This convinces us that sparse coding based classifiers can be very effective for classifying high-throughput biological data.
A more detailed description of maximum-margin based classifiers can be found in Schölkopf and Smola [14] for further reading.
Implementation details regarding the developed SVM-based classifier can be found in [9].
In more concrete terms, a ligand-based classifier can be viewed as an algorithm that appropriately associates ligand j denoted by a descriptor vector x j with protein target(s) – often referred to as classes/labels, denoted by l 1, …, l k, …, l |L| – against which x j may show activity.
Thus, combined with experimentally verified or computationally predicted singly modified sites, our phos-based classifier and O-GlcNAcylated-based classifier can be used for proteomic-wide screening and systematic examination for in situ O-GlcNAc/Phos cross talk sites.
Since genome-wide RNAi screens used for discovering functional genes in stem-cell self-renewal and pluripotency maintenance produce candidate lists that are inherently noisy, the SVM-based classifier can be applied to prioritize experimental choices when facing with a large list of candidate genes to verify and further functionally characterize.
One advantage of this schema is that adding or deleting the base classifiers can be performed relatively easily since no communications are required between them in the first level.
Since each training data is very big and split across a set of computing nodes, there is no way that the final base classifiers can be trained on the whole training datasets.
In this space, the vector quantization based meta classifier can be viewed as an ECOC decoder, where each learned prototype of a class can be seen as a codeword of the class in the new representation space.
