Sparse and smooth canonical correlation analysis through rank-1 matrix approximation

EURASIP Journal on Advances in Signal Processing

Table 2 Simulation results part 1

		θ _x (rad)	θ _y (rad)	θ _x (rad)	θ _y (rad)	θ _x (rad)	θ _y (rad)
	Method	N=50		N=100		N=200
Scenario 1:	CCA	0.5395	0.5033	0.3468	0.3475	0.2273	0.2388
	LS CCA	0.4161	0.3697	0.2649	0.2650	0.1784	0.1872
	CCA LB	0.5172	0.5151	0.3310	0.3341	0.2250	0.2228
	PMD	0.2203	0.2420	0.0908	0.0506	0.0207	0.0175
	Algorithm 2	0.5074	0.5189	0.3123	0.3140	0.2225	0.2202
	Algorithm 3	0.2011	0.2191	0.0491	0.0273	0.0044	0.0057
Scenario 2:	CCA	0.5091	0.6682	0.3108	0.4123	0.2089	0.2771
	LS CCA	0.3481	0.5083	0.2285	0.3247	0.1605	0.2182
	CCA LB	0.3000	0.3761	0.0227	0.0228	0.0008	0.0009
	PMD	0.2061	0.3068	0.0230	0.0706	0.0043	0.0443
	Algorithm 2	0.5064	0.6462	0.3062	0.4111	0.2061	0.2792
	Algorithm 3	0.1162	0.1508	0.0012	0.0015	0.0001	0.0001
Scenario 3:	CCA	0.8699	1.0281	0.6800	0.8254	0.4823	0.6009
	LS CCA	0.6398	0.8314	0.4608	0.6139	0.3116	0.4348
	CCA LB	0.8681	1.0285	0.6575	0.8122	0.3859	0.4938
	PMD	0.7690	0.9080	0.5382	0.6736	0.2736	0.4811
	Algorithm 2	0.8465	0.9876	0.6654	0.8078	0.4345	0.5839
	Algorithm 3	0.3424	0.4571	0.0393	0.0628	0.0001	0.0016