flâneur — a map of the web's best reading

Encapsulating Capsule Networks: Everything You Need To Know

gebob19.github.io · 5,008 words · saved by 1 readers

This Tutorial and blog post covers everything from intuition to the basics to a complicated distribution routing algorithm. As long as you don’t have a medical phobia of capsules, you’ll be able to swallow all the knowledge in this post.

Photo Link When applying Convolutional Neural Networks (CNNs) (LeCun et al., 1990) to a computer vision task, a change in viewpoint (change in orientation, position, shear, etc.) is likely to lead to drastically different network activations, hindering the model’s ability to generalize. To solve this problem, current CNNs require a large number of parameters, datasets and computational power. This lead to the introduction of Capsule Networks (Hinton et al., 2011) . Capsule Networks aim to generalize to different viewpoints by taking advantage of the fact that the relationship between parts of

Explore this link on the map →

related reading