Smart Augmentation - Learning an Optimal Data Augmentation Strategy

Abstract

A recurring problem faced when training neural networks is that there is typically not enough data to maximize the generalization capability of deep neural networks(DNN). There are many techniques to address this, including data augmentation, dropout, and transfer learning. In this paper, we introduce an additional method which we call Smart Augmentation and we show how to use it to increase the accuracy and reduce overfitting on a target network. Smart Augmentation works by creating a network that learns how to generate augmented data during the training process of a target network in a way that reduces that networks loss. This allows us to learn augmentations that minimize the error of that network. Smart Augmentation has shown the potential to increase accuracy by demonstrably significant measures on all datasets tested. In addition, it has shown potential to achieve similar or improved performance levels with significantly smaller network sizes in a number of tested cases.

References in corpus (5)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Explaining and Harnessing Adversarial Examples
• Synthetic Data and Artificial Neural Networks for Natural Scene Text Recognition
• Return of the Devil in the Details: Delving Deep into Convolutional Nets

Cited by in corpus (57)

• Improved Regularization of Convolutional Neural Networks with Cutout
• Albumentations: fast and flexible image augmentations
• AutoAugment: Learning Augmentation Policies from Data
• A Comprehensive Survey of Image Augmentation Techniques for Deep Learning
• Further advantages of data augmentation on convolutional neural networks
• An End to End Deep Neural Network for Iris Segmentation in Unconstraint Scenarios
• A Survey on Face Data Augmentation
• Cross-Domain Few-Shot Classification via Learned Feature-Wise Transformation
• Recent Advances in Object Detection in the Age of Deep Convolutional Neural Networks
• A survey of synthetic data augmentation methods in computer vision
• Data augmentation instead of explicit regularization
• SaliencyMix: A Saliency Guided Data Augmentation Strategy for Better Regularization
• An overview of mixing augmentation methods and augmentation strategies
• Unsupervised Histopathology Image Synthesis
• Adversarial Examples Improve Image Recognition
• Sampling Strategies for GAN Synthetic Data
• Data Augmentation Revisited: Rethinking the Distribution Gap between Clean and Augmented Data
• An embedded system for the automated generation of labeled plant images to enable machine learning applications in agriculture
• An Analysis of an Integrated Mathematical Modeling -- Artificial Neural Network Approach for the Problems with a Limited Learning Dataset
• Adapting Semantic Segmentation Models for Changes in Illumination and Camera Perspective
• GuidedMixup: An Efficient Mixup Strategy Guided by Saliency Maps
• Domain Randomization and Pyramid Consistency: Simulation-to-Real Generalization without Accessing Target Domain Data
• PointAugment: an Auto-Augmentation Framework for Point Cloud Classification
• Augmented Hard Example Mining for Generalizable Person Re-Identification
• Role-Wise Data Augmentation for Knowledge Distillation
• PseudoAugment: Learning to Use Unlabeled Data for Data Augmentation in Point Clouds
• On Data-Augmentation and Consistency-Based Semi-Supervised Learning
• Enhancing Robustness of Deep Neural Networks Against Adversarial Malware Samples: Principles, Framework, and AICS'2019 Challenge
• BdSL36: A Dataset for Bangladeshi Sign Letters Recognition
• Efficient CNN Implementation for Eye-Gaze Estimation on Low-Power/Low-Quality Consumer Imaging Systems
• Improving 3D Object Detection through Progressive Population Based Augmentation
• Image Augmentation for Multitask Few-Shot Learning: Agricultural Domain Use-Case
• Online Hyper-parameter Learning for Auto-Augmentation Strategy
• Data Augmentation for Object Detection via Differentiable Neural Rendering
• Data Augmentation Using Adversarial Training for Construction-Equipment Classification
• Adversarial Learning of General Transformations for Data Augmentation
• Face Synthesis with Landmark Points from Generative Adversarial Networks and Inverse Latent Space Mapping
• Parallel Grid Pooling for Data Augmentation
• Robust and Accurate Object Detection via Adversarial Learning
• Learning Optimal Data Augmentation Policies via Bayesian Optimization for Image Classification Tasks
• Smart Inference for Multidigit Convolutional Neural Network based Barcode Decoding
• Deep Complementary Joint Model for Complex Scene Registration and Few-shot Segmentation on Medical Images
• Data augmentation and image understanding
• EEG Classification by factoring in Sensor Configuration
• The Penalty Imposed by Ablated Data Augmentation
• DTWSSE: Data Augmentation with a Siamese Encoder for Time Series
• Improving Augmentation and Evaluation Schemes for Semantic Image Synthesis
• Automatically Learning Data Augmentation Policies for Dialogue Tasks
• Rapid Identification of X-ray Diffraction Spectra Based on Very Limited Data by Interpretable Convolutional Neural Networks
• Controllable Data Augmentation Through Deep Relighting
• What augmentations are sensitive to hyper-parameters and why?
• Global and Local Texture Randomization for Synthetic-to-Real Semantic Segmentation
• A novel multiple instance learning framework for COVID-19 severity assessment via data augmentation and self-supervised learning
• Diabetic Retinopathy detection by retinal image recognizing
• Cross-modal Spectrum Transformation Network For Acoustic Scene classification
• Safe Augmentation: Learning Task-Specific Transformations from Data
• Training Lightweight CNNs for Human-Nanodrone Proximity Interaction from Small Datasets using Background Randomization