search

yxgeee / MMT

[ICLR-2020] Mutual Mean-Teaching: Pseudo Label Refinery for Unsupervised Domain Adaptation on Person Re-identification.
https://yxgeee.github.io/projects/mmt
MIT License
469 stars 73 forks source link

about classifier weight initial #36

Closed Youskrpig closed 4 years ago

Youskrpig commented 4 years ago

Hi, thanks for your great work, i noticed in base_train_dbscan.py, line 171
model.module.classifier.weight.data[:args.numclusters].copy(F.normalize(cluster_centers, dim=1).float().cuda()) what does it mean and why not directly setting classifier weight==0? Looking for your reply.

yxgeee commented 4 years ago

We use cluster centroids to initialize the classifier weights for better convergence.

Youskrpig commented 4 years ago

get it