
few-shot-image-classification-on-cifar-fs-5-way-5-shotPEMnE-BMS*#5Accuracy: 91.86
few-shot-image-classification-on-cub-200-5-way-1-shotPEMnE-BMS*#5Accuracy: 94.78
few-shot-image-classification-on-mini-imagenet-5-way-5-shotPEMnE-BMS*(transductive)#5Accuracy: 91.53
few-shot-image-classification-on-mini-imagenet-5-way-5-shotPEMbE-NCM (inductive)#26Accuracy: 84.67
few-shot-image-classification-on-mini-imagenet-5-way-1-shotPEMnE-BMS* (transductive)#7Accuracy: 85.54
few-shot-image-classification-on-mini-imagenet-5-way-1-shotPEMbE-NCM (inductive)#37Accuracy: 68.43