This feature uses singular value decomposition (SVD) to reduce the dimensionality of the inputted hidden markov model matrix. SVD factorizes a matrix C of dimensions \(i, j\) to \(U[i, r] \times \Sigma[r, r] \times V[r, j]\). The diagonal values of \(\Sigma\) are known as the singular values of matrix C, and are what are returned with this function.
References
Song, X., Chen, Z., Sun, X., You, Z., Li, L., & Zhao, Y. (2018). An Ensemble Classifier with Random Projection for Predicting Protein–Protein Interactions Using Sequence and Evolutionary Information. Applied Sciences, 8(1), 89.
Examples
h<- hmm_svd(system.file("extdata", "1DLHA2-7", package="protHMM"))