co-regularied-alignment-for-domain-adaptation

This is an UNOFFICIAL implemnetation of Co-regularized Alignment for Unsupervised Domain Adaptation

---

Prerequisite

• python 3.6
• PyTorch 1.1
• tensorboardX
• numpy
• cv2
• tqdm

---

Training losses & results from MNIST->SVHN task

agreement loss	diverse loss

---

---

net_num	validation	test
net1
net2

---

TODO:

1. DIRT-T model

---

Note: See https://github.com/lyakaap/VAT-pytorch for original implementation of VAT loss