Graph embedding ensemble methods based on the heterogeneous network for lncRNA-miRNA interaction prediction

BMC Genomics

Table 1 Parameter settings for proposed methods

Methods	Components	Parameters
Graph embedding methods	Representation vector	dimension θ: 120
	GraRep	k-step: 1
	DeepWalk	walk length t: 80, walk per vertex γ: 30, window size w: 30
	GAE	variational Autoencoder, hidden size β: 512
GEEL-PI	Random Forest	default parameters
GEEL-PI	Logistic Regression	L2 regulation with default parameters
GEEL-FI	DANN	hidden layers μ: 2, hidden size β: {240,120}
	Representation vector	embeddings ε: {GraRep, LE}
	Pair feature	dimension θ: 160
	Random Forest	estimators η: 2000

ISSN: 1471-2164