Research map / Learning theory and algorithms / Prediction and decision losses
Optimally Controllable Perceptual Lossy Compression
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 |
|---|---|---|---|
| Interpolate MMSE and perfect-perception decoder outputs with increasing α (ablation) | Framework A and Framework B · KODAK low (0.13 bpp) and high (0.41 bpp) | PSNR: 36.89 -> 38.27; 38.71 -> 40.13 | Helped |
| Interpolate MMSE and perfect-perception decoder outputs with increasing α (ablation) | Framework A and Framework B · KODAK low (0.13 bpp) and high (0.41 bpp) | PI: 2.05 -> 3.32; 2.16 -> 2.73 | Hurt |
| Use Framework A at α=0 (perfect-perception decoder) instead of HiFiC | KODAK low (0.13 bpp) · KODAK low (0.13 bpp) | PI: 2.10 -> 2.05 | Helped |
| Use Framework A at α=0 (perfect-perception decoder) instead of HiFiC | KODAK high (0.41 bpp) · KODAK high (0.41 bpp) | PI: 2.24 -> 2.16 | Helped |
| Increase λ during Framework A training (ablation) | Framework A · KODAK | PSNR: 36.89 -> 38.27 | Helped |
| Decrease λ during Framework A training (ablation) | Framework A · KODAK | PI: 4.54 -> 1.98 | Helped |
| Add MMSE-decoder-distance supervision to conditional adversarial training | Perfect-perception decoder training · RGB images | perceptual quality | Helped |
| Use HiFiC instead of interpolation at α=0.5 | Framework A and Framework B · KODAK low and high bit rates | PI: 2.47 vs 2.10; 2.29 vs 2.24 | Helped |
Closest papers by meaning
- Controllable Distortion-Perception Tradeoff Through Latent Diffusion for Neural Image Compression (AAAI 2025 · similarity 0.88)
- Correcting Diffusion-Based Perceptual Image Compression with Privileged End-to-End Decoder (ICML 2024 · similarity 0.88)
- Optimal Neural Compressors for the Rate-Distortion-Perception Tradeoff (NeurIPS 2025 · similarity 0.88)
- Traversing Distortion-Perception Tradeoff using a Single Score-Based Generative Model (CVPR 2025 · similarity 0.87)
- Towards image compression with perfect realism at ultra-low bitrates (ICLR 2024 · similarity 0.87)
- Good, Cheap, and Fast: Overfitted Image Compression with Wasserstein Distortion (CVPR 2025 · 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.