AlignMixup (CVPR 2022)

This repo consists of the official Pytorch code for our CVPR 2022 paper AlignMixup: Improving Representations By Interpolating Aligned Features (http://arxiv.org/abs/2103.15375)

Requirements

This code has been tested with
python 3.8.11
torch 1.10.1
torchvision 0.11.2
numpy==1.21.0

Additional package versions

cuda 11.3.1
cudnn 8.2.0.53-11.3
tar==1.34
py-virtualenv==16.7.6

Dataset Preparation

1. For CIFAR-10/100, the dataset will automatically be downloaded, if there does not exist any CIFAR-10/100 directory in the path specified while executing the code.
2. For Tiny-Imagenet-200, you can download the dataset from here. Unzip it and specify its path in the code.

Alternatively, you can run the following command in your terminal if you have wget installed to download it to your current directory:

wget http://cs231n.stanford.edu/tiny-imagenet-200.zip

How to run experiments for CIFAR-10

AlignMixup PreActResnet18

cd cifar10_100  

python main.py --dataset cifar10 --data_dir path_to_cifar10_directory \
		--save_dir path_to_save_checkpoints --network resnet --epochs 2000 \
		--alpha 2.0 --num_classes 10 --manualSeed 8492 

AlignMixup WRN 16x8

cd cifar10_100  

python main.py --dataset cifar10 --data_dir path_to_cifar10_directory \
		--save_dir path_to_save_checkpoints --network wideresnet --epochs 2000 \
		--alpha 2.0 --num_classes 10 --manualSeed 8492

How to run experiments for CIFAR-100

AlignMixup PreActResnet18

cd cifar10_100  

python main.py --dataset cifar100 --data_dir path_to_cifar100_directory \
		--save_dir path_to_save_checkpoints --network resnet --epochs 2000 \
		--alpha 2.0 --num_classes 100 --manualSeed 8492 

AlignMixup WRN 16x8

cd cifar10_100  

python main.py --dataset cifar100 --data_dir path_to_cifar100_directory \
		--save_dir path_to_save_checkpoints --network wideresnet --epochs 2000 \
		--alpha 2.0 --num_classes 100 --manualSeed 8492 

How to run experiments for Tiny-Imagenet-200

AlignMixup PreActResnet18

cd tiny_imgnet  

python main.py  --train_dir path_to_train_directory \
		--val_dir path_to_val_directory \
		--save_dir path_to_save_checkpoints --epochs 1200 \
		--alpha 2.0 --num_classes 200 --manualSeed 8492

How to run experiments for Imagenet

To run on a subset of training set (i.e approx 20% images per class)

cd imagenet  

python main.py --data_dir path_to_imagenet_directory --save_dir path_to_save_checkpoints \
		--mini_imagenet True --subset 260 --num_classes 1000 --epochs 300 --alpha 2.0 --batch_size 1024

To run on a full imagenet

cd imagenet  

python main.py --data_dir path_to_imagenet_directory --save_dir path_to_save_checkpoints \
		--mini_imagenet False --num_classes 1000 --epochs 300 --alpha 2.0 --batch_size 1024

TODO

Imagenet using Distributed data parallel (multiple nodes) - coming soon

Results

Dataset	Network	AlignMixup
CIFAR-10	Resnet-18	97.05%	log
CIFAR-10	WRN 16x8	96.91%	log
CIFAR-100	Resnet-18	81.71%	log
CIFAR-100	WRN 16x8	81.23%	log
Tiny-Imagenet	Resnet-18	66.87%	log
Imagenet	Resnet-50	79.32%	log

Acknowledgement

The code for Sinkhorn-Knopp algorithm is adapted and modified based on this amazing repository by Daniel Daza

Citation

If you find this work useful and use it on your own research, please cite our paper

@inproceedings{venkataramanan2021alignmix,
  title={AlignMixup: Improving Representations By Interpolating Aligned Features},
  author={Venkataramanan, Shashanka and Kijak, Ewa and Amsaleg, Laurent and Avrithis, Yannis},
  booktitle={IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year={2022}
}