Std and mean not clear at all

Hello fellas,
What is the purpose of find mean and std, is from each image or whole dataset, what is use for, how to find it? not clear yet and i have see that is a must in all models or am I wrong? thanks a lot

Normalizing the data is often beneficial and the standardization or whitening of the data by subtracting the mean and dividing by the stddev shows commonly a superior convergence of the model. I’m sure there might be exceptions, but it’s quite the standard now. I couldn’t find a quick reference but am sure that Bishop explains it in Pattern Recognition and Machine Learning.

1 Like