Replace Add with concatenate+conv. Why?

NeuromorphicProcessorProject / snn_toolbox

Toolbox for converting analog to spiking neural networks (ANN to SNN), and running them in a spiking neuron simulator.

MIT License

362 stars 105 forks source link

Replace Add with concatenate+conv. Why? #74

Closed joaromi closed 3 years ago

joaromi commented 3 years ago

Hi, I was wondering why is it necessary to replace the Add layer for concatenate+conv layers. As it is a layer with no spiking dynamics wouldn't it be easy to implement it in INI? Did you do it to account for other simulation environments?