AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

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

Clément Chadebec, Elina Thibeau–Sutre, Ninon Burgos, Stéphanie Allassonnière

TPAMI 2022 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Regularizing VAE latent information · 120 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add Hamiltonian flows and Riemannian metric to VAEVAE · MNIST testlog-likelihood: -92.94 -> -82.64Improved
add geometry-aware metric to HVAEHVAE · MNIST testlog-likelihood: -85.33 -> -82.64Improved
add Riemannian metric to HVAEHVAE · MNIST testKL to true posterior: 3.61 -> 3.57Improved
sample with metric-volume distribution instead of priorRHVAE · reduced MNIST (balanced)GAN-train: 82.0 -> 90.1Improved
sample with metric-volume distribution instead of priorRHVAE · reduced MNIST (balanced)GAN-test: 63.1 -> 88.1Improved
sample with metric-volume distribution instead of priorRHVAE · reduced MNIST (unbalanced)GAN-train: 69.3 -> 86.2Improved
replace prior sampling with GMM(20) on RAE vs proposedRAE · reduced EMNISTGAN-train: 82.6 -> 79.5Hurt
increase GMM components in RAERAE · reduced MNIST (balanced)GAN-test: 91.7 -> 96.6Hurt
add 1k synthetic samples per class to real dataOurs · CIFAR (reduced)accuracy: 42.6 -> 49.2Improved
add 1k synthetic samples per class to real dataOurs · MNIST unbalanced (reduced)accuracy: 81.5 -> 89.7Improved
train only on 1k synthetic samples per classOurs · MNIST (reduced)accuracy: 89.9 -> 90.1Improved
add 1k VAE-GMM samples to real dataVAE GMM · CIFAR (reduced)accuracy: 42.6 -> 38.9Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy