Figure 1.

Comparison of ensemble of SVMs and single SVM from unbalanced data for N-linked glycosylation using local sequence identity. ROC curves for ensemble of SVMs and single SVM trained on the "natural" distribution of the data extracted from the original glycoprotein sequence dataset for N-linked glycosylation using local sequence identity with 0/1 String Kernel.

Caragea et al. BMC Bioinformatics 2007 8:438   doi:10.1186/1471-2105-8-438
Download authors' original image