项目作者: Therickysen08

项目描述 :
Triplet loss on MNIST
高级语言: Jupyter Notebook
项目地址: git://github.com/Therickysen08/Triplet_loss_on_MNIST.git
创建时间: 2021-01-31T10:54:02Z
项目社区:https://github.com/Therickysen08/Triplet_loss_on_MNIST

开源协议:

下载


Triplet loss on MNIST

Triplet loss is a loss function for machine learning algorithms where a baseline (anchor) input is compared to a positive (truthy) input and a negative (falsy) input. Triplet loss was first introduced in the paper FaceNet: A Unified Embedding for Face Recognition and Clustering.

The Triplet Loss minimizes the distance between an anchor and a positive, both of which have the same identity, and
maximizes the distance between the anchor and a negative of a different identity.

paper info

The same CNN is used to get embeddings of the anchor, positive and negative images and then the triplet loss is calculated.

paper info

For MNIST data, TSNE was used to reduce the dimensionality of embeddings and then the datapoints were projected in low dimensional space.

2D:

paper info

3D:

paper info