Research map / Probabilistic and scientific modeling / Variational and generative sampling
Regularization with Latent Space Virtual Adversarial Training
Genki Osada, Budrul Ahsan, Revoti Prasad Bora, Takashi Nishide
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use Glow instead of VAE as the latent transformer in LVAT (ablation) | LVAT · CIFAR-10 (SSL, 4000 labels) | error rate (%): 13.90 -> 9.94 | Improved |
| use Glow instead of VAE as the latent transformer in LVAT (ablation) | LVAT · CIFAR-10 w/ aug (SSL, 4000 labels) | error rate (%): 14.64 -> 7.34 | Improved |
| use Glow instead of VAE as the latent transformer in LVAT (ablation) | LVAT · SVHN (SSL, 1000 labels) | error rate (%): 4.44 -> 4.20 | Improved |
| perturb in latent space (LVAT-Glow) instead of input space (VAT) | VAT · CIFAR-10 (SSL, 4000 labels) | error rate (%): 16.92 -> 9.94 | Improved |
| perturb in latent space (LVAT-Glow) instead of input space (VAT) | VAT · SVHN (SSL, 1000 labels) | error rate (%): 5.77 -> 4.20 | Improved |
| perturb in latent space (LVAT-VAE) instead of input space (VAT) | VAT · CIFAR-10 w/ aug (SSL, 4000 labels) | error rate (%): 11.36 -> 14.64 | Hurt |
| perturb in latent space (LVAT-VAE) instead of input space (VAT) | VAT · CIFAR-10 (SSL, 4000 labels) | error rate (%): 16.92 -> 13.90 | Improved |
| perturb in latent space (LVAT-Glow) instead of input space (VAT) | VAT · CIFAR-10 w/ aug (SL) | error rate (%): 5.81 -> 3.94 | Improved |
Closest papers by meaning
- Tangent-Normal Adversarial Regularization for Semi-Supervised Learning (CVPR 2019 · similarity 0.87)
- Virtual Adversarial Training: A Regularization Method for Supervised and Semi-Supervised Learning (TPAMI 2018 · similarity 0.86)
- Consistency Regularization for Adversarial Robustness (AAAI 2022 · similarity 0.86)
- EQ-VAE: Equivariance Regularized Latent Space for Improved Generative Image Modeling (ICML 2025 · similarity 0.86)
- Rethinking Invariance Regularization in Adversarial Training to Improve Robustness-Accuracy Trade-off (ICLR 2025 · similarity 0.85)
- Kernel Learning with Adversarial Features: Numerical Efficiency and Adaptive Regularization (NeurIPS 2025 · similarity 0.85)
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.
