Skip to content

Re-implementation of the vector capsule with dynamic routing

License

Notifications You must be signed in to change notification settings

ZhenchaoTang/VectorCapsule

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

VectorCapsule

Re-implementation of the vector capsule with dynamic routing

We implement the vector capsule and dynamic routing via graph neural network. We trained the capsule network on MNIST, the record of top1 accuracy on test dataset is:

Epoch Accuracy
1 Test set: Average loss: 0.0006, Accuracy: 9556/10000 (96%)
2 Test set: Average loss: 0.0003, Accuracy: 9812/10000 (98%)
3 Test set: Average loss: 0.0002, Accuracy: 9835/10000 (98%)
4 Test set: Average loss: 0.0002, Accuracy: 9875/10000 (99%)
5 Test set: Average loss: 0.0001, Accuracy: 9900/10000 (99%)

About

Re-implementation of the vector capsule with dynamic routing

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages