|
|
|
|
|
|
by p1esk
2424 days ago
|
|
|
Yes, that's a good move. Let me know when you have CIFAR-10 results, I will try to match your accuracy using the same number of parameters in a regular convnet. I actually implemented the original, vector based capsnet a while ago: https://github.com/michaelklachko/CapsNet but I haven't really explored it. Your success on CIFAR-10 would definitely provide motivation for me to do so. |
|
|
FWIW, a while back I reimplemented and tinkered a bit with the Sabour et al. version too... and did not see much promise in it.
Note that the routing algorithm I've proposed generalizes to vectors (by setting the dimension of the covector space d_cov to 1).