# GVSS-07 — Online Visual System Identification and Diagnostic Model Learning

## Core learned object

$$
\boxed{
\mathsf{DM}_\nu=(\widehat T_\nu,\widehat Q_\nu,\mathcal U_\nu,\mathsf{Prov}_\nu).
}
$$

## Core distinction

$$
\boxed{
\text{controlled/labeled identification}\neq\text{latent POMDP model learning}.
}
$$

## Core no-go

$$
\boxed{
\text{perfect self-label agreement}\not\Rightarrow\text{diagnostic correctness}.
}
$$

## Core control link

$$
\boxed{
TV(b^{a,y},\widehat b^{a,y})
\le
\frac{2\varepsilon_T+\varepsilon_Q}{\zeta}.
}
$$

## Package contents

- canonical paper;
- literature audit;
- roadmap;
- GVSS-08 handoff;
- theorem index;
- regression tests;
- validation report;
- checksums.