Can we use DeepInsight or CNN with very few features?

I have a dataset with n=891 number of samples and each sample is described by 'd'=6 number of features/dimensions. If I will calculate tSNE on those 6 features then I'll end up having only 6 pixels coordinates therefore it's not really something CNN could process, there is not enough data to perform convolution on those 6 pixels.


  1. Add comments here.

Leave a Comment