A multi-feature selection approach for gender identification of handwriting based on kernel mutual information | Litlas