2

I am a python Intermediate and Keras/NN beginner. I have implemented Keras models before from tutorials and want to try implementing one from a research paper. What does this schematic mean in relation to a Keras model?

enter image description here

Where "Input Cross-presence Matrix" is a incremental count of the number of times every pair of numbers (1-45) appeared together eg (1-2, 1-3 .. 1-45).

"Other input" is the incremental count of the number of times a number (1-45) appeared.

The output is a non-linear sigmoid of the presence or not a number (1-45) in an 8 wide array.

Thanks for any help.

Thiedent
  • 126
  • 3
  • 9

1 Answers1

3

You want a multi-input model. The first input will go through a first Convolutional layer with kernel size 3. The output of this layer will go through 2 more of these layers, and then be added back to the original output of the first layer (basically a residual/skip connection implementation). Then this output will be concatenated with an auxiliary input and fed to a fully-connected Dense layer, and then again to the last one.

Hope this helps as a start!

Michele Tonutti
  • 4,298
  • 1
  • 21
  • 22