I have some multichannel data. For convenience, I need to make sure the number of samples in each channel are same. If I remove the outliers, the number of samples are not uniform and filling outliers leads to false data points. What is the solution?