search

WenbinLee / ADM

The Pytorch code of "Asymmetric Distribution Measure for Few-shot Learning", IJCAI 2020.
Other
15 stars 2 forks source link

ADM'ecperiments on deeper models #2

Open Fei-Long121 opened 3 years ago

Fei-Long121 commented 3 years ago

Great job! Many thanks for sharing! But I want to know that Have you experimented with deeper models,such as resnet12? If so, could you share the results on miniImageNet and Tiered_imagenet?

WenbinLee commented 3 years ago

Thanks. I have not tried to use a deeper backbone because of the computational cost. If you want to obtain the corresponding results and you also have enough computing resource, maybe you can try to replace the shallow backbone with a deeper backbone. It's also worth mentioning that because ADM needs calculate covariance matrix, the resnet12 backbone needs to be slightly modified. Hope this can help you.

Fei-Long121 commented 3 years ago

Thanks. I have not tried to use a deeper backbone because of the computational cost. If you want to obtain the corresponding results and you also have enough computing resource, maybe you can try to replace the shallow backbone with a deeper backbone. It's also worth mentioning that because ADM needs calculate covariance matrix, the resnet12 backbone needs to be slightly modified. Hope this can help you.

ok, Thanks a lot!!!