Siamese-Network-with-Triplet-Loss-Implementation

Siamese Network with Triplet Loss in Keras

Implementing a Triplet Loss function, creating a Siamese Network and train the network with the Triplet Loss function. With this training process, the network will learn to produce Embedding of different classes from a given dataset in a way that Embedding of examples from different classes will start to move away from each other in the vector space

Applying this implemetation on the default keras MNIST dataset

This appraoch is taken from the popular FaceNet paper.