Multiview Embedding with Partial Labels to Recognize Users of Devices Based on Unified Transformer

Table 5

The main parameters of MVEPL.

Type	Parameters	Value
Word2Vec	vector_size	128
	Windows	2
	sg	0
	hs	0
	Epochs	10
	min_count	1
One-hot	vector_size	128
Label encoder	vector_size	128
Multiviews embeddings (using GraphSAGE)	Aggregator	Mean
	Node vector_size	256
	vector_size	128
	view_number	4
Unified transformer	Hidden size	256
	Heads	4
	vector_size	128
	Transformer	1
	Block number
	Cotransformer	1
	Block number
Training	Epochs	20
	batch_size	64
	drop_out	0.25
	learning_rate	0.001
	masked_label_ratio	0.2