Improving Transferability of Representations via Augmentation-Aware Self-Supervision
Accepted to NeurIPS 2021
TL;DR: Learning augmentation-aware information by predicting the difference between two augmented samples improves the transferability of representations.
Dependencies
conda create -n AugSelf python=3.8 pytorch=1.7.1 torchvision=0.8.2 cudatoolkit=10.1 ignite -c pytorch
conda activate AugSelf
pip install scipy tensorboard kornia==0.4.1 sklearn
Checkpoints
We provide ImageNet100-pretrained models in this Dropbox link.
Pretraining
We here provide SimSiam+AugSelf pretraining scripts. For training the baseline (i.e., no AugSelf), remove --ss-crop and --ss-color options. For using other frameworks like SimCLR, use the --framework option.
STL-10
CUDA_VISIBLE_DEVICES=0 python pretrain.py \
--logdir ./logs/stl10/simsiam/aug_self \
--framework simsiam \
--dataset stl10 \
--datadir DATADIR \
--model resnet18 \
--batch-size 256 \
--max-epochs 200 \
--ss-color 1.0 --ss-crop 1.0
ImageNet100
python pretrain.py \
--logdir ./logs/imagenet100/simsiam/aug_self \
--framework simsiam \
--dataset imagenet100 \
--datadir DATADIR \
--batch-size 256 \
--max-epochs 500 \
--model resnet50 \
--base-lr 0.05 --wd 1e-4 \
--ckpt-freq 50 --eval-freq 50 \
--ss-crop 0.5 --ss-color 0.5 \
--num-workers 16 --distributed
Evaluation
Our main evaluation setups are linear evaluation on fine-grained classification datasets (Table 1) and few-shot benchmarks (Table 2).
linear evaluation
CUDA_VISIBLE_DEVICES=0 python transfer_linear_eval.py \
--pretrain-data imagenet100 \
--ckpt CKPT \
--model resnet50 \
--dataset cifar10 \
--datadir DATADIR \
--metric top1
few-shot
CUDA_VISIBLE_DEVICES=0 python transfer_few_shot.py \
--pretrain-data imagenet100 \
--ckpt CKPT \
--model resnet50 \
--dataset cub200 \
--datadir DATADIR
10 Mar 22, 2022
3 Apr 07, 2022
288 Oct 28, 2022
131 Dec 25, 2022
187 Dec 26, 2022
10 Jul 14, 2022
3.2k Jan 05, 2023
127 Dec 30, 2022
0 Feb 09, 2022
112 Jan 02, 2023
39 Dec 08, 2022
216 Dec 18, 2022
10 Dec 12, 2022
189 Dec 07, 2022
90 Jan 06, 2023
26 Nov 30, 2022
1.8k Jan 01, 2023
2.1k Jan 03, 2023
16 Aug 30, 2022
14 Dec 19, 2022