Research map / Probabilistic and scientific modeling / Variational and generative sampling
DAVA: Disentangling Adversarial Variational Autoencoder
Benjamin Estermann, Roger Wattenhofer
This paper tried a common change: Add a reconstruction / auxiliary loss
Add reconstruction loss to PIPE · AbstractDSprites · correlation with supervised metrics Hurt
What 987 accepted papers report for this change
- 882 improved
- 41 no clear effect
- 20 mixed
- 44 hurt
- OPTION: Optimal Transport–Guided Flow Matching for Incomplete and Unaligned Multi-View Clustering ICML 2026 · add reconstruction loss L_rec · Scene-15 (aligned, 70% missing) · ACC: 32.4 -> 37.5Improved
- Detached Skip-Links and $R$-Probe: Decoupling Feature Aggregation from Gradient Propagation for MLLM OCR ICML 2026 · add R-probe auxiliary loss · OCRBench · score: 714.0 -> 721.0Improved
- Rethinking Time-Series Imputation as Conditional Inference along Temporal Evolution ICML 2026 · add reconstruction loss on observed entries (joint vs missing-only supervision) · ETTh1 10% missing · Masked MAE: 0.2002 -> 0.1658Improved
See this change in detailMatched by the wording of each paper's own result, read by AI from the full text. Papers report gains more often than failures, so this shows what gets published, not the chance a change works.
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 DAVA instead of the mean-tuned baseline | VAE · Shapes3D | MIG: 0.52±0.02 -> 0.62±0.05 | Improved |
| Use DAVA instead of the mean-tuned baseline | VAE · Shapes3D | FVAE: 0.92±0.01 -> 0.82±0.03 | Hurt |
| Use DAVA instead of the mean-tuned baseline | VAE · Shapes3D | PIPE: 0.22±0.04 -> 0.61±0.04 | Improved |
| Use DAVA instead of the mean-tuned baseline | VAE · AbstractDSprites | MIG: 0.15±0.01 -> 0.23±0.04 | Improved |
| Use DAVA instead of the mean-tuned baseline | VAE · AbstractDSprites | PIPE: 0.20±0.03 -> 0.35±0.03 | Improved |
| Use DAVA instead of the mean-tuned baseline | VAE · Mpi3d Toy | DCI: 0.23±0.01 -> 0.30±0.03 | Improved |
| Use DAVA instead of the mean-tuned baseline | VAE · Mpi3d Toy | FVAE: 0.50±0.02 -> 0.41±0.04 | Hurt |
| Add adaptive adversarial regularization to a VAE (ablation) | DAVA · multiple datasets | disentanglement performance | Improved |
| Add adaptive capacity control guided by discriminator accuracy (ablation) | DAVA · multiple datasets | disentanglement performance | Improved |
| Add reconstruction loss to PIPE (ablation) | PIPE metric · AbstractDSprites | correlation with supervised metrics | Hurt |
| Use instance normalization in the discriminator (ablation) | DAVA · multiple datasets | adversarial gradient signal | Improved |
| Replace absolute capacity deviation with its fourth power (ablation) | DAVA · multiple datasets | capacity control | Improved |
Closest papers by meaning
- Disentangling Disentanglement in Variational Autoencoders (ICML 2019 · similarity 0.88)
- Disentangling Factors of Variation with Cycle-Consistent Variational Auto-Encoders (ECCV 2018 · similarity 0.88)
- Demystifying Inductive Biases for (Beta-)VAE Based Architectures (ICML 2021 · similarity 0.88)
- Purify Unlearnable Examples via Rate-Constrained Variational Autoencoders (ICML 2024 · similarity 0.86)
- $\alpha$TC-VAE: On the relationship between Disentanglement and Diversity (ICLR 2024 · similarity 0.85)
- Tripod: Three Complementary Inductive Biases for Disentangled Representation Learning (ICML 2024 · 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.
