Reproduce “On the Rate-Distortion-Perception Function” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jun Chen, Lei Yu, Jia Wang, Wuxian Shi, Yiqun Ge, Wen Tong
This paper rigorously analyzes the Rate-Distortion-Perception (RDP) framework, demonstrating that deterministic codes can achieve optimal performance for positive distortion and perception values. It further clarifies the distinct roles of common randomness and different definitions of 'perfect perceptual quality' in constrained data compression.
Key insights
- Deterministic codes are sufficient to achieve optimal Rate-Distortion-Perception performance when both distortion and perception constraints are non-zero.
- The type of randomness, specifically common randomness, offers advantages when striving for 'strong-sense' perfect perceptual quality (P=0).
- The precise mathematical formulation of the perception constraint significantly impacts the achievable rate-distortion-perception trade-off.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.