Recently, I made a tutorial on Capsule Networks! Capsule Networks were introduced by Geoffrey Hinton and have been shown to be more robust to adversarial attacks, achieve higher accuracy, all while requiring significantly fewer parameters.

The tutorial covers how CNNs and Capsule networks account for viewpoint variance (a massive problem in computer vision), an introduction to Capsule Networks, and two routing algorithms, Dynamic Routing Between Capsules, and Matrix Capsules with EM Routing.

The post was written from zero knowledge to a complete understanding of the intuition and algorithm details in an easy to understand way. I’m sharing the post since it would be helpful to others wanting to learn about the topic, enjoy!

​

https://gebob19.github.io/capsule-networks/