Table 1.

Model result comparison using 5 evaluation metrics: mean accuracy, class 0 (cat) accuracy, class 1 (dog) accuracy, balanced accuracy, and log loss.

Models	Total accuracy	Class 0 (Cat)	Class 1 (Dog)	Balanced accuracy	Log loss
Cleansed data
Baseline	98.44	98.62	98.26	98.44	0.061
${\mathbb{M}}^1$	98.42	98.49	98.35	98.42	0.045
${\mathbb{M}}^2$	98.62	98.13	99.11	98.62	0.038
Noisy data
Baseline	75.31	57.34	93.32	75.33	1.235
${\mathbb{M}}^1$	78.00	57.68	98.35	78.01	0.421
${\mathbb{M}}^2$	73.78	48.49	99.11	73.80	0.598