Research map / Adversarial robustness and security / Attacks on text and graphs
Resisting Adversarial Attacks Using Gaussian Mixture Variational Autoencoders
Partha Ghosh, Arpan Losalka, Michael J. Black
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 |
|---|---|---|---|
| Apply encoder and decoder thresholds | GMVAE supervised · MNIST test | accuracy: 99.67% -> 97.97% | Hurt |
| Apply encoder and decoder thresholds | GMVAE supervised · SVHN test | accuracy: 95.06% -> 92.80% | Hurt |
| Apply encoder and decoder thresholds | GMVAE supervised · COIL-100 test | accuracy: 99.89% -> 98.40% | Hurt |
| Apply encoder and decoder thresholds | GMVAE semi-supervised · MNIST test | accuracy: 99.1% -> 98.17% | Hurt |
| Apply encoder and decoder thresholds | GMVAE semi-supervised · SVHN test | accuracy: 86.42% -> 83.54% | Hurt |
| Use constant rather than variable encoder-distribution entropy (ablation) | GMVAE · MNIST FGSM detection | adversarial sample detection | Improved |
| Use constant rather than variable encoder-distribution entropy (ablation) | GMVAE · CelebA | FID: 58.3 -> 50.4 | Improved |
| Reclassify rejected test samples using decoder latent optimization | GMVAE supervised · MNIST test | accuracy: 99.07% | Improved |
| Threshold encoder and decoder outputs | GMVAE · CleverHans attacks on SVHN and COIL-100 | adversarial sample rejection | Improved |
| Threshold encoder and decoder outputs | GMVAE · FGSM, Momentum Iterative Method, and Projected Gradient Descent | adversarial sample rejection | Improved |
Closest papers by meaning
- Variational Adversarial Defense: A Bayes Perspective for Adversarial Training (TPAMI 2023 · similarity 0.87)
- Towards Deep Learning Models Resistant to Adversarial Attacks (ICLR 2018 · similarity 0.87)
- Alleviating Adversarial Attacks on Variational Autoencoders with MCMC (NeurIPS 2022 · similarity 0.86)
- Ensemble Adversarial Training: Attacks and Defenses (ICLR 2018 · similarity 0.86)
- Strong Transferable Adversarial Attacks via Ensembled Asymptotically Normal Distribution Learning (CVPR 2024 · similarity 0.85)
- Enhancing the Adversarial Robustness via Manifold Projection (AAAI 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.
