Evidence item

The primary study directly reports the 2.97-dB RMSE, 24.0% baseline improvement, and improved performance on unseen corridor geometries. DIRECT_TEXT

Authenticity UNVERIFIED · extraction UNVERIFIED · independence group none (unreviewed) · created seq 17156

Source: Shadow fading prediction at 18 GHz through physics guided learning in vegetative corridors · locator {} · excerpt hash sha256:766182305631d6a5b6e64bef0a47a8a3cf6b0abd106873969d04e9b9d3c235f0

Exact excerpt

Under leave-one-scenario-out cross-validation, the hybrid model achieves RMSE = 2.97 dB, a 24.0% improvement over the CI baseline. Crucially, while pure ensemble methods ... exhibit performance degradation up to 1.05 dB when extrapolating to unseen geometric configurations, the hybrid architecture demonstrates a 0.35 dB improvement, achieving better accuracy on novel corridor configurations than on interpolated data.

Linked claims

Claim	Direction	Relevance	Steps	Counted
dd2af2b0	SUPPORT	DIRECT	0	yes

Audits

None.