TensorFlow 1.x Deep Learning Cookbook

There's more...

TensorFlow provides a technique called embeddings, which is a mapping of objects into vectors. Embedding Projector of TensorBoard allows us to interactively visualize embeddings from our model. Embedding Projector provides three ways for dimensionality reduction: PCA, t-SNE, and Custom. We can use the Embedding Projector of TensorBoard to achieve a similar result as the previous one. We will need to import the projector class from tensorflow.contrib.tensorboard.plugins to do the same from tensorflow.contrib.tensorboard.plugins import projector. We can do it in three simple steps:

1. Load the data whose embeddings you want to explore:

mnist = input_data.read_data_sets('MNIST_data')images = tf.Variable(mnist.test.images, name= ...