In this paper, we raise important issues on scalability and the required degree of supervision of existing Mahalanobis metric learning methods.
確定! 回上一頁