Skip to main content

Visualizing how t-SNE works by TensorBoard (image embedding)



Each image has 300 dimensions vectors those are representing “features”. And the image distances are representing “similarity”.

source

Leave a Reply

Your email address will not be published. Required fields are marked *