Data Augmentation Generative Adversarial Networks
arXiv:1711.04340
Abstract
Effective training of neural networks requires much data. In the low-data regime, parameters are underdetermined, and learnt networks generalise poorly. Data Augmentation alleviates this by using existing data more effectively. However standard data augmentation produces only limited plausible alternative data. Given there is potential to generate a much broader set of augmentations, we design and train a generative model to do data augmentation. The model, based on image conditional Generative Adversarial Networks, takes data from a source domain and learns to take any data item and generalise it to generate other within-class data items. As this generative process does not depend on the classes themselves, it can be applied to novel unseen classes of data. We show that a Data Augmentation Generative Adversarial Network (DAGAN) augments standard vanilla classifiers well. We also show a DAGAN can enhance few-shot learning systems such as Matching Networks. We demonstrate these approaches on Omniglot, on EMNIST having learnt the DAGAN on Omniglot, and VGG-Face data. In our experiments we can see over 13% increase in accuracy in the low-data regime experiments in Omniglot (from 69% to 82%), EMNIST (73.9% to 76%) and VGG-Face (4.5% to 12%); in Matching Networks for Omniglot we observe an increase of 0.5% (from 96.9% to 97.4%) and an increase of 1.8% in EMNIST (from 59.5% to 61.3%).
10 pages
References in corpus (2)
Cited by in corpus (178)
- AutoML: A Survey of the State-of-the-Art
- AutoAugment: Learning Augmentation Policies from Data
- A Comprehensive Survey of Image Augmentation Techniques for Deep Learning
- Regularized Deep Networks in Intelligent Transportation Systems: A Taxonomy and a Case Study
- GAN Augmentation: Augmenting Training Data using Generative Adversarial Networks
- Delta-encoder: an effective sample synthesis method for few-shot object recognition
- A Closer Look at Few-shot Classification
- Selective Synthetic Augmentation with HistoGAN for Improved Histopathology Image Classification
- Further advantages of data augmentation on convolutional neural networks
- Interventional Few-Shot Learning
- Recent Advances in Object Detection in the Age of Deep Convolutional Neural Networks
- Understanding Back-Translation at Scale
- Data Augmentation in High Dimensional Low Sample Size Setting Using a Geometry-Based Variational Autoencoder
- Balanced Semi-Supervised Generative Adversarial Network for Damage Assessment from Low-Data Imbalanced-Class Regime
- Data augmentation instead of explicit regularization
- Assuring the Machine Learning Lifecycle: Desiderata, Methods, and Challenges
- Generative Adversarial Networks for Image-to-Image Translation on Multi-Contrast MR Images - A Comparison of CycleGAN and UNIT
- How to train your MAML
- Deep Feature Augmentation for Occluded Image Classification
- SESAME: Semantic Editing of Scenes by Adding, Manipulating or Erasing Objects
- Bias Correction of Learned Generative Models using Likelihood-Free Importance Weighting
- Free-form tumor synthesis in computed tomography images via richer generative adversarial network
- Machine Learning in NextG Networks via Generative Adversarial Networks
- Small Sample Learning in Big Data Era
- Augment your batch: better training with larger batches
- Model Patching: Closing the Subgroup Performance Gap with Data Augmentation
- MelanoGANs: High Resolution Skin Lesion Synthesis with GANs
- Deep Robust Clustering by Contrastive Learning
- ResizeMix: Mixing Data with Preserved Object Information and True Labels
- Contrastive Learning with Stronger Augmentations
- SymmetryGAN: Symmetry Discovery with Deep Learning
- Generative Adversarial Network for Radar Signal Generation
- Generative Image Translation for Data Augmentation of Bone Lesion Pathology
- Break-It-Fix-It: Unsupervised Learning for Program Repair
- Defining Benchmarks for Continual Few-Shot Learning
- Deep Generative Models for Library Augmentation in Multiple Endmember Spectral Mixture Analysis
- Sequence-to-Sequence Data Augmentation for Dialogue Language Understanding
- High-resolution Deep Convolutional Generative Adversarial Networks
- Weakly-supervised Compositional FeatureAggregation for Few-shot Recognition
- Regularizing Deep Networks with Semantic Data Augmentation
- Task Augmentation by Rotating for Meta-Learning
- Viewmaker Networks: Learning Views for Unsupervised Representation Learning
- Data Augmentation and Classification of Sea-Land Clutter for Over-the-Horizon Radar Using AC-VAEGAN
- Sill-Net: Feature Augmentation with Separated Illumination Representation
- Towards generative adversarial networks as a new paradigm for radiology education
- Random Matrix Theory Proves that Deep Learning Representations of GAN-data Behave as Gaussian Mixtures
- Style-transfer GANs for bridging the domain gap in synthetic pose estimator training
- DC-Check: A Data-Centric AI checklist to guide the development of reliable machine learning systems
- Local Augmentation for Graph Neural Networks
- Making Invisible Visible: Data-Driven Seismic Inversion with Spatio-temporally Constrained Data Augmentation
- MetaSAug: Meta Semantic Augmentation for Long-Tailed Visual Recognition
- Imperfect ImaGANation: Implications of GANs Exacerbating Biases on Facial Data Augmentation and Snapchat Selfie Lenses
- Conditional Infilling GANs for Data Augmentation in Mammogram Classification
- Synthesizing lesions using contextual GANs improves breast cancer classification on mammograms
- Counterfactual Vision-and-Language Navigation via Adversarial Path Sampling
- Dynamic Relevance Learning for Few-Shot Object Detection
- Learning Stage-wise GANs for Whistle Extraction in Time-Frequency Spectrograms
- Synthetic Observational Health Data with GANs: from slow adoption to a boom in medical research and ultimately digital twins?
