For Reducing the dataset exact Independent Component Analysis (ICA) doesn’t exist but you may usericafor doing the dimensionality reduction. The Reconstruction ICA work closely as ICA but an added feature of optimizing non-linear objective function that also include penalty while reconstructing the output. This can improve the convergence speed as compared to ICA and the features you get are globally orthogonal. Reconstruction ICA is supported in R2017a and later versions.
Thank you so much for your answer, but I need to keep the dimensions same as reducing its dimensions will affect the labelled data. I wish to first do the ICA and then extract features for specific range of values.