Table 2.

Summaries of machine learning in haptic sensing device

Category	Reference	Algorithms	Task	Application	Accuracy
Piezoresistive	[26]	CNN	Classification and regression	Body motion recognition	63.76%
	[68]	MLP	Classification	Hand trajectory	98.8%
	[93]	SVM	Classification	Hand gesture identification	92.67%
	[94]	RBFNN	Classification	Hand gesture identification	93.3%
	[100]	CNN	Classification, clustering, and dimension reduction	Gesture recognition	96.7%
	[104]	RSL	Classification and regression	Human action recognition	96.2%
	[107]	CNN	Classification, cluster, and dimension reduction	Object identification	96.9%
	[116]	XGB	Classification	Material recognition	93.3%
	[122]	CNN	Classification, regression, clustering, and dimension reduction	Object identification	89.4%
Piezoelectric	[119]	-	Classification	Texture recognition	99.1%
Triboelectric	[88]	1D-CNN	Classification	Body motion recognition	96.67%
	[101]	SVM	Classification and feature extraction	Hand gesture identification	98.63%
	[83]	1D-CNN	Classification	Body movement trajectory	85.67%
	[86]	ANN	Classification	Gait signal identification	-
	[89]	CNN	Classification	Human action recognition	91.47%
	[106]	1D CNN	Classification	Body motion recognition	97.5%
	[112]	LDA	Classification	Object identification	96.8%
	[113]	VGG	Classification	Material recognition	96.62%
Hybrid (piezoelectric and triboelectric)	[96]	CNN	Classification	Hand gesture identification	94.16%
	[102]	LDA	Classification	Human action recognition	92.6%
Magnetic	[82]	-	Classification, clustering, and dimension reduction	Human action recognition	-
Electromyography	[103]	AM	Dimension reduction	Hand gesture identification	97.12%
Photoelectric	[117]	ANN	Classification	Texture recognition	94.1%
Thermal conductive	[121]	MLP	Classification	Object recognition	94%