Section 9, paragraph 3Conclusion: what the saved experiments do and do not show
Informal interpretation Motivation, interpretation or scope; not a formal claim.
1 Paper text
The saved floating-point experiments show a large supervised two-frame restoration advantage and perfect finite-test classification by a learned multi-head controller. The latter coexists with imperfect individual channel recovery. The restoration plateaus and checkpoint diagnostics also show that learning can proceed in stages after transition information is supplied. These observations do not certify the formal gate assumptions, worst-case mastery, or self-play discovery.
Generated from paper_jmlr.tex, lines 1280–1280.
2 Explain this
The third paragraph says what the saved experiments show: a large supervised two-frame restoration advantage, and perfect finite-test classification by a learned multi-head controller despite imperfect individual channels; learning can proceed in stages even after transition information is supplied. It also says what they do not show: the formal gate assumptions, worst-case mastery, or self-play discovery.
3 How this is formalised
A summary of the empirical records Section 7.2.5, “One-frame does not learn restoration; two-frame eventually does”, Section 7.3, paragraph 2 and Section 7.2.5, “Learning in stages: the plateau and subsequent improvement”.
| Paper anchor | sec:conclusion-p3 (website anchor; the manuscript has no label here) |
|---|---|
| Displayed as | Section 9, paragraph 3 |
| Claim | Conclusion: what the saved experiments do and do not show |
| Coverage category | Informal interpretation |
| Statement source | paper_jmlr.tex lines 1280–1280; SHA-256 of the normalised block source 1c97c652320cd6f4… |
| Depends on | Section 7.2.5, “One-frame does not learn restoration; two-frame eventually does”, Section 7.3, paragraph 2 |
| Explanations | explain/c-conclusion-experiments, formal/c-conclusion-experiments |
| Build evidence | fresh isolated rebuild, audit report |
4 Lean statement and source
No Lean statement corresponds to this claim. This is interpretation or motivation, not a formal mathematical claim.
Build evidence: fresh isolated rebuild of all 38 bundled modules with Lean (version 4.26.0, started 2026-10-10T15:21:56Z, finished 2026-10-10T15:27:00Z: 38/38 modules compiled, 1588 axiom reports, no errors. See Build for the commands and the independent audit of 2026-10-10 (dated evidence).