/torch-TripletEmbedding

TripletLoss used in Google's FaceNet paper

Primary LanguageLuaMIT LicenseMIT

TripletEmbedding Criterion

This aims to reproduce the loss function used in Google's FaceNet paper.

criterion = nn.TripletEmbeddingCriterion([alpha])

The cost function can be expressed as follow

1 __ 2 2 L({a, p, n}) = - \ max(0, ||a - p || + alpha - ||a - n || ) N /__ i i i i i

where a, p and n are batches of the embedding of ancore, positive and negative samples respectively.

If the margin alpha is not specified, it is set to 0.2 by default.

Test

In order to test the criterion, someone can run the test script as

th test.lua

which shows how to use the criterion and checks the correctness of the gradient.

Training

The folder xmp contains two examples which show how a network can be trained with this criterion.

  • recycle-embedding recycles the embedding of the positive and negative sample from the previous epoch (faster training, less accurate)
  • fresh-embedding computes the updated embedding of all ancore, positive and negative training samples (correct algorithm, thrice slower)

Triplet construction

The folder data contains a package for generating triplets to feed to a network.

To test the data script, run data-test.lua, but you need to have a dataset in the format described in data.lua. In this same file is provided a snippet from the training script.