Research map / Probabilistic and scientific modeling / Variational and generative sampling
Evaluating Lossy Compression Rates of Deep Generative Models
Sicong Huang, Alireza Makhzani, Yanshuai Cao, Roger Grosse
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 |
|---|---|---|---|
| Increase GAN code size (ablation) | GANs · MNIST and CIFAR-10 | rate-distortion tradeoff | Improved |
| Increase GAN depth (ablation) | GANs · MNIST | distortion | Improved |
| Increase GAN depth (ablation) | GANs · MNIST low-rate regime | distortion | No clear effect |
| Use different GAN variants at the same code size (ablation) | GANs · CIFAR-10 | rate-distortion curve | No clear effect |
| Increase VAE latent capacity (ablation) | VAEs · MNIST high-rate regime | distortion | Improved |
| Increase VAE latent capacity (ablation) | VAEs · MNIST low-rate regime | distortion | Hurt |
| Use AAE instead of GAN (ablation) | GANs · MNIST low-rate regime | distortion | No clear effect |
| Use AAE instead of GAN (ablation) | GANs · MNIST high-rate regime | distortion | Improved |
| Use AAE instead of VAE (ablation) | VAEs · MNIST low-rate regime | distortion | Improved |
| Use AAE instead of VAE (ablation) | VAEs · MNIST high-rate regime | distortion | Hurt |
| Replace the VAE prior with a mixture including a poor prior (ablation) | VAE-10 · MNIST low-rate regime | distortion | Hurt |
| Blur the VAE decoder output (ablation) | VAE · MNIST high-rate regime | distortion | Hurt |
Closest papers by meaning
- High-Fidelity Generative Image Compression (NeurIPS 2020 · similarity 0.88)
- Adaptive Density Estimation for Generative Models (NeurIPS 2019 · similarity 0.87)
- Assessing Generative Models via Precision and Recall (NeurIPS 2018 · similarity 0.87)
- Singular Value Scaling: Efficient Generative Model Compression via Pruned Weights Refinement (AAAI 2025 · similarity 0.86)
- Statistically Optimal Generative Modeling with Maximum Deviation from the Empirical Distribution (ICML 2024 · similarity 0.86)
- Unified Generation, Reconstruction, and Representation: Generalized Diffusion with Adaptive Latent Encoding-Decoding (ICML 2024 · similarity 0.86)
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.
