Divergence domain adaptation methods.

This example illustrates the DeepCoral method from [1] on a simple image classification task.

[1]

Baochen Sun and Kate Saenko. Deep coral: Correlation alignment for deep domain adaptation. In ECCV Workshops, 2016.

# Author: Théo Gnassounou
#
# License: BSD 3-Clause
# sphinx_gallery_thumbnail_number = 4

from skorch import NeuralNetClassifier
from torch import nn

from skada.datasets import load_mnist_usps
from skada.deep import DeepCoral
from skada.deep.modules import MNISTtoUSPSNet

Load the image datasets

dataset = load_mnist_usps(n_classes=2, n_samples=0.5, return_dataset=True)
X, y, sample_domain = dataset.pack(
    as_sources=["mnist"], as_targets=["usps"], mask_target_labels=True
)
X_test, y_test, sample_domain_test = dataset.pack(
    as_sources=[], as_targets=["usps"], mask_target_labels=False
)

  0%|          | 0.00/9.91M [00:00<?, ?B/s]
100%|██████████| 9.91M/9.91M [00:00<00:00, 151MB/s]

  0%|          | 0.00/28.9k [00:00<?, ?B/s]
100%|██████████| 28.9k/28.9k [00:00<00:00, 41.4MB/s]

  0%|          | 0.00/1.65M [00:00<?, ?B/s]
100%|██████████| 1.65M/1.65M [00:00<00:00, 75.7MB/s]

  0%|          | 0.00/4.54k [00:00<?, ?B/s]
100%|██████████| 4.54k/4.54k [00:00<00:00, 14.3MB/s]
/home/circleci/project/skada/datasets/_mnist_usps.py:72: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.detach().clone() or sourceTensor.detach().clone().requires_grad_(True), rather than torch.tensor(sourceTensor).
  mnist_target = torch.tensor(mnist_dataset.targets)

  0%|          | 0.00/1.83M [00:00<?, ?B/s]
  2%|▏         | 32.8k/1.83M [00:00<00:10, 165kB/s]
  5%|▌         | 98.3k/1.83M [00:00<00:06, 261kB/s]
 11%|█         | 197k/1.83M [00:00<00:04, 366kB/s]
 16%|█▌        | 295k/1.83M [00:00<00:03, 417kB/s]
 23%|██▎       | 426k/1.83M [00:00<00:02, 504kB/s]
 29%|██▊       | 524k/1.83M [00:01<00:03, 372kB/s]
 45%|████▍     | 819k/1.83M [00:01<00:01, 666kB/s]
 52%|█████▏    | 950k/1.83M [00:01<00:01, 616kB/s]
 57%|█████▋    | 1.05M/1.83M [00:01<00:01, 629kB/s]
 93%|█████████▎| 1.70M/1.83M [00:02<00:00, 1.40MB/s]
100%|██████████| 1.83M/1.83M [00:02<00:00, 833kB/s]

Train a classic model

model = NeuralNetClassifier(
    MNISTtoUSPSNet(),
    criterion=nn.CrossEntropyLoss(),
    batch_size=128,
    max_epochs=5,
    train_split=False,
    lr=1e-2,
)
model.fit(X[sample_domain > 0], y[sample_domain > 0])
model.score(X_test, y=y_test)

  epoch    train_loss     dur
-------  ------------  ------
      1        1.0683  3.9993
      2        0.1451  3.5001
      3        0.0613  3.6998
      4        0.0406  3.4957
      5        0.0318  3.5025

0.9003215434083601

Train a DeepCoral model

model = DeepCoral(
    MNISTtoUSPSNet(),
    layer_name="fc1",
    batch_size=128,
    max_epochs=5,
    train_split=False,
    reg=1,
    lr=1e-2,
)
model.fit(X, y, sample_domain=sample_domain)
model.score(X_test, y_test, sample_domain=sample_domain_test)

  epoch    train_loss     dur
-------  ------------  ------
      1        1.4604  8.2720
      2        0.3677  7.0954
      3        0.1435  6.6957
      4        0.0882  7.1028
      5        0.0665  7.4022

0.9035369774919614