- Unified Adversarial Invariance
- Pathological Retinal Region Segmentation From OCT Images Using Geometric Relation Based Augmentation
- Data Augmentation for Deep Learning-based Radio Modulation Classification
- Joint Contrastive Learning with Infinite Possibilities
- Imbalanced Data Learning by Minority Class Augmentation using Capsule Adversarial Networks
- Enhancing the Performance of Practical Profiling Side-Channel Attacks Using Conditional Generative Adversarial Networks
- Online Hyper-parameter Learning for Auto-Augmentation Strategy
- Targeted Nonlinear Adversarial Perturbations in Images and Videos
- General-to-Detailed GAN for Infrequent Class Medical Images
- Data Augmentation for Skin Lesion using Self-Attention based Progressive Generative Adversarial Network
- Dynamics of Fourier Modes in Torus Generative Adversarial Networks
- Multi-modal AsynDGAN: Learn From Distributed Medical Image Data without Sharing Private Information
- Negative Data Augmentation
- DADA: Deep Adversarial Data Augmentation for Extremely Low Data Regime Classification
- Data-Efficient GAN Training Beyond (Just) Augmentations: A Lottery Ticket Perspective
- Meta-learning algorithms for Few-Shot Computer Vision
- Imagining an Engineer: On GAN-Based Data Augmentation Perpetuating Biases
- Revisiting Few-Shot Learning for Facial Expression Recognition
- Data augmentation with Symbolic-to-Real Image Translation GANs for Traffic Sign Recognition
- Developing Future Human-Centered Smart Cities: Critical Analysis of Smart City Security, Interpretability, and Ethical Challenges
- Addressing Class Imbalance in Classification Problems of Noisy Signals by using Fourier Transform Surrogates
- A Plug-in Method for Representation Factorization in Connectionist Models
- Machine learning with limited data
- Attentive Normalization for Conditional Image Generation
- Assessing Intelligence in Artificial Neural Networks
- Diffusion-Weighted Magnetic Resonance Brain Images Generation with Generative Adversarial Networks and Variational Autoencoders: A Comparison Study
- GAN-based Data Augmentation for Chest X-ray Classification
- ACGAN-based Data Augmentation Integrated with Long-term Scalogram for Acoustic Scene Classification
- DeltaGAN: Towards Diverse Few-shot Image Generation with Sample-Specific Delta
- Deep Learning Models in Software Requirements Engineering
- Selective Synthetic Augmentation with Quality Assurance
- OnlineAugment: Online Data Augmentation with Less Domain Knowledge
- Learning Optimal Data Augmentation Policies via Bayesian Optimization for Image Classification Tasks
- MatchingGAN: Matching-based Few-shot Image Generation
- Effective Data Augmentation with Multi-Domain Learning GANs
- Snore-GANs: Improving Automatic Snore Sound Classification with Synthesized Data
- Generative Interventions for Causal Learning
- SB-MTL: Score-based Meta Transfer-Learning for Cross-Domain Few-Shot Learning
- Learning Prototype-oriented Set Representations for Meta-Learning
- Influence Estimation for Generative Adversarial Networks
- Conditional Distribution Modelling for Few-Shot Image Synthesis with Diffusion Models
- Learning from Data-Rich Problems: A Case Study on Genetic Variant Calling
- Automatic Data Augmentation by Learning the Deterministic Policy
- A Survey of Techniques All Classifiers Can Learn from Deep Networks: Models, Optimizations, and Regularization
- Improving the Performance of Fine-Grain Image Classifiers via Generative Data Augmentation
- F2GAN: Fusing-and-Filling GAN for Few-shot Image Generation
- Learning Compositional Representation for Few-shot Visual Question Answering
- Data augmentation and image understanding
- Augmentation-Interpolative AutoEncoders for Unsupervised Few-Shot Image Generation
- Featurized Density Ratio Estimation
- Unsupervised Text Embedding Space Generation Using Generative Adversarial Networks for Text Synthesis
- Distinguishing rule- and exemplar-based generalization in learning systems
- BrainNetGAN: Data augmentation of brain connectivity using generative adversarial network for dementia classification
- Red-GAN: Attacking class imbalance via conditioned generation. Yet another perspective on medical image synthesis for skin lesion dermoscopy and brain tumor MRI
- Joint Learning of Generative Translator and Classifier for Visually Similar Classes
- Signed Laplacian Deep Learning with Adversarial Augmentation for Improved Mammography Diagnosis
- Adversarial Recommendation: Attack of the Learned Fake Users
- Automatically Learning Data Augmentation Policies for Dialogue Tasks
- Why Mixup Improves the Model Performance
- Hallucinated Adversarial Learning for Robust Visual Tracking
