Enhancing energy consumption prediction in smart homes: a convergence-aware federated transfer learning approach | Science and Technology for Energy Transition (STET)

Emerging Advances in Hybrid Renewable Energy Systems and Integration

Open Access

Table 4

Comparison of proposed FedDNN and FedLSTM models with existing models in terms of mean squared error.

Algorithm	C1	C2	C3	C4	C5	C6	C7	C8	Average
LinearRegression	134.12 ± 2.50	154.23 ± 2.43	161.34 ± 1.39	144.80 ± 1.49	144.11 ± 1.47	173.59 ± 2.74	133.68 ± 2.12	154.94 ± 2.16	150.10 ± 2.04
LGBMRegressor	26.43 ± 1.29	22.54 ± 2.81	23.15 ± 2.05	16.44 ± 2.20	16.37 ± 1.06	24.34 ± 1.60	24.56 ± 2.61	24.56 ± 2.76	22.30 ± 2.05
XGBRegressor	27.16 ± 2.50	25.22 ± 2.26	24.45 ± 1.25	16.43 ± 2.42	15.29 ± 1.53	23.45 ± 2.33	23.43 ± 1.15	23.43 ± 1.61	22.36 ± 1.88
CatBoostRegressor	26.64 ± 1.10	24.54 ± 2.14	25.65 ± 2.02	17.30 ± 2.74	13.32 ± 1.67	22.65 ± 1.73	24.35 ± 2.92	24.23 ± 2.24	22.33 ± 2.07
SGDRegressor	137.25 ± 1.28	153.23 ± 2.73	164.13 ± 2.74	146.73 ± 2.69	143.12 ± 2.39	171.23 ± 1.91	133.34 ± 1.37	154.45 ± 2.17	150.44 ± 2.16
KernelRidge	143.64 ± 2.95	165.75 ± 1.12	171.56 ± 1.66	152.51 ± 1.47	151.74 ± 2.15	182.43 ± 2.51	142.54 ± 2.37	165.56 ± 2.23	159.47 ± 2.06
BayesianRidge	134.17 ± 1.13	151.34 ± 2.35	163.53 ± 1.86	143.43 ± 2.65	144.11 ± 1.84	171.45 ± 2.95	132.34 ± 2.97	153.34 ± 2.39	149.21 ± 2.27
GradientBoostingRegressor	33.85 ± 2.54	28.62 ± 1.79	31.72 ± 2.81	22.37 ± 2.80	23.34 ± 2.37	34.34 ± 1.67	31.34 ± 2.06	32.65 ± 2.83	29.78 ± 2.36
SVR	135.34 ± 4.69	156.35 ± 2.86	165.15 ± 2.83	141.54 ± 1.58	142.12 ± 2.97	171.34 ± 1.58	133.45 ± 1.79	156.58 ± 2.49	150.23 ± 2.60
DNN	24.35 ± 2.01	24.56 ± 2.00	24.67 ± 2.97	16.23 ± 2.96	15.57 ± 1.73	25.07 ± 1.62	23.52 ± 1.00	23.53 ± 1.21	22.19 ± 1.94
FedDNN-full	23.62 ± 2.64	22.42 ± 1.43	23.73 ± 2.69	14.34 ± 2.60	15.24 ± 1.49	24.67 ± 1.48	22.87 ± 2.01	22.72 ± 2.50	21.20 ± 2.11
FedDNN-par	21.58 ± 1.61	17.15 ± 2.75	22.25 ± 1.46	11.24 ± 1.19	12.64 ± 1.69	23.96 ± 2.44	22.30 ± 2.75	20.32 ± 1.63	18.93 ± 1.94
LSTM	31.15 ± 2.61	22.93 ± 1.13	25.14 ± 1.24	16.45 ± 2.77	15.52 ± 2.30	24.46 ± 1.60	23.15 ± 2.83	23.37 ± 2.32	22.77 ± 2.10
FedLSTM-full	28.64 ± 2.29	22.56 ± 1.78	24.56 ± 1.46	14.34 ± 1.91	14.13 ± 1.13	22.58 ± 2.09	22.75 ± 1.75	22.67 ± 2.14	21.53 ± 1.82
FedLSTM-par	19.23 ± 2.75	18.15 ± 1.25	23.65 ± 2.70	12.22 ± 1.54	13.34 ± 1.71	19.49 ± 1.56	21.32 ± 2.90	19.15 ± 1.84	18.32 ± 2.03

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.