https://github.com/pfnet-research/chainer-ADDA

Adversarial Discriminative Domain Adaptation in Chainer
https://github.com/pfnet-research/chainer-ADDA

adda adversarial chainer chainer-adda domain-adaptation mnist svhn

Last synced: 2 days ago
JSON representation

Adversarial Discriminative Domain Adaptation in Chainer

• Host: GitHub
• URL: https://github.com/pfnet-research/chainer-ADDA
• Owner: pfnet-research
• License: mit
• Created: 2017-06-21T09:07:40.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2017-11-20T05:26:36.000Z (almost 7 years ago)
• Last Synced: 2024-08-02T12:21:52.271Z (3 months ago)
• Topics: adda, adversarial, chainer, chainer-adda, domain-adaptation, mnist, svhn
• Language: Python
• Size: 1.99 MB
• Stars: 25
• Watchers: 5
• Forks: 4
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# chainer-ADDA

Implementation of [Adversarial Discriminative Domain Adaptation](https://arxiv.org/abs/1702.05464) in [Chainer](https://github.com/chainer/chainer).

**Note** this code depends on [this](https://github.com/chainer/chainer/tree/96753116c96e26d33e4d747673114b06b86ebb6b) version of Chainer (or newer). Please check out the source from that link rather than installing via pip.

## Results

The following results are for the SVHN to MNIST domain adaptation task.

![loss](loss.png)

| Training | % accuracy (Paper) | % accuracy (This implementation) |

| -------- | ----- | ------------------- |

| source only   | 0.601 | 0.575          |

| ADDA          | 0.760 | 0.800          |

## Usage

Run `python train.py -g 0` to train everything using GPU 0. SVHN and MNIST datasets will download automatically. If a classifier pretrained on the source (SVHN) domain is not found, one will be trained first, then continue on to do ADDA.

# Resources

- https://arxiv.org/pdf/1702.05464.pdf

- https://github.com/erictzeng/adda

- https://github.com/davidtellez/adda_mnist64