- Learning More with Less: Conditional PGGAN-based Data Augmentation for Brain Metastases Detection Using Highly-Rough Annotation on MR Images
- Asymmetric Distribution Measure for Few-shot Learning
- Sample Amplification: Increasing Dataset Size even when Learning is Impossible
- CT Image Synthesis Using Weakly Supervised Segmentation and Geometric Inter-Label Relations For COVID Image Analysis
- ColdGAN: Resolving Cold Start User Recommendation by using Generative Adversarial Networks
- Improving Augmentation and Evaluation Schemes for Semantic Image Synthesis
- Mixup-breakdown: a consistency training method for improving generalization of speech separation models
- An Artificial Intelligence-Based System to Assess Nutrient Intake for Hospitalised Patients
- EVAGAN: Evasion Generative Adversarial Network for Low Data Regimes
- Semantically Robust Unpaired Image Translation for Data with Unmatched Semantics Statistics
- Improving Text Relationship Modeling with Artificial Data
- Adversarial AutoMixup
- Uncertainty-Aware Few-Shot Image Classification
- Unsupervised Domain Alignment to Mitigate Low Level Dataset Biases
- OCT-GAN: Neural ODE-based Conditional Tabular GANs
- Mixture Representation Learning with Coupled Autoencoders
- EMIXER: End-to-end Multimodal X-ray Generation via Self-supervision
- Visualizing Deep Learning-based Radio Modulation Classifier
- Fine-grained Image-to-Image Transformation towards Visual Recognition
- GCCN: Global Context Convolutional Network
- Multi-stage domain adversarial style reconstruction for cytopathological image stain normalization
- Flow-based Generative Models for Learning Manifold to Manifold Mappings
- COIN: Contrastive Identifier Network for Breast Mass Diagnosis in Mammography
- End-to-end Generative Zero-shot Learning via Few-shot Learning
- Defuse: Harnessing Unrestricted Adversarial Examples for Debugging Models Beyond Test Accuracy
- Learning Classifiers for Domain Adaptation, Zero and Few-Shot Recognition Based on Learning Latent Semantic Parts
- Defect-GAN: High-Fidelity Defect Synthesis for Automated Defect Inspection
- ReMarNet: Conjoint Relation and Margin Learning for Small-Sample Image Classification
- Learning End-to-End Action Interaction by Paired-Embedding Data Augmentation
- FusionNet: Enhanced Beam Prediction for mmWave Communications Using Sub-6GHz Channel and A Few Pilots
- Open-Ended Content-Style Recombination Via Leakage Filtering
- The MIDI Degradation Toolkit: Symbolic Music Augmentation and Correction
- 3D-Aided Data Augmentation for Robust Face Understanding
- A Generative Approach Towards Improved Robotic Detection of Marine Litter
- Exploring Frequency Domain Interpretation of Convolutional Neural Networks
- Learning Augmentation Distributions using Transformed Risk Minimization
- Generative Adversarial Networks for Black-Box API Attacks with Limited Training Data
- Twin Systems for DeepCBR: A Menagerie of Deep Learning and Case-Based Reasoning Pairings for Explanation and Data Augmentation
- Labeled Data Generation with Inexact Supervision
- Data Augmentation with Variational Autoencoders and Manifold Sampling
- Semantic Generative Augmentations for Few-Shot Counting
- MA 3 : Model Agnostic Adversarial Augmentation for Few Shot learning
- A Diversity-Enhanced and Constraints-Relaxed Augmentation for Low-Resource Classification
- Towards Fair Cross-Domain Adaptation via Generative Learning
- Learning of Inter-Label Geometric Relationships Using Self-Supervised Learning: Application To Gleason Grade Segmentation
- A Series of Unfortunate Counterfactual Events: the Role of Time in Counterfactual Explanations
- Image Translation for Medical Image Generation -- Ischemic Stroke Lesions
- Informative sample generation using class aware generative adversarial networks for classification of chest Xrays
- Generative Residual Attention Network for Disease Detection
- Restrained Generative Adversarial Network against Overfitting in Numeric Data Augmentation
- Decoupling Features and Coordinates for Few-shot RGB Relocalization
- Regularising Deep Networks with Deep Generative Models
- Distance in Latent Space as Novelty Measure
- Safe Augmentation: Learning Task-Specific Transformations from Data
- GAN based ball screw drive picture database enlargement for failure classification
- Implicit Label Augmentation on Partially Annotated Clips via Temporally-Adaptive Features Learning
- Man versus Machine: AutoML and Human Experts' Role in Phishing Detection
- ProtoShotXAI: Using Prototypical Few-Shot Architecture for Explainable AI
- Pathology-Aware Generative Adversarial Networks for Medical Image Augmentation
- Synthetic Convolutional Features for Improved Semantic Segmentation