Alternating Direction Multiplier Method for Matrix -Norm Optimization in Multitask Feature Learning Problems

Table 2

Numerical results for the real problem.

Demo	b, X	ADMM-BB			IADM-MFL
	(m, n, t)	ITER	TIME	RelChg	ITER	TIME	RelChg
20%	(80, 500, 10)	178	1.34	9.95e − 04	365	2.47	9.97e − 04
30%	(120, 500, 10)	282	2.23	9.98e − 04	412	2.54	9.93e − 04
40%	(160, 500, 10)	266	2.33	9.96e − 04	415	2.73	9.92e − 04
50%	(200, 500 10)	300	2.75	9.95e − 04	404	2.80	9.73e − 04
60%	(240, 500, 10)	306	2.86	9.99e − 04	427	3.13	9.95e − 04
70%	(280, 500, 10)	349	3.30	9.99e − 04	460	3.47	9.97e − 04