# Face Recognition as Binary Classification

Face Recognition can also be looked at as a binary classification problem.

We know that the Siamese Network outputs a pair of encodings, one for each image.

These encodings can be used to train a Logistic Regression classifier with labels 0 (not same person) and 1 (same person), instead of using the triplet loss function for training.

The Logistic Regression equation for this problem would be:

$$\hat{y} = \sigma (\sum\_{k=1}^{num\_features} w\_i |f(x^{(i)})\_k-f(x^{(j)})\_k| + b)$$