I want to train a SOM and I was just thinking which normalization to use. In literature (Data Mining and Predictive Analytics by Daniel T. Larose) I found that I need to use the zscore where the mean is equal to zero and the standard deviation is one of the normalized dataset. But a SOM is a special neural network which initializes weights at the beginning from 0-1. So like for neural networks I should normalize my data on the 0-1 scale (MinMax). Any suggestions or links to explanations which normalization to use? Or do both work just fine for the purpose?