a deep neural expression based classifier demonstrated to fit a unbalanced dataset, expression datasets across the samples. a deep neural expression based classifier demonstrated to fit a unbalanced dataset, expression datasets across the samples where summed and then divided by the length of the total replicates and then i classify them as those which have expression greater than 100 as 1 and those which have lower than 100 as 0 and build this neural network. Another optimization would be to the dropoutlayer and also present a normalizing approach, which can make the expression unbiasness looks straight. I am working on the same so that the deviations caused by the change in the wide expression is not a problem.
Gaurav
Academic Staff Member
Bioinformatics
Institute for Biochemistry and Biology
University of Potsdam
Potsdam,Germany