An Inexact Penalty Decomposition Method for Sparse Optimization

Figure 3

Averaged results of the penalty decomposition methods for the compressed sensing problem with different sampling numbers on 100 instances: (a) CPU time over sampling number; (b) recovered rate over sampling number; (c) MSE over sampling number; (d) data fidelity over sampling number; (e) number of nonzero components over sampling number.

(a)

(b)

(c)

(d)

(e)