search

Mephisto405 / Learning-Loss-for-Active-Learning

Reproducing experimental results of LL4AL [Yoo et al. 2019 CVPR]
215 stars 50 forks source link

About the performance of active learning methods #7

Closed Kyle-YK-Yu closed 4 years ago

Kyle-YK-Yu commented 4 years ago

I try some methods including random, entropy, core-set, VAAL and yours with Resnet-18 on Cifar-100 using fixed initial indices. But icannot get an ideal result. Random, entropy, VAAL and LL4AL have the almost same accuracies. Is this possible? I check my settings many times.

Mephisto405 commented 4 years ago

Yeah, please take a look at this issue: The test accuracy is not matched with your image.

In fact, I think LL4AL only achieves comparable or slightly better results. Main advantage besides the performance is its task-agnostic nature.