make wrapper for architecture

[geoffwoollard]

write downsampling code for

• mean/median/max pooling
• arbitrary input size to arbitrary output size
• implement for our size (380) to 128

[geoffwoollard]

The deep consensus architecture repeats

• max pooling (kernel size param)
• conv2d
• conv2d (filter depth param)
• activation
• batch normalization

The wrapper should take tuples of the kernel size param and filter depth param.

The user can build the model architecture and do model.summary() to check the number of trainable params.

[Davjes15]

All the parameters can be customizable. The final layer should have 8X8X64 Depending on the size of the image user can decide to add extra layers to get the best result

[Davjes15]

I made wrapper that you can customize

• Number of layers
  • Kernel size
  • max pooling size
  • ave pooling size
  • Number of nodes in each layer
  • Dense layer
  • drop rate

[geoffwoollard]

@Davjes15 please commit and push your code to the repo

[Davjes15]

I just did it please let me know if you can see it

[geoffwoollard]

Yes, closing issue.