April 2020
Beginner to intermediate
156 pages
4h 47m
English
The whole idea of using Siamese architecture is not to classify between classes but to learn to discriminate between inputs. So, it requires a differentiating form of the loss function known as the contrastive loss function. This is given as follows:

In this equation,
,
represents the Siamese neural network, and m represents the margin.
Let's solve the loss equation further. Take for similar pairs:
If both inputs ...
Read now
Unlock full access