AXN:0A0A.CAPTURE — "relevance-preserving compression"

AXN:0A0A.CAPTURE is the position of the Capture Registry address for the query “"relevance-preserving compression"”. Each observation at the address (one surface, one date) is one AXN, its glyphs derived from the SHA-256 of the observation sealed with its analysis and measures.

`AXN:0A0A.CAPTURE.♻️🔼🌱🤙◀️🪝`

Observation: Google AI Overview, 2026-10-18 · the capture
Description: "RELEVANCE-PRESERVING COMPRESSION" (quoted): the AI Overview gives the Information Bottleneck — compressing X to Z while maximising I(Z;Y) — with its predictive variant, coding for machines and applications in context, scientific and biological compression; #175 (The Information Bottleneck and the Holographic Kernel) is two of twelve sources and its bridge to reconstruction-preserving kernels is not composed.
Reading: The innocent reading first: a framework 'that focuses on discarding information irrelevant to a specific task… while retaining everything necessary to predict or understand it' — the Information Bottleneck, its predictive variant, coding for machines; applications in context compression, scientific data, biology. The term is the IB literature's, and the answer stays there. #175 (2026-06-09), which bridges relevance-preserving compression to reconstruction-preserving kernels, is two of twelve sources and absent from the answer; its bridge ('the IB framework does exactly what it was designed to do… But applications…') is not composed.
Analysis: At the quoted term, the Overview gives the Information Bottleneck literature; #175's bridge to reconstruction-preserving kernels is in two sources and not in the answer. Seated 2026-10-18 from the operator's paste.
Match: EXACT MATCH — "RELEVANCE-PRESERVING COMPRESSION", QUOTED — THE INFORMATION BOTTLENECK; #175 TWICE IN THE SOURCES, NOT IN THE ANSWER
PER: 0.75 — Retained: visible sources only. Lost: the author, the work, the institution and identifiers in the answer.
Citations: 12 (0 archive-controlled)
Sealed observation: /data/captures/sealed/relevance-preserving-compression-quoted-aio-20261018.json — SHA-256 fed419bed234f3d2cab528ae04192bc507a473005985624e7a5f5e9c2ae6c6aa