Table 3. Parameter settings for C-Atlas...

Table 3.

Parameter settings for C-Atlas and C-Atlas+

Method	SLIC parameters	Bagging parameters	SVM parameters
C-Atlas	$C o m p t = 0.001$ $n I t e r = 200$ $K = 100$ –1000	N/A	Kernel: linear $C = 1.0$ penalty = l2 $m = 1000$
C-Atlas+	$C o m p t = 0.05$ $n I t e r = 200$ $K = 100$ –1000	$N = 10$ bootstrap = yes random seed = fixed stratified sampling = yes $α = 0.8$	Kernel: linear $C = 1.0$ penalty = l2 $m = 1000$

Method	SLIC parameters	Bagging parameters	SVM parameters
C-Atlas	$C o m p t = 0.001$ $n I t e r = 200$ $K = 100$ –1000	N/A	Kernel: linear $C = 1.0$ penalty = l2 $m = 1000$
C-Atlas+	$C o m p t = 0.05$ $n I t e r = 200$ $K = 100$ –1000	$N = 10$ bootstrap = yes random seed = fixed stratified sampling = yes $α = 0.8$	Kernel: linear $C = 1.0$ penalty = l2 $m = 1000$

Note(s): Compt, compactness parameter; nIter, number of maskSLIC iterations; K, number of clusters; N, number of bagging estimators; $α$ ⁠, bagging aggregation weight; m, maximum number of SVM iterations; C, regularization parameter

Source(s): Authors’ own work

[ViewLarge]

Sharing Unavailable