One Shot Learning with Siamese Networks in PyTorch

Deep neural networks are the go to algorithm when it comes to image classification. This is partly because they can have arbitrarily large number of trainable parameters. However, this comes at a cost of requiring a large amount of data, which is sometimes not available. I will discuss One Shot Learning, which aims to mitigate such an issue, and how to implement a Neural Net capable of using it,in PyTorch. This article assumes some familiarity with neural networks.