刘宗游

How did you evaluate the performance of your algorithm in your paper without groud truth labels for real datasets you used?

Thank you！