Normalization is the process of transforming raw values to standard range of values, typically [0, 1]

• It increases learning speed; hence fast convergence
• If no normalization, then big numeric features dominate the learning