Table 4.

Probability Calibration Performance for MLP and VGG16 Models Using Brier Score and Expected Calibration Error

Model/Variant	Brier Score	ECE	Group
MLP + Isotonic	0.0434	0.0486	MLP — 10-Fold Cross-Validated Probabilities
MLP (Uncalibrated)	0.0452	0.0535	MLP — 10-Fold Cross-Validated Probabilities
MLP + Platt (Sig.)	0.0481	0.0801	MLP — 10-Fold Cross-Validated Probabilities
VGG16 + Isotonic	0.1142	0	CNN Models — Validation Set Probabilities
VGG16 + Platt	0.123	0.0655	CNN Models — Validation Set Probabilities
VGG16	0.1338	0.0944	CNN
VGG16 (Uncalibrated)	0.1338	0.0944	CNN Models — Validation Set Probabilities