Skip to main content

Table 2 Hyperparameters for the different classifiers. For all of them random undersampling [23] was used

From: Stepwise approach for combining many sources of evidence for site-recognition in genomic sequences

Classifier Inputs Hyperparameters
Decision trees Raw sequence Pruned trees
Position weight matrix Raw sequence None
Stop codon method Raw sequence None
Support vector Raw sequence C{1,10,100,100},d{12,24}
k nearest neighbor Raw sequence Hamming distance, k[1,100]