Research map / Probabilistic and scientific modeling / Variational and generative sampling
Relaxing Bijectivity Constraints with Continuously Indexed Normalising Flows
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 continuously indexed bijections instead of a single flow bijection | ResFlow (10 layers) · Power | average test log-likelihood (nats): -2.73 ± 0.03 -> 1.60 ± 0.21 | Helped |
| Use continuously indexed bijections instead of a single flow bijection | ResFlow (10 layers) · Gas | average test log-likelihood (nats): 4.16 ± 0.08 -> 12.12 ± 0.10 | Helped |
| Use continuously indexed bijections instead of a single flow bijection | ResFlow (10 layers) · Hepmass | average test log-likelihood (nats): -20.68 ± 0.02 -> -13.74 ± 0.03 | Helped |
| Use continuously indexed bijections instead of a single flow bijection | ResFlow (10 layers) · Miniboone | average test log-likelihood (nats): -14.2 ± 0.10 -> -8.10 ± 0.04 | Helped |
| Use continuously indexed bijections instead of a single flow bijection | ResFlow (10 layers) · BSDS300 | average test log-likelihood (nats): 123.51 ± 0.09 -> 160.50 ± 0.08 | Helped |
| Use continuously indexed bijections | MAF · Power | average test log-likelihood (nats): 0.19 ± 0.02 -> 0.48 ± 0.01 | Helped |
| Use continuously indexed bijections | MAF · Gas | average test log-likelihood (nats): 9.23 ± 0.07 -> 12.02 ± 0.10 | Helped |
| Use continuously indexed bijections | NSF · Power | average test log-likelihood (nats): 0.69 ± 0.00 -> 0.68 ± 0.01 | No clear effect |
| Use continuously indexed bijections | NSF · Gas | average test log-likelihood (nats): 13.01 ± 0.02 -> 12.94 ± 0.01; 13.01 ± 0.02 -> 13.08 ± 0.00 | Mixed |
| Use continuously indexed bijections | ResFlow · MNIST | average test bits per dimension: 1.074 -> 0.922 | Helped |
| Use continuously indexed bijections | ResFlow · CIFAR-10 | average test bits per dimension: 3.474 -> 3.334 | Helped |
| Use continuously indexed bijections | RealNVP · Fashion-MNIST | average test bits per dimension: 2.944 ± 0.003 -> 2.823 ± 0.003 | Helped |
Closest papers by meaning
- On the Universality of Volume-Preserving and Coupling-Based Normalizing Flows (ICML 2024 · similarity 0.86)
- Lifting Architectural Constraints of Injective Flows (ICLR 2024 · similarity 0.84)
- Asymptotically exact variational flows via involutive MCMC kernels (NeurIPS 2025 · similarity 0.84)
- Flow Matching from Viewpoint of Proximal Operators (NeurIPS 2026 · similarity 0.83)
- Physics-Constrained Flow Matching: Sampling Generative Models with Hard Constraints (NeurIPS 2025 · similarity 0.83)
- How to build a consistency model: Learning flow maps via self-distillation (NeurIPS 2025 · similarity 0.83)
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.