Dataset Augmentation in Feature Space

Abstract

Dataset augmentation, the practice of applying a wide array of domain-specific transformations to synthetically expand a training set, is a standard tool in supervised learning. While effective in tasks such as visual recognition, the set of transformations must be carefully designed, implemented, and tested for every new domain, limiting its re-use and generality. In this paper, we adopt a simpler, domain-agnostic approach to dataset augmentation. We start with existing data points and apply simple transformations such as adding noise, interpolating, or extrapolating between them. Our main insight is to perform the transformation not in input space, but in a learned feature space. A re-kindling of interest in unsupervised representation learning makes this technique timely and more effective. It is a simple proposal, but to-date one that has not been tested empirically. Working in the space of context vectors generated by sequence-to-sequence models, we demonstrate a technique that is effective for both static and sequential data.

Cited by in corpus (20)

• mixup: Beyond Empirical Risk Minimization
• Learning Representations for Neural Network-Based Classification Using the Information Bottleneck Principle
• ShakeDrop Regularization for Deep Residual Learning
• Further advantages of data augmentation on convolutional neural networks
• Cross-Domain Few-Shot Classification via Learned Feature-Wise Transformation
• Recent Advances in Object Detection in the Age of Deep Convolutional Neural Networks
• Data augmentation instead of explicit regularization
• An overview of mixing augmentation methods and augmentation strategies
• Dual Attention Matching Network for Context-Aware Feature Sequence based Person Re-Identification
• Sampling Strategies for GAN Synthetic Data
• PointAugment: an Auto-Augmentation Framework for Point Cloud Classification
• Adversarial Learning of General Transformations for Data Augmentation
• Deep Learning-Based Strategy for Macromolecules Classification with Imbalanced Data from Cellular Electron Cryotomography
• Data augmentation and image understanding
• Augmentation Inside the Network
• Image Augmentation Using a Task Guided Generative Adversarial Network for Age Estimation on Brain MRI
• Fast Mesh Data Augmentation via Chebyshev Polynomial of Spectral filtering
• Mitigating Generation Shifts for Generalized Zero-Shot Learning
• ReMix: Towards Image-to-Image Translation with Limited Data
• CoopSubNet: Cooperating Subnetwork for Data-Driven Regularization of Deep Networks under Limited Training Budgets