PatternLab for proteomics
pattern recognition software
 
  Main :: ACFold & TFold:: CPM :: GOEx :: SEPro :: Venn Diagram:: YADA :: XDIA :: MudPIT Simulator :: Downloads

natural Suport Vector Machines (nSVMs)

nSVMs address experiments that comprise three or more readings from each class. Its name was given because of its roots in evolutionary computing and in statistical learning theory. nSVM’s niche comprises projects aiming the selection of a minimum set of proteins for classification problems.  An example would be to aid in selecting a minimal but optimal combination of bona fide markers to develop an early detection kit for a given pathology. Below are a few of the parameters that can be used to tune the genetic algorithm.

Description of the GUI Parameters
fileDB Type the full path to the location of your trainning dataset.
noThreadsToUse The GA will generate as many populations as the number indicated in this parameter. Each population will "live" in a thread or "island". We suggest matching the number of threads with the number of computing cores available.
migration By checking this checkbox the GA will allow individuals to migrate from one island to another.
noInd2Migrate The number of individuals that will be exchanged between islands when the GA raises a migration event.
secBetweenMig The number of seconds necessary so the GA can raise a migration event.
populationSize How many individuals will exist in a population.
conversionAttempts The GA ceases to produce new generations after there is no increase in the fitness of the most fit individual during a user specified amount of generations specified in the conversionAttempts parameter.
noBestToKeep The user can also configure the GA to allow elitism; this parameter indicates the number of fittest individuals to continue in the new population.
mutationIndex The maxium number of mutations that can be applied to ones genome
mutInd1After When the GA has reduced the number of features, one might wish to reduce the mutation index of the genetic algorithm as well. This parameter forces the GA to allow up to only 1 mutation per new individual when his number of "active aleles" is inferior to the number indicated by this parameter.
mSprint Stands for "mutation sprint". If the GA is having difficulties to reduce the number of features, he is probably trapped in a "feature lock". By elevating the mutations in the individuals, he could escape from this feature lock.
Times to run GA To predict the amount of relevant features, the GA must be executed several times. This parameter indicates how many times the GA will be executed.
delegate2Maestro This feature is still under construction, but it will allow nSVM to use remote computing resources to speed up the algorithm. This is a grid computing module