The Rate-Distortion-Perception Trade-Off with Algorithmic Realism

2025-06-22

International Symposium on Information Theory

AI Summary

This paper addresses the challenge of jointly optimizing perceptual quality (realism) and rate-distortion performance in lossy image compression at high compression ratios. We propose a novel perception-constrained framework that requires no substantial public randomness. Methodologically, we introduce a universal discriminator-driven realism constraint—applicable to either individual reconstructions or mini-batches—which formalizes perceptual quality as a verifiable discriminative boundary, eliminating reliance on shared random seeds or intricate priors. Leveraging high-rate compression theory and information-theoretic tools, our theoretical analysis proves that the optimal rate-distortion trade-off is asymptotically achievable under this constraint, with practical scalability to batch size. Our key contribution is bridging the gap between theoretical optimality and engineering feasibility: the framework significantly reduces implementation complexity while providing an analyzable, deployable paradigm for perception-driven compression.

Technology Category

• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Non-convex Optimization
• Constraint Satisfaction and Optimization: Constraint Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Realism constraints (or constraints on perceptual quality) have received considerable recent attention within the context of lossy compression, particularly of images. Theoretical studies of lossy compression indicate that high-rate common randomness between the compressor and the decompressor is a valuable resource for achieving realism. On the other hand, the utility of significant amounts of common randomness has not been noted in practice. We offer an explanation for this discrepancy by considering a realism constraint that requires satisfying a universal critic that inspects realizations of individual compressed reconstructions, or batches thereof. We characterize the optimal rate-distortion-perception trade-off under such a realism constraint, and show that it is asymptotically achievable without any common randomness, unless the batch size is impractically large.

Problem

• The paper explains why common randomness is not needed for realistic image compression
• It characterizes optimal rate-distortion trade-offs under universal realism constraints
• Shows high-rate common randomness becomes unnecessary unless batch sizes are impractical

Innovation

• Realism constraints require universal critic satisfaction
• Optimal rate-distortion trade-off characterized under realism
• Achievable without common randomness for practical batches