Table 1.

The performance of the proposed method with different values for the mixing parameter p on the DREAM3 problems.

		Ecoli1	Ecoli2	Yeast1	Yeast2	Yeast3
		AVG	AVG	AVG	AVG	AVG
		± STD	± STD	± STD	± STD	± STD
Proposed method	(p = 1.0)	0.41910	0.54478	0.50084	0.39486	0.31297
		±0.00390	±0.00586	±0.00287	±0.00344	±0.00224
Proposed method	(p = 0.9)	0.42143	0.54539	0.50594	0.40047	0.32093
		±0.00378	±0.00563	±0.00289	±0.00370	±0.00221
Proposed method	(p = 0.8)	0.42307	0.54607	0.50825	0.40261	0.32234
		±0.00380	±0.00530	±0.00301	±0.00371	±0.00245
Proposed method	(p = 0.7)	0.42422	0.54674	0.50984	0.40384	0.32256
		±0.00395	±0.00523	±0.00314	±0.00377	±0.00263
Proposed method	(p = 0.6)	0.42493	0.54736	0.51055	0.40446	0.32223
		±0.00420	±0.00537	±0.00347	±0.00384	±0.00272
Proposed method	(p = 0.5)	0.42532	0.54772	0.51060	0.40419	0.32151
		±0.00437	±0.00542	±0.00395	±0.00393	±0.00278
Proposed method	(p = 0.4)	0.42520	0.54767	0.50996	0.40321	0.32054
		±0.00465	±0.00579	±0.00446	±0.00393	±0.00291
Proposed method	(p = 0.3)	0.42410	0.54689	0.50856	0.40179	0.31975
		±0.00533	±0.00658	±0.00472	±0.00378	±0.00297
Proposed method	(p = 0.2)	0.42216	0.54514	0.50655	0.40059	0.31922
		±0.00663	±0.00766	±0.00489	±0.00386	±0.00293
Proposed method	(p = 0.1)	0.42034	0.54344	0.50332	0.40046	0.31881
		±0.00757	±0.00813	±0.00507	±0.00390	±0.00265
Proposed method	(p = 0.0)	0.07094	0.07486	0.09892	0.13139	0.13949
		±0.00195	±0.00206	±0.00252	±0.00232	±0.00284
Random-forest-based inference method		0.41918	0.54477	0.50083	0.39482	0.31291
(Kimura et al., 2019)		±0.00388	±0.00586	±0.00285	±0.00344	±0.00223

The performance of the random-forest-based inference method (Kimura et al., 2019) is also shown. AVG and STD represent the averaged AURPC and its standard deviation, respectively.