Sharpness-Aware Minimization for Efficiently Improving Generalization

In today's heavily overparameterized models, the value of the training loss provides few guarantees on model generalization ability. Indeed, optimizing only the training loss value, as is commonly done, can easily lead to suboptimal model quality... (read more)

PDF Abstract ICLR 2021 PDF ICLR 2021 Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK USES EXTRA
TRAINING DATA
RESULT BENCHMARK
Fine-Grained Image Classification Birdsnap EffNet-L2 (SAM) Accuracy 90.07% # 1
Image Classification CIFAR-10 PyramidNet (SAM) Percentage correct 98.6 # 12
Percentage error 1.4 # 9
Image Classification CIFAR-10 EffNet-L2 (SAM) Percentage correct 99.70 # 1
Percentage error 0.30 # 1
Image Classification CIFAR-100 EffNet-L2 (SAM) Percentage correct 96.08 # 1
Percentage error 3.92 # 1
Image Classification CIFAR-100 PyramidNet (SAM) Percentage correct 89.7 # 7
Image Classification Fashion-MNIST Shake-Shake (SAM) Accuracy 96.41 # 2
Fine-Grained Image Classification FGVC Aircraft EffNet-L2 (SAM) Top-1 Error Rate 4.82 # 1
Image Classification Flowers-102 EffNet-L2 (SAM) Accuracy 99.65% # 1
Fine-Grained Image Classification Food-101 EffNet-L2 (SAM) Accuracy 96.18 # 1
Image Classification ImageNet ResNet-152 (SAM) Top 1 Accuracy 81.6% # 51
Top 5 Accuracy 95.65% # 34
Image Classification ImageNet EfficientNet-L2-475 (SAM) Top 1 Accuracy 88.61% # 3
Number of params 480M # 4
Fine-Grained Image Classification Oxford-IIIT Pets EffNet-L2 (SAM) Top-1 Error Rate 2.90% # 1
Accuracy 97.10% # 3
Fine-Grained Image Classification Stanford Cars EffNet-L2 (SAM) Accuracy 95.96% # 3
Image Classification SVHN WRN28-10 (SAM) Percentage error 0.99 # 1

Methods used in the Paper


METHOD TYPE
🤖 No Methods Found Help the community by adding them if they're not listed; e.g. Deep Residual Learning for Image Recognition uses ResNet