SimilarityLearning

Various Embeddings

Reproduce some embedding methods on MNIST:

• Siamese loss in the paper Learning a Similarity Metric Discriminatively, with Application to Face Verification.
• Cosine loss
• Triplet loss in the paper FaceNet: A Unified Embedding for Face Recognition and Clustering
• Softmax triplet loss in the paper Deep Metric Learning using Triplet Network
• Center loss in the paper A Discriminative Feature Learning Approach for Deep Face Recognition

Usage:

# to train:
./mnist-embeddings.py --algorithm [siamese/cosine/triplet/softtriplet/center]
# to visualize:
./mnist-embeddings.py --algorithm [siamese/cosine/triplet/softtriplet/center] --visualize --load train_log/mnist-embeddings/checkpoint