[WIP] Pytorch implementation of Adversarial Discriminative Domain Adaptation
- code reference: https://github.com/corenel/pytorch-adda
- source & target classifier: LeNet 5 (followed implementation in Caffe)
- discriminator: 3-layer MLP (500-500-1) with ReLU activation
- This does not converge for some reasons.
Envirionment
- Python 3.6
- PyTorch 0.4.0
- torchvision 0.2.1
Result
| SVHN(Source) | MNIST(Target) | |
|---|---|---|
| Source Classifier | 0.9168 | 0.633 |
| Target Classifier | - | WIP |
LeNet 5 implementation
