On estimation of entropy and mutual information of continuous distributions