Condense targeted analysis tables

manuscript/targeted/main.tex

@@ -327,8 +327,8 @@ Completed blocks + observed tail & 70.85 & 59.17 & -- \\
 
 \fi
 \begin{table}[t]
-\caption{Pending paired $\G\rightarrow\G+\A$ transfer (unmeasured).
-EC is EasyCom; MiniCPM first needs readout/refresh/history audits.}
+\caption{Pending paired $\G\rightarrow\G+\A$ comparison on EasyCom
+across Qwen2.5-Omni backbone sizes (unmeasured).}
 \label{tab:backbone}
 \centering\normalsize
 \setlength{\tabcolsep}{3pt}
@@ -336,10 +336,8 @@ EC is EasyCom; MiniCPM first needs readout/refresh/history audits.}
 \toprule
 Backbone & $\G$ & $\G+\A$ & $\Delta$ \\
 \midrule
-EC: Omni 3B & -- & -- & -- \\
-EC: Omni 7B & -- & -- & -- \\
-EC: MiniCPM-o 4.5 & -- & -- & -- \\
-Ego4D: Omni 3B & -- & -- & -- \\
+Omni 3B & -- & -- & -- \\
+Omni 7B & -- & -- & -- \\
 \bottomrule
 \end{tabular}
 \end{table}
@@ -508,14 +506,16 @@ shared GRU, while head capacity and training batch differ; raw boundary
 metrics and repeat seeds are unavailable. Thus the +0.66/+0.43-point
 difference does not isolate visual motion or floor behavior.
 
-The analysis-only audit in Table~\ref{tab:oracle} substitutes two seconds
+The lower block of Table~\ref{tab:visual} is an analysis-only audit that
+substitutes two seconds
 of transcript-derived, strictly past visible-person activity on the 5-Hz
-grid. It tests visibility alone, per-person histories, shuffled identities,
-a one-second delay, and OR-pooled activity. These privileged labels are
+grid. The displayed conditions test visibility alone, a one-second delay,
+and OR-pooled activity. These privileged labels are
 neither deployable visual features nor a performance upper bound.
-Aggregate visible
-activity improves over the audio baseline by 4.826 points at 0.2~s and 1.775
-mean-mAP points.  Visibility alone changes mean mAP by only +0.040, and
+The three-seed audio baseline is 81.87\% at 0.2~s and 65.18\% mean mAP;
+aggregate visible activity improves on it by 4.826 and 1.775 points,
+respectively.
+Visibility alone changes mean mAP by only +0.040, and
 one-second-delayed activity by +0.164.  Thus availability or extra parameters
 cannot explain the oracle gain.  Per-person activity exceeds shuffled identity
 by only 0.200 mean-mAP points and is itself 0.291 below the aggregate

manuscript/targeted/tables_compact.tex

@@ -31,43 +31,27 @@ Omni $\G+\A+\mathrm{SVA}$ & -- & -- & -- & -- & -- & -- & -- & -- & -- & -- & --
 \end{table*}
 
 \begin{table}[t]
-\caption{EasyCom visual-encoding screens (\%). Branch names are defined
-in Sec.~\ref{sec:method}. Available rows are exploratory and not
-capacity- or training-matched.}
+\caption{EasyCom local-vision and information-audit screens (\%). The
+upper rows are exploratory encoder-based runs; the lower rows use
+non-deployable past activity (three-seed mean $\pm$ SD). Their training
+protocols differ, so cross-section differences are not matched effects.
+Branch names are defined in Sec.~\ref{sec:method}.}
 \label{tab:visual}
 \centering\ninept
 \renewcommand{\arraystretch}{0.92}
-\setlength{\tabcolsep}{4pt}
+\setlength{\tabcolsep}{3pt}
 \begin{tabular}{@{}lcc@{}}
 \toprule
-Visual branch on $\G+\A$ & mAP$_{.2}$ & Mean \\
+Condition & mAP$_{.2}$ & Mean \\
 \midrule
-None & 81.72 & 65.11 \\
-Pool & 82.03 & 65.40 \\
-SVA & -- & -- \\
-TSVA & 82.38 & 65.54 \\
-\bottomrule
-\end{tabular}
-\end{table}
-
-\begin{table}[t]
-\caption{EasyCom privileged past-activity audit (\%). Three-seed mean
-$\pm$ sample SD. The extra activity cues are not deployable inputs; W is
-wearer AP at 0.2~s.}
-\label{tab:oracle}
-\centering\ninept
-\renewcommand{\arraystretch}{0.92}
-\setlength{\tabcolsep}{3pt}
-\begin{tabular}{@{}lccc@{}}
-\toprule
-Extra condition & mAP$_{.2}$ & W$_{.2}$ & Mean \\
+$\G+\A$ & 81.72 & 65.11 \\
+$\G+\A+$Pool & 82.03 & 65.40 \\
+$\G+\A+$SVA & -- & -- \\
+$\G+\A+$TSVA & 82.38 & 65.54 \\
 \midrule
-Audio baseline & $81.87{\pm}.05$ & $85.26{\pm}.11$ & $65.18{\pm}.06$ \\
-Visibility only & $81.95{\pm}.10$ & $85.20{\pm}.14$ & $65.22{\pm}.05$ \\
-Delayed 1 s & $81.95{\pm}.09$ & $85.27{\pm}.14$ & $65.34{\pm}.05$ \\
-Shuffled IDs & $85.92{\pm}.09$ & $85.58{\pm}.25$ & $66.46{\pm}.09$ \\
-Per-person & $86.27{\pm}.18$ & $85.61{\pm}.22$ & $66.66{\pm}.14$ \\
-Visible OR & $\mathbf{86.70{\pm}.09}$ & $\mathbf{85.84{\pm}.15}$ & $\mathbf{66.95{\pm}.02}$ \\
+Visibility only & $81.95{\pm}.10$ & $65.22{\pm}.05$ \\
+Delayed 1 s & $81.95{\pm}.09$ & $65.34{\pm}.05$ \\
+Visible OR & $\mathbf{86.70{\pm}.09}$ & $\mathbf{66.95{\pm}.02}$ \\
 \bottomrule
 \end{tabular}
 \end{table}