Capsule Neural networks (CapsNets)
A Capsule Neural Network (CapsNet) is a machine learning system that is a type of artificial neural network (ANN) that can be used to better model hierarchical relationships. The approach is an attempt to more closely mimic biological neural organization
# Resources
- https://en.wikipedia.org/wiki/Capsule_neural_network.
- The main failure of CNNs is that they do not carry any information about the relative relationships between features (CNNs since they are based on the convolution operation applied to scalar values).
- Capsules introduce a new building block that can be used in deep learning to better model relationships inside the network. The key to this richer feature representation is the use of vectors rather than scalars.
- A capsule is an abstract idea of having a group of neurons with an activity vector that contains more information about the object. There are many ways to implement this. Hinton et al chose one particular way to implement this, which allows using “dynamic routing”.
- https://towardsdatascience.com/a-simple-and-intuitive-explanation-of-hintons-capsule-networks-b59792ad46b1
- https://towardsdatascience.com/capsule-neural-networks-are-here-to-finally-recognize-spatial-relationships-693b7c99b12
- https://towardsdatascience.com/capsule-neural-networks-part-2-what-is-a-capsule-846d5418929f
- https://www.freecodecamp.org/news/understanding-capsule-networks-ais-alluring-new-architecture-bdb228173ddc/