Data augmentation for improving deep learning in image classification problem

By Agnieszka Mikołajczyk, Michal Grochowski. 2018

1,460 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Biophysics, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Biophysics

In Computer Vision and Pattern Recognition

Citations in the corpus

Style transfer-based image synthesis as an efficient regularization technique in deep learning (2019, methodology): “significantly improve the learning abilities of algorithm used as well as further generalization abilities of the networks [10].”
Low-Resource Expressive Text-To-Speech Using Data Augmentation (2020, background): “Especially in computer vision, creating synthetic data to train better models has been thoroughly investigated [14–16].”
Data Augmentation techniques in time series domain: a survey and taxonomy (2022, methodology): “Traditionally, in fields such as image recognition, different transformations have been applied to data such as cropping [23, 24, 25, 26], scaling [24, 25], mirroring [23, 27, 26], colour augmentation [23, 25, 28] or translation [27].”
Machine learning applied in production planning and control: a state-of-the-art in the era of industry 4.0 (2020, background): “To tackle this issue, some authors have proposed solutions such as data augmentation, a common practice in computer vision that consists of artificially creating new training examples by modifying existent observations (Perez and Wang 2017; Mikołajczyk and Grochowski 2018).”
Addressing bias in big data and AI for health care: A call for open science (2021, background): “60 This can allow neural networks to be trained with more samples of data that may be underrepresented, such as data of ethnic minorities.”
Deep neural network architecture search using network morphism (2019, positive): “This cause need of a special method to deal with this phenomenon e.g. usage of upsampling or downsampling, weighted loss function, or the use of advanced data augmentation methods i.e. Generative Adversarial Networks [17] or Neural Style Transfer [18], [19].”
CNN-based transfer learning–BiLSTM network: A novel approach for COVID-19 infection detection (2020, methodology): “While this technique provides data diversity and more accurate classification, it also includes disadvantages such as additional memory, conversion calculation costs, and additional training time [50,51].”
Improving robustness against common corruptions by covariate shift adaptation (2020, methodology): “This includes data augmentation like Gaussian noise [44], optimized mixtures of data augmentations in conjunction with a consistency loss [30], training on stylized images [28, 42, 45] or against adversarial noise distributions [29].”
Data Augmentation in Classification and Segmentation: A Survey and New Strategies (2023, methodology): “The work in [58] developed an image analogue technique, using pixel resampling to transfer textures from one image to another [59].”
A review of medical image data augmentation techniques for deep learning applications (2021, background): “10 However, these articles mainly focus on generative adversarial networks (GANs) based augmentation, 9,10 a speci ﬁ c task of either image classi ﬁ cation or segmentation, 6,7 or general data augmentation for computer vision.”
FUTURE-AI: international consensus guideline for trustworthy and deployable artificial intelligence in healthcare (2023, neutral): “Additionally, it is important to show that the AI tool is safe and does not cause harm to individuals (or specific groups), such as through a randomised clinical trial.”
A Simple Way to Make Neural Networks Robust Against Diverse Image Corruptions (2020, methodology): “A popular method to decrease overﬁtting and help the network generalize better to unseen data is to augment the training dataset by applying a set of (randomized) manipulations to the images [26].”
CovidGAN: Data Augmentation Using Auxiliary Classifier GAN for Improved Covid-19 Detection (2020, methodology): “Current data augmentation techniques use simple modifications to incorporate affinity like image transformations and color adjustments, such as scaling, flipping, converting, improving contrast or brightness, blurring, and sharpening, white balance, etc [8].”
Label-Only Membership Inference Attacks (2020, background): “Data augmentation is commonly used in machine learning to prevent a model from overfitting, in particular in low data regimes [29, 54, 55, 56, 57].”