Research map / Probabilistic and scientific modeling / Variational and generative sampling
Diffusion bridges vector quantized variational autoencoders
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 |
|---|---|---|---|
| Replace PixelCNN prior with OU diffusion bridge prior over continuous latents (T=1000) | VQ-VAE latent prior · miniImageNet | NLL (latent inpainting): 1.00 -> 0.94 | Helped |
| Replace PixelCNN prior with OU diffusion bridge prior | VQ-VAE latent prior · miniImageNet | FID: 98 -> 99 | No clear effect |
| Replace PixelCNN prior with OU diffusion bridge prior | VQ-VAE latent prior · miniImageNet | sampling time per sample: 10.6s -> 1.7s | Helped |
| Replace PixelCNN prior with OU diffusion bridge prior | VQ-VAE latent prior · CIFAR10 | NLL (latent inpainting): 1.41 -> 1.33 | Helped |
| Replace PixelCNN prior with OU diffusion bridge prior | VQ-VAE latent prior · CIFAR10 | FID: 109 -> 104 | Helped |
| Replace PixelCNN prior with OU diffusion bridge prior | VQ-VAE latent prior · CIFAR10 | sampling time per sample: 0.21s -> 0.05s | Helped |
| End-to-end training of VQ-VAE with diffusion prior (vs. fixed VQ-VAE) (ablation) | diffusion bridge prior · CIFAR10 | FID: 104 -> 92 | Helped |
| Replace PixelCNN prior with diffusion prior on latent KL | VQ-VAE latent prior · Latent KL (feature map) | KL: 0.809 -> 0.713 | Helped |
Closest papers by meaning
- Neural Discrete Representation Learning (NeurIPS 2017 · similarity 0.91)
- Generating Diverse High-Fidelity Images with VQ-VAE-2 (NeurIPS 2019 · similarity 0.91)
- Restructuring Vector Quantization with the Rotation Trick (ICLR 2025 · similarity 0.90)
- Hierarchical Quantized Autoencoders (NeurIPS 2020 · similarity 0.90)
- Generalization in VAE and Diffusion Models: A Unified Information-Theoretic Analysis (ICLR 2025 · similarity 0.89)
- DiffEnc: Variational Diffusion with a Learned Encoder (ICLR 2024 · similarity 0.89)
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.