Understanding data augmentation for classification: when to warp?

2016/09/28 by Wong, Sebastien C., Gatt, Adam, Stamatescu, Victor +1 · 6 citations
#Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #I.4.7 #I.5.2

paper · doi:10.48550/arxiv.1609.08764

Abstract

In this paper we investigate the benefit of augmenting data with synthetically created samples when training a machine learning classifier. Two approaches for creating additional training samples are data warping, which generates additional samples through transformations applied in the data-space, and synthetic over-sampling, which creates additional samples in feature-space. We experimentally evaluate the benefits of data augmentation for a convolutional backpropagation-trained neural network, a convolutional support vector machine and a convolutional extreme learning machine classifier, using the standard MNIST handwritten digit dataset. We found that while it is possible to perform generic augmentation in feature-space, if plausible transforms for the data are known then augmentation in data-space provides a greater benefit for improving performance and reducing overfitting.

Cited by

• Oya: Deep Learning for Accurate Global Precipitation Estimation
• Archaeological Classification of Small Datasets Using Meta- and Transfer Learning Methods: A Case Study on Hittite Stele Fragments
• An Analysis of Causal Effect Estimation using Outcome Invariant Data Augmentation
• Bridging AI Innovation and Healthcare Needs: Lessons Learned from Incorporating Modern NLP at The BC Cancer Registry
• DRIFT: Data Reduction via Informative Feature Transformation- Generalization Begins Before Deep Learning starts
• CONCORD: Concept-Informed Diffusion for Dataset Distillation

Related

• Adversarial Learning of General Transformations for Data Augmentation
• A Kernel Theory of Modern Data Augmentation
• Untapped Potential of Data Augmentation: A Domain Generalization Viewpoint
• A Preliminary Study on Data Augmentation of Deep Learning for Image Classification
• E-Stitchup: Data Augmentation for Pre-Trained Embeddings