A Novel Sparse Least Squares Support Vector Machines

Figure 3

Comparison between standard SVMs and FLSA-SVMs: (a) two-spiral pattern recognized by the FLSA-SVM trained on 134 data points with 𝜆=1 in (36); (b) two-spiral pattern recognized by the standard SVM trained on 164 data points. The penalty constant 𝐶=32 and 𝜎=0.5.

(a)

(b)