Data Augmentation in High Dimensional Low Sample Size Setting Using a Geometry-Based Variational Autoencoder

Abstract

In this paper, we propose a new method to perform data augmentation in a reliable way in the High Dimensional Low Sample Size (HDLSS) setting using a geometry-based variational autoencoder. Our approach combines a proper latent space modeling of the VAE seen as a Riemannian manifold with a new generation scheme which produces more meaningful samples especially in the context of small data sets. The proposed method is tested through a wide experimental study where its robustness to data sets, classifiers and training samples size is stressed. It is also validated on a medical imaging classification task on the challenging ADNI database where a small number of 3D brain MRIs are considered and augmented using the proposed VAE framework. In each case, the proposed method allows for a significant and reliable gain in the classification metrics. For instance, balanced accuracy jumps from 66.3% to 74.3% for a state-of-the-art CNN classifier trained with 50 MRIs of cognitively normal (CN) and 50 Alzheimer disease (AD) patients and from 77.7% to 86.3% when trained with 243 CN and 210 AD while improving greatly sensitivity and specificity metrics.

arXiv commentaccepted to IEEE transactions on pattern analysis and machine intelligence (TPAMI)

References in corpus (30)

• SMOTE: Synthetic Minority Over-sampling Technique
• Auto-Encoding Variational Bayes
• Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms
• GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
• GAN-based Synthetic Medical Image Augmentation for increased CNN Performance in Liver Lesion Classification
• Generative Adversarial Network in Medical Imaging: A Review
• Variational Inference with Normalizing Flows
• Convolutional Neural Networks for Classification of Alzheimer's Disease: Overview and Reproducible Evaluation
• Data Augmentation Generative Adversarial Networks
• CovidGAN: Data Augmentation Using Auxiliary Classifier GAN for Improved Covid-19 Detection
• Deep Unsupervised Clustering with Gaussian Mixture Variational Autoencoders
• Markov Chain Monte Carlo and Variational Inference: Bridging the Gap
• Deep Variational Information Bottleneck
• BAGAN: Data Augmentation with Balancing GAN
• Variational Lossy Autoencoder
• Importance Weighted Autoencoders
• Inference Suboptimality in Variational Autoencoders
• Cardiac MRI Segmentation with Strong Anatomical Guarantees
• Explorations in Homeomorphic Variational Auto-Encoding
• Metrics for Deep Generative Models
• Continuous Hierarchical Representations with Poincaré Variational Auto-Encoders
• Lung Segmentation from Chest X-rays using Variational Data Imputation
• High-Resolution Mammogram Synthesis using Progressive Generative Adversarial Networks
• Riemannian Manifold Hamiltonian Monte Carlo
• A Contrastive Divergence for Combining Variational Inference and MCMC
• Poincaré Wasserstein Autoencoder
• A Contrastive Learning Approach for Training Variational Autoencoder Priors
• Geometry-Aware Hamiltonian Variational Auto-Encoder
• Latent Space Cartography: Generalised Metric-Inspired Measures and Measure-Based Transformations for Generative Models
• A prior-based approximate latent Riemannian metric

Cited by in corpus (5)

• Deep Learning Approaches for Data Augmentation in Medical Imaging: A Review
• Discriminative Hamiltonian Variational Autoencoder for Accurate Tumor Segmentation in Data-Scarce Regimes
• Exploring Privacy and Fairness Risks in Sharing Diffusion Models: An Adversarial Perspective
• Pythae: Unifying Generative Autoencoders in Python -- A Benchmarking Use Case
• Learning to Solve Inverse Problems for Perceptual Sound Matching