We can take a Siamese Network and have both the computed embedding and input these into a logistic regression unit to the make a prediction. Where the target output will be 1 if both are the same person and 0 if different people.
If is the output, then we can define it as :
In some networks this definition is also used :
Now that we have covered the mathematics and theory, we can look at the Code Implementation of Face Recognition.
Now let us look at now we can use Neural Style Transfer to make artwork.