{"id":54716,"date":"2022-02-23T00:04:05","date_gmt":"2022-02-22T18:34:05","guid":{"rendered":"https:\/\/www.tothenew.com\/blog\/?p=54716"},"modified":"2022-02-23T00:04:05","modified_gmt":"2022-02-22T18:34:05","slug":"oneshot-learning","status":"publish","type":"post","link":"https:\/\/www.tothenew.com\/blog\/oneshot-learning\/","title":{"rendered":"OneShot Learning"},"content":{"rendered":"

Written by <\/b>Tushar Raj Verma<\/b><\/a> &<\/strong>\u00a0<\/b>Abhishek Kumar Upadhyay<\/b><\/a><\/em><\/p>\n

Deep Convolutional Neural Networks have become the state of the art methods for image classification tasks. <\/span>However, one of the biggest challenges is that they require a lot of labeled data to train the model. In many applications, collecting this much data is sometimes not feasible. One-Shot Learning aims to solve this problem. <\/span>To understand <\/span>One-Shot Learning<\/span>, we first need to understand Siamese networks.<\/span><\/p>\n

Siamese Networks<\/b><\/p>\n

This is a type of model network architecture that contains two or more identical subnetworks which are used to generate feature vectors(or embeddings) for each input and compare them. <\/span>Siamese Networks can be applied in use cases, like face recognition, detecting duplicates, and finding anomalies. <\/span>Our goal is for the model to learn to estimate the similarity between images. For this, we will provide three images to the model, where two of them will be similar (<\/span>anchor<\/span><\/i> and <\/span>positive<\/span><\/i> samples), and the third will be unrelated (a <\/span>negative<\/span><\/i> example). <\/span>For the network to learn, we use a triplet loss function.<\/span><\/p>\n

Triplet Loss<\/b><\/p>\n

Its introduction was first introduced in the <\/span>FaceNet paper<\/span><\/a>. TripletLoss is a loss function that trains a neural network to closely embed features of the same class while maximizing the distance between embeddings of different classes. To do this an anchor is chosen along with one negative and one positive sample.<\/span><\/p>\n

\"\"Figure: The Triplet loss minimizes the distance between an anchor and a positive<\/em><\/p>\n

The loss function is described as a Euclidean distance function:<\/b><\/p>\n

\"\"<\/p><\/blockquote>\n

Here, A is our anchor input, P is the positive sample input, N is the negative sample input, and \u201cmargin\u201d which you use to specify when a triplet has become too “easy” and you no longer want to adjust the weights from it.<\/span><\/p>\n

Semi-Hard Online Learning<\/b><\/p>\n

The best results are from the triplets known as \u201cSemi-Hard\u201d. These are defined as triplets where the negative is farther from the anchor than the positive but still produces a positive loss. To efficiently find these triplets you utilize online learning and only train from the Semi-Hard examples in each batch.<\/span><\/p>\n

WorkFlow<\/b><\/p>\n

\"\"<\/p>\n

There are two phases in the flow of OneShot<\/strong><\/p>\n

1. Training Phase<\/strong><\/p>\n