Journal of Cloud Computing

Advances, Systems and Applications

Table 3 Comparison of our proposed models and relevant baselines on MovieLens and Adressa

From: A split-federated learning and edge-cloud based efficient and privacy-preserving large-scale item recommendation model

Models	MovieLens			Adressa
Models	AUC	Recall@10	nDCG@10	AUC	Recall@10	nDCG@10
DSSM	76.15 ± 0.91	15.01 ± 0.41	35.84 ± 1.11	69.24 ± 0.18	24.36 ± 0.41	40.17 ± 1.20
FCF	70.56 ± 0.41	10.84 ± 0.08	25.27 ± 0.71	54.49 ± 0.03	17.61 ± 0.06	27.09 ± 1.72
FedMVMF	72.15 ± 0.35	12.03 ± 0.88	27.15 ± 1.31	60.73 ± 0.01	20.91 ± 0.10	30.13 ± 0.58
FedNewsRec	75.81 ± 0.51	15.42 ± 0.01	34.22 ± 0.38	70.95 ± 0.18	26.30 ± 0.01	42.68 ± 1.84
FedDSSM	75.93 ± 0.25	14.63 ± 0.68	33.41 ± 0.54	68.45 ± 1.03	23.11 ± 0.22	39.13 ± 0.48
SpFedRec	75.55 ± 1.21	14.48 ± 1.73	32.96 ± 1.61	68.10 ± 1.25	22.79 ± 1.03	38.65 ± 1.71
SpFedRec-SENet	76.93 ± 0.13	15.54 ± 1.11	36.21 ± 0.75	69.91 ± 0.22	25.43 ± 0.55	41.34 ± 1.01

Back to article page