Figure 3 | Scientific Reports

Figure 3

From: An adaptive shortest-solution guided decimation approach to sparse high-dimensional linear regression

Figure 3

Performance comparison on the original SSD (left), SSD1 which is SSD with early-stopping, and ASSD. The measurement matrix is \(200\times 1000\), while the noise level is \(\sigma ^2 = 1\). The blue crosses mark the 30 nonzero elements of the true coefficient vector \(\varvec{\beta }^0\), the red circles mark the elements of the estimated coefficient vector \(\varvec{\beta }\). In ASSD, we set \(R=20\) and \(\eta =0.1\).

Back to article page