Table 2.

Mean NLL and standard error of the predictions. The entries in bold denote the best performing model(s).

Dataset	BLR	VI	FedAG${}_{\mathit{N}=0}^{(\mathit{t}=1)}$	FedAG${}_{\mathit{N}=1}^{(\mathit{t}=5)}$	DE
Boston	$3.07\pm 0.03$	$2.90\pm 0.07$	$3.02\pm 0.03$	$\mathbf{2.58}\pm \mathbf{0.06}$	$\mathbf{2.41}\pm \mathbf{0.25}$
Concrete	$3.78\pm 0.02$	$3.39\pm 0.02$	$3.76\pm 0.03$	$\mathbf{3.21}\pm \mathbf{0.04}$	$\mathbf{3.06}\pm \mathbf{0.18}$
Energy	$5.12\pm 0.05$	$2.39\pm 0.03$	$5.31\pm 0.06$	$2.07\pm 0.04$	$\mathbf{1.38}\pm \mathbf{0.22}$
Kin8nm	$1.17\pm 0.04$	$-0.90\pm 0.01$	$1.03\pm 0.04$	$-0.87\pm 0.01$	$-\mathbf{1.20}\pm \mathbf{0.02}$
Naval Propulsion	$-3.55\pm 0.02$	$-3.73\pm 0.12$	$-3.45\pm 0.01$	$-3.21\pm 0.01$	$-\mathbf{5.63}\pm \mathbf{0.05}$
Power Plant	$2.97\pm 0.01$	$2.89\pm 0.01$	$2.94\pm 0.01$	$2.92\pm 0.01$	$\mathbf{2.79}\pm \mathbf{0.04}$
Protein	$3.07\pm 0.00$	$2.99\pm 0.01$	$3.08\pm 0.00$	$2.95\pm 0.00$	$\mathbf{2.83}\pm \mathbf{0.04}$
Red Wine	$1.50\pm 0.07$	$\mathbf{0.98}\pm \mathbf{0.01}$	$\mathbf{1.01}\pm \mathbf{0.03}$	$\mathbf{0.99}\pm \mathbf{0.02}$	$\mathbf{0.94}\pm \mathbf{0.12}$
Yacht	$3.63\pm 0.05$	$3.44\pm 0.16$	$4.02\pm 0.07$	$1.92\pm 0.06$	$\mathbf{1.18}\pm \mathbf{0.21}$
Year Prediction	$3.73\pm \mathrm{NA}$	$3.86\pm \mathrm{NA}$	$3.72\pm \mathrm{NA}$	$3.66\pm \mathrm{NA}$	$\mathbf{3.35}\pm \mathbf{NA }$