Table 5.

Statistical comparison of methods across metrics (mean ± 95% CI, p-values and effect sizes for Ours.Variational vs. best SOTA, n = 5).

Metric	Ours.Variational	Best SOTA (FedEntropy)	p-value	Cohen’s d
Accuracy (%)	89.6 ± 1.8	85.3 ± 2.0	0.021	0.91
Compression Ratio	2.42 ± 0.28	1.72 ± 0.33	0.036	0.85
KL-Divergence	0.106 ± 0.024	0.157 ± 0.030	0.049	0.76
Entropy Reduction	2.05 ± 0.25	1.23 ± 0.27	0.018	0.93
Latency (ms)	24.9 ± 3.8	28.1 ± 4.1	0.061	0.67
Memory (KB)	44.6 ± 5.1	37.5 ± 4.9	0.072	0.63