AutoAugment

Introduced by Cubuk et al. in AutoAugment: Learning Augmentation Policies from Data

AutoAugment is an automated approach to find data augmentation policies from data. It formulates the problem of finding the best augmentation policy as a discrete search problem. It consists of two components: a search algorithm and a search space.

At a high level, the search algorithm (implemented as a controller RNN) samples a data augmentation policy $S$, which has information about what image processing operation to use, the probability of using the operation in each batch, and the magnitude of the operation. The policy $S$ is used to train a neural network with a fixed architecture, whose validation accuracy $R$ is sent back to update the controller. Since $R$ is not differentiable, the controller will be updated by policy gradient methods.

The operations used are from PIL, a popular Python image library: all functions in PIL that accept an image as input and output an image. It additionally uses two other augmentation techniques: Cutout and SamplePairing. The operations searched over are ShearX/Y, TranslateX/Y, Rotate, AutoContrast, Invert, Equalize, Solarize, Posterize, Contrast, Color, Brightness, Sharpness, Cutout and Sample Pairing.

Source: AutoAugment: Learning Augmentation Policies from Data

Latest Papers

PAPER DATE
A Technical Report for VIPriors Image Classification Challenge
Zhipeng LuoGe LiZhiguang Zhang
2020-07-17
Hypernetwork-Based Augmentation
Chih-Yang ChenChe-Han ChangEdward Y. Chang
2020-06-11
AutoCLINT: The Winning Method in AutoCV Challenge 2019
| Woonhyuk BaekIldoo KimSungwoong KimSungbin Lim
2020-05-09
On the Generalization Effects of Linear Transformations in Data Augmentation
| Sen WuHongyang R. ZhangGregory ValiantChristopher Ré
2020-05-02
Supervised Contrastive Learning
| Prannay KhoslaPiotr TeterwakChen WangAaron SarnaYonglong TianPhillip IsolaAaron MaschinotCe LiuDilip Krishnan
2020-04-23
ResNeSt: Split-Attention Networks
| Hang ZhangChongruo WuZhongyue ZhangYi ZhuZhi ZhangHaibin LinYue SunTong HeJonas MuellerR. ManmathaMu LiAlexander Smola
2020-04-19
UniformAugment: A Search-free Probabilistic Data Augmentation Approach
| Tom Ching LingChenAva KhonsariAmirreza LashkariMina Rafi NazariJaspreet Singh SambeeMario A. Nascimento
2020-03-31
TResNet: High Performance GPU-Dedicated Architecture
| Tal RidnikHussam LawenAsaf NoyItamar FriedmanEmanuel Ben BaruchGilad Sharir
2020-03-30
Circumventing Outliers of AutoAugment with Knowledge Distillation
Longhui WeiAn XiaoLingxi XieXin ChenXiaopeng ZhangQi Tian
2020-03-25
SuperMix: Supervising the Mixing Data Augmentation
| Ali DaboueiSobhan SoleymaniFariborz TaherkhaniNasser M. Nasrabadi
2020-03-10
DADA: Differentiable Automatic Data Augmentation
| Yonggang LiGuosheng HuYongtao WangTimothy HospedalesNeil M. RobertsonYongxin Yang
2020-03-08
Compounding the Performance Improvements of Assembled Techniques in a Convolutional Neural Network
| Jungkyu LeeTaeryun WonTae Kwan LeeHyemin LeeGeonmo GuKiho Hong
2020-01-17
GridMask Data Augmentation
| Pengguang ChenShu LiuHengshuang ZhaoJiaya Jia
2020-01-13
ReMixMatch: Semi-Supervised Learning with Distribution Matching and Augmentation Anchoring
| Anonymous
2020-01-01
Resizable Neural Networks
Anonymous
2020-01-01
Adversarial AutoAugment
Xinyu ZhangQiang WangJian ZhangZhao Zhong
2019-12-24
ReMixMatch: Semi-Supervised Learning with Distribution Alignment and Augmentation Anchoring
| David BerthelotNicholas CarliniEkin D. CubukAlex KurakinKihyuk SohnHan ZhangColin Raffel
2019-11-21
Adversarial Examples Improve Image Recognition
| Cihang XieMingxing TanBoqing GongJiang WangAlan YuilleQuoc V. Le
2019-11-21
Faster AutoAugment: Learning Augmentation Strategies using Backpropagation
| Ryuichiro HatayaJan ZdenekKazuki YoshizoeHideki Nakayama
2019-11-16
RandAugment: Practical automated data augmentation with a reduced search space
| Ekin D. CubukBarret ZophJonathon ShlensQuoc V. Le
2019-09-30
Automatically Learning Data Augmentation Policies for Dialogue Tasks
Tong NiuMohit Bansal
2019-09-27
SCARLET-NAS: Bridging the gap between Stability and Scalability in Weight-sharing Neural Architecture Search
| Xiangxiang ChuBo ZhangJixiang LiQingyuan LiRuijun Xu
2019-08-16
Greedy AutoAugment
Alireza NaghizadehMohammadsajad AbavisaniDimitris N. Metaxas
2019-08-02
A Fourier Perspective on Model Robustness in Computer Vision
Dong YinRaphael Gontijo LopesJonathon ShlensEkin D. CubukJustin Gilmer
2019-06-21
Improving Robustness Without Sacrificing Accuracy with Patch Gaussian Augmentation
Raphael Gontijo LopesDong YinBen PooleJustin GilmerEkin D. Cubuk
2019-06-06
AutoAugment: Learning Augmentation Strategies From Data
Ekin D. Cubuk Barret Zoph Dandelion Mane Vijay Vasudevan Quoc V. Le
2019-06-01
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
| Mingxing TanQuoc V. Le
2019-05-28
Population Based Augmentation: Efficient Learning of Augmentation Policy Schedules
| Daniel HoEric LiangIon StoicaPieter AbbeelXi Chen
2019-05-14
Fast AutoAugment
| Sungbin LimIldoo KimTaesup KimChiheon KimSungwoong Kim
2019-05-01
CondConv: Conditionally Parameterized Convolutions for Efficient Inference
| Brandon YangGabriel BenderQuoc V. LeJiquan Ngiam
2019-04-10
MultiGrain: a unified image embedding for classes and instances
| Maxim BermanHervé JégouAndrea VedaldiIasonas KokkinosMatthijs Douze
2019-02-14
Learning data augmentation policies using augmented random search
Mingyang GengKele XuBo DingHuaimin WangLei Zhang
2018-11-12
AutoAugment: Learning Augmentation Policies from Data
| Ekin D. CubukBarret ZophDandelion ManeVijay VasudevanQuoc V. Le
2018-05-24

Categories