Siamese Networks for One-Shot Learning

A reimplementation of the original paper in pytorch with
training and testing on the Omniglot dataset.

requirement

• pytorch
• torchvision
• python3.5+
• python-gflags

See requirements.txt

run step

• download dataset

git clone https://github.com/brendenlake/omniglot.git
cd omniglot/python
unzip images_evaluation.zip
unzip images_background.zip
cd ../..
# setup directory for saving models
mkdir models

• train and test by running

python3 train.py --train_path omniglot/python/images_background \
                 --test_path  omniglot/python/images_evaluation \
                 --gpu_ids 0 \
                 --model_path models

experiment result

Loss value is sampled after every 200 batches

My final precision is 89.5% a little smaller than the result of the paper (92%).

The small result difference might be caused by some difference between my implementation and the paper’s. I list these differences as follows:

• learning rate

instead of using SGD with momentum I just use ADAM.

• parameters initialization and settings

Instead of using individual initialization methods, learning rates and regularization rates at different layers I simply use the default setting of pytorch and keep them same.