Research map / Probabilistic and scientific modeling / Variational and generative sampling
ACFlow: Flow Models for Arbitrary Conditional Likelihoods
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 ACFlow instead of PixelCNN | arbitrary conditional image likelihood modeling · MNIST, Omniglot, and CelebA | NLL: MNIST: 313.93$\pm$3.09 -> 271.99$\pm$3.10; Omniglot: 160.20$\pm$1.97 -> 159.32$\pm$1.25; CelebA: 8806.02$\pm$34.95 -> 83 | Helped |
| use ACFlow instead of VAEAC | arbitrary conditional image likelihood modeling · MNIST, Omniglot, and CelebA | NLL: MNIST: 1034.86$\pm$6.05 -> 271.99$\pm$3.10; Omniglot: 1014.62$\pm$2.53 -> 159.32$\pm$1.25; CelebA: 17638.83$\pm$15.21 -> | Helped |
| train ACFlow on conditional tasks and evaluate the joint distribution | ACFlow · MNIST and CelebA joint likelihood | bits per dimension: MNIST: 0.87 -> 0.86; CelebA: 3.02 -> 2.85 | Helped |
| train ACFlow on conditional tasks and evaluate the joint distribution | ACFlow · Omniglot and CIFAR10 joint likelihood | bits per dimension: Omniglot: 0.52 -> 0.52; CIFAR10: 3.49 -> 3.50 | Mixed |
| add the best-guess auxiliary penalty (ablation) | ACFlow · MNIST, Omniglot, and CelebA inpainting | PSNR: MNIST: 17.349$\pm$0.018 -> 20.828$\pm$0.031; Omniglot: 15.572$\pm$0.031 -> 18.838$\pm$0.009; CelebA: 22.393$\pm$0.040 -> | Helped |
| add the best-guess auxiliary penalty (ablation) | ACFlow · MNIST, Omniglot, and CelebA imputation distributions | PRD | No clear effect |
| add the best-guess auxiliary penalty (ablation) | ACFlow · UCI feature imputation | NRMSE | Helped |
| use ACFlow instead of DCSPN | conditional face imputation · Olivetti Face left and bottom masks | MSE: left: 455 -> 415; bottom: 503 -> 434 | Helped |
Closest papers by meaning
- A Likelihood Based Approach to Distribution Regression Using Conditional Deep Generative Models (ICML 2025 · similarity 0.85)
- DiffPuter: Empowering Diffusion Models for Missing Data Imputation (ICLR 2025 · similarity 0.85)
- Multivariate Latent Recalibration for Conditional Normalizing Flows (NeurIPS 2025 · similarity 0.85)
- CAR-Flow: Condition-Aware Reparameterization Aligns Source and Target for Better Flow Matching (NeurIPS 2025 · similarity 0.85)
- Generative Modeling under Non-Monotone MAR Missingness via Approximate Wasserstein Gradient Flows (NeurIPS 2026 · similarity 0.84)
- Rethinking the Diffusion Models for Missing Data Imputation: A Gradient Flow Perspective (NeurIPS 2024 · similarity 0.84)
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.