2020

SuperMix: Supervising the Mixing Data Augmentation

Dabouei, Ali, Soleymani, Sobhan, Taherkhani, Fariborz et al.

Understand

This paper presents a supervised mixing augmentation method termed SuperMix, which exploits the salient regions within input images to construct mixed training samples.

SuperMix is designed to obtain mixed images rich in visual features and complying with realistic image priors.
To enhance the efficiency of the algorithm, we develop a variant of the Newton iterative method, $65\times$ faster than gradient descent on this problem.
We validate the effectiveness of SuperMix through extensive evaluations and ablation studies on two tasks of object classification and knowledge distillation.

Built on

Probability of error of some adaptive pattern-recognition machines
H Scudder · 1965
Earlier work this paper cites.
Unsupervised word sense disambiguation rivaling supervised methods
David Yarowsky · 1995
Earlier work this paper cites.
Statistical learning theory, 1998
Vladimir Vapnik and Vlamimir Vapnik · 1998
Earlier work this paper cites.
Semi-supervised self-training of object detection models
Chuck Rosenberg, Martial Hebert, and Henry Schneiderman · 2005
Earlier work this paper cites.
Model compression
Cristian Bucilu, Rich Caruana, and Alexandru Niculescu-Mizil · 2006
Earlier work this paper cites.
Nonlinear optimization
Andrzej P Ruszczyński and Andrzej Ruszczynski · 2006
Earlier work this paper cites.
Imagenet: A large-scale hierarchical image database
Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei · 2009
Earlier work this paper cites.
Learning multiple layers of features from tiny images
Alex Krizhevsky, Geoffrey Hinton, et al · 2009
Earlier work this paper cites.
Optimol: automatic online picture collection via incremental model learning
Li-Jia Li and Li Fei-Fei · 2010
Earlier work this paper cites.
Neil: Extracting visual knowledge from web data
Xinlei Chen, Abhinav Shrivastava, and Abhinav Gupta · 2013
Earlier work this paper cites.
Convolutional neural networks for no-reference image quality assessment
Le Kang, Peng Ye, Yi Li, and David Doermann · 2014
Earlier work this paper cites.
Dropout: a simple way to prevent neural networks from overfitting
Nitish Srivastava, Geoffrey Hinton, Alex Krizhevsky, Ilya Sutskever, and Ruslan Salakhutdinov · 2014
Earlier work this paper cites.

Similar

Distilling the knowledge in a neural network
Original
Geoffrey Hinton, Oriol Vinyals, and Jeff Dean · 2015
Cited alongside, same era.
Deepfool: a simple and accurate method to fool deep neural networks
Seyed-Mohsen Moosavi-Dezfooli, Alhussein Fawzi, and Pascal Frossard · 2016
Cited alongside, same era.
Understanding deep learning requires rethinking generalization
Original
Chiyuan Zhang, Samy Bengio, Moritz Hardt, Benjamin Recht, and Oriol Vinyals · 2016
Cited alongside, same era.
Learning deep features for discriminative localization
Bolei Zhou, Aditya Khosla, Agata Lapedriza, Aude Oliva, and Antonio Torralba · 2016
Cited alongside, same era.
A downsampled variant of imagenet as an alternative to the cifar datasets
Original
Patryk Chrabaszcz, Ilya Loshchilov, and Frank Hutter · 2017
Cited alongside, same era.
Real time image saliency for black box classifiers
Piotr Dabkowski and Yarin Gal · 2017
Cited alongside, same era.
Improved regularization of convolutional neural networks with cutout
Original
Terrance DeVries and Graham W Taylor · 2017
Cited alongside, same era.
Interpretable explanations of black boxes by meaningful perturbation
Ruth C Fong and Andrea Vedaldi · 2017
Cited alongside, same era.
Deep pyramidal residual networks
Dongyoon Han, Jiwhan Kim, and Junmo Kim · 2017
Cited alongside, same era.
Deep learning scaling is predictable, empirically
Original
Joel Hestness, Sharan Narang, Newsha Ardalani, Gregory Diamos, Heewoo Jun, Hassan Kianinejad, Md Patwary, Mostofa Ali, Yang Yang, and Yanqi Zhou · 2017
Cited alongside, same era.
Smart augmentation learning an optimal data augmentation strategy
Joseph Lemley, Shabab Bazrafkan, and Peter Corcoran · 2017
Cited alongside, same era.
The effectiveness of data augmentation in image classification using deep learning
Original
Luis Perez and Jason Wang · 2017
Cited alongside, same era.

Then

mixup: Beyond empirical risk minimization
Original
Hongyi Zhang, Moustapha Cisse, Yann N Dauphin, and David Lopez-Paz · 2017
Later among the works it cites.
Adversarially robust generalization requires more data
Ludwig Schmidt, Shibani Santurkar, Dimitris Tsipras, Kunal Talwar, and Aleksander Madry · 2018
Later among the works it cites.
Between-class learning for image classification
Yuji Tokozume, Yoshitaka Ushiku, and Tatsuya Harada · 2018
Later among the works it cites.
Autoaugment: Learning augmentation strategies from data
Ekin D Cubuk, Barret Zoph, Dandelion Mane, Vijay Vasudevan, and Quoc V Le · 2019
Later among the works it cites.
Randaugment: Practical data augmentation with no separate search
Original
Ekin D Cubuk, Barret Zoph, Jonathon Shlens, and Quoc V Le · 2019
Later among the works it cites.
Smoothfool: An efficient framework for computing smooth adversarial perturbations
Original
Ali Dabouei, Sobhan Soleymani, Fariborz Taherkhani, Jeremy Dawson, and Nasser M Nasrabadi · 2019
Later among the works it cites.
Mixup as locally linear out-of-manifold regularization
Hongyu Guo, Yongyi Mao, and Richong Zhang · 2019
Later among the works it cites.
Population based augmentation: Efficient learning of augmentation policy schedules
Original
Daniel Ho, Eric Liang, Ion Stoica, Pieter Abbeel, and Xi Chen · 2019
Later among the works it cites.
Fast autoaugment
Sungbin Lim, Ildoo Kim, Taesup Kim, Chiheon Kim, and Sungwoong Kim · 2019
Later among the works it cites.
Contrastive representation distillation
Original
Yonglong Tian, Dilip Krishnan, and Phillip Isola · 2019
Later among the works it cites.
Cutmix: Regularization strategy to train strong classifiers with localizable features
Original
Sangdoo Yun, Dongyoon Han, Seong Joon Oh, Sanghyuk Chun, Junsuk Choe, and Youngjoon Yoo · 2019
Later among the works it cites.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…

Reading the bibliography…