Researchers have introduced a new framework for understanding signal compression, moving beyond traditional rate-distortion theory. This new approach, termed synonymous variational inference (SVI), reformulates perceptual reconstruction by focusing on recovering any acceptable sample within an ideal "synonymous set" rather than a specific source sample. The framework establishes a consistency principle between synonymity and perception, theoretically aligning semantic information identification with perceptual optimization. This work clarifies the theoretical origins of distributional divergence in rate-distortion-perception models and suggests potential benefits for synonymous source coding. AI
RANK_REASON Academic paper published on arXiv detailing a new theoretical framework for signal compression. [lever_c_demoted from research: ic=1 ai=0.4]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
