Using Proximity Graph Cut for Fast and Robust Instance-Based Classification in Large Datasets

Table 1

Comparison of classification accuracy for different proximity graphs against SVM classifier with polynomial kernel.

Dataset	Accuracy (%)
	SVM	kNN	RNG	MST	Gabriel	NSW
Dermatology (k = 10)	96.9	98.6	95.6	86.7	65.4	95.65
Isolet (k = 7)	96.0	91.9	88.1	N/A	N/A	88.5
Image Segmentation (k = 1)	92.9	90.3	88.8	76.7	82.1	87.5

refers to the best kNN hyperparameter. NSW is our choice. Bold shows our study implementations.