Research map / Probabilistic and scientific modeling / Variational and generative sampling
The Gaussian Discriminant Variational Autoencoder (GdVAE): A Self-Explainable Model with Counterfactual Explanations
Anselm Haselhoff, Kevin Trelenberg, Fabian Küppers, Jonas Schneider
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 EM-based inference in latent space instead of importance sampling in image space (ablation) | GdVAE · CIFAR-10 | accuracy (%): 55.0 -> 65.1 | Improved |
| use EM-based inference instead of importance sampling (ablation) | GdVAE · CelebA - Gender | reconstruction MSE (x10^2): 1.77 -> 0.91 | Improved |
| use EM-based inference instead of importance sampling (ablation) | GdVAE · MNIST | accuracy (%): 99.0 -> 99.0 | No clear effect |
| GdVAE (transparent) versus black-box jointly trained CVAE classifier | GdVAE · CIFAR-10 | accuracy (%): 69.0 -> 65.1 | Hurt |
| use ProtoVAE's augmentation and preprocessing with GdVAE | GdVAE · CIFAR-10 | accuracy (%): 65.1 -> 76.8 | Improved |
| class-conditional GdVAE instead of unconditional ProtoVAE | GdVAE · CelebA - Gender | reconstruction MSE (x10^2): 1.32 -> 0.71 | Improved |
| use global counterfactuals (move toward opposing prototype) instead of local (ablation) | GdVAE CFs · CelebA Smiling CFs | FID: 85.56 -> 128.93 | Hurt |
| use global counterfactuals instead of local-M (ablation) | GdVAE CFs · MNIST binary 0/1 CFs | consistency ACC (%): 44.6 -> 54.2 | Improved |
| use Mahalanobis (local-M) instead of L2 (local-L2) metric for local CFs (ablation) | GdVAE CFs · CelebA Smiling CFs | proximity MSE (x10^2): 0.99 -> 0.92 | No clear effect |
| train latent space jointly with the classifier so that CFs use a linear explainer | GdVAE local CFs · CelebA Smiling CFs | FID: 95.35 -> 85.52 | Improved |
Closest papers by meaning
- Rethinking Visual Counterfactual Explanations Through Region Constraint (ICLR 2025 · similarity 0.90)
- V-CECE: Visual Counterfactual Explanations via Conceptual Edits (NeurIPS 2025 · similarity 0.88)
- Towards Fine-Grained Interpretability: Counterfactual Explanations for Misclassification with Saliency Partition (CVPR 2025 · similarity 0.88)
- CounterFlowNet: From Minimal Changes to Meaningful Counterfactual Explanations (NeurIPS 2026 · similarity 0.88)
- Counterfactual Explanations for Misclassified Images: How Human and Machine Explanations Differ (Abstract Reprint) (AAAI 2024 · similarity 0.87)
- CF-OPT: Counterfactual Explanations for Structured Prediction (ICML 2024 · similarity 0.87)
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.
