Selective multiple kernel learning for classification with ensemble strategy

Tao Sun, Licheng Jiao, Fang Liu, Shuang Wang, Jie Feng

Abstract: Highlights•We obtain a competitive result with MKL, meanwhile owning sparsity.•We propose a new kernel evaluation method with quantified result.•We save the memory to optimize MKL and extend the scale of problem.•We accelerate MKL optimization by using Lp-norm(p≥2)<math><mi mathvariant="italic" is="true">Lp</mi><mi mathvariant="normal" is="true">-</mi><mi mathvariant="italic" is="true">norm</mi><mo stretchy="false" is="true">(</mo><mi is="true">p</mi><mo is="true">≥</mo><mn is="true">2</mn><mo stretchy="false" is="true">)</mo></math>.•A fast SMKL with L∞-norm<math><mi is="true">L</mi><mo is="true">∞</mo><mi mathvariant="normal" is="true">-</mi><mi mathvariant="italic" is="true">norm</mi></math> is proposed, without MKL optimization.