Research map / Learning theory and algorithms / Sample complexity
Faster Relative Entropy Coding with Greedy Rejection Coding
Gergely Flamich, Stratis Markou, José Miguel Hernández-Lobato
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 |
|---|---|---|---|
| Encode GRCD heap indices with ζ coding instead of δ coding (ablation) | VAE compression pipeline with ELBO training · MNIST test, 20 latent dimensions | total BPP: 1.482 ± 0.004 -> 1.472 ± 0.004 | Improved |
| Encode GRCD heap indices with ζ coding instead of δ coding (ablation) | VAE compression pipeline with ELBO training · MNIST test, 50 latent dimensions | total BPP: 1.530 ± 0.003 -> 1.511 ± 0.003 | Improved |
| Encode GRCD heap indices with ζ coding instead of δ coding (ablation) | VAE compression pipeline with ELBO training · MNIST test, 100 latent dimensions | total BPP: 1.600 ± 0.003 -> 1.523 ± 0.003 | Improved |
| Encode GRCD heap indices with ζ coding instead of δ coding (ablation) | VAE compression pipeline with modified ELBO training · MNIST test, 20 latent dimensions | total BPP: 1.478 ± 0.004 -> 1.470 ± 0.004 | Improved |
| Encode GRCD heap indices with ζ coding instead of δ coding (ablation) | VAE compression pipeline with modified ELBO training · MNIST test, 50 latent dimensions | total BPP: 1.514 ± 0.003 -> 1.484 ± 0.003 | Improved |
| Encode GRCD heap indices with ζ coding instead of δ coding (ablation) | VAE compression pipeline with modified ELBO training · MNIST test, 100 latent dimensions | total BPP: 1.579 ± 0.003 -> 1.485 ± 0.003 | Improved |
| Train with the modified ELBO including logarithmic overhead instead of regular ELBO (ablation) | VAE compression pipeline using ζ coding · MNIST test, 20 latent dimensions | total BPP: 1.472 ± 0.004 -> 1.470 ± 0.004 | Improved |
| Train with the modified ELBO including logarithmic overhead instead of regular ELBO (ablation) | VAE compression pipeline using ζ coding · MNIST test, 50 latent dimensions | total BPP: 1.511 ± 0.003 -> 1.484 ± 0.003 | Improved |
| Train with the modified ELBO including logarithmic overhead instead of regular ELBO (ablation) | VAE compression pipeline using ζ coding · MNIST test, 100 latent dimensions | total BPP: 1.523 ± 0.003 -> 1.485 ± 0.003 | Improved |
| Train with the modified ELBO including logarithmic overhead instead of regular ELBO (ablation) | VAE compression pipeline using δ coding · MNIST test, 20 latent dimensions | total BPP: 1.482 ± 0.004 -> 1.478 ± 0.004 | Improved |
| Train with the modified ELBO including logarithmic overhead instead of regular ELBO (ablation) | VAE compression pipeline using δ coding · MNIST test, 50 latent dimensions | total BPP: 1.530 ± 0.003 -> 1.514 ± 0.003 | Improved |
| Train with the modified ELBO including logarithmic overhead instead of regular ELBO (ablation) | VAE compression pipeline using δ coding · MNIST test, 100 latent dimensions | total BPP: 1.600 ± 0.003 -> 1.579 ± 0.003 | Improved |
Closest papers by meaning
- Accelerating Relative Entropy Coding with Space Partitioning (NeurIPS 2024 · similarity 0.91)
- Universal Sample Coding (NeurIPS 2024 · similarity 0.87)
- Channel Simulation and Distributed Compression with Ensemble Rejection Sampling (NeurIPS 2025 · similarity 0.87)
- Minimum Entropy Coupling with Bottleneck (NeurIPS 2024 · similarity 0.87)
- Entropy Coding of Unordered Data Structures (ICLR 2024 · similarity 0.85)
- Approximate Gradient Coding for Distributed Learning with Heterogeneous Stragglers (NeurIPS 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.
