Figure 1From: Learning Interpretable SVMs for Biological Sequence ClassificationIn this "figure matrix", columns correspond to the noise level, i.e. different numbers of nucleotides randomly substituted in the motif of the toy data set (cf. Appendix A.1). Each sub-figure shows a matrix with each element corresponding to one kernel weight: columns correspond to weights used at a certain sequence position (1–50) and rows to the oligomer length used at that position (1–7). The first row of the figure matrix shows the kernel weights that are significant, while the second row depicts the likelihood of every weight to be rejected under ℋ 0 MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBamrtHrhAL1wy0L2yHvtyaeHbnfgDOvwBHrxAJfwnaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaWaaeGaeaaakeaaimaacqWFlecsdaWgaaWcbaGae8hmaadabeaaaaa@3874@ .Back to article page