search

comob-project / snn-sound-localization

Training spiking neural networks for sound localization
https://comob-project.github.io/snn-sound-localization/
MIT License
60 stars 26 forks source link

Figure in main text: cleaned up sketch of model architecture #144

Closed tfiers closed 2 months ago

tfiers commented 3 months ago

[I'm on holiday these weeks but should have time for this on the train this Saturday the 6th].

I will then also remove the sketch figure in @thesamovar's section "A minimal trainable model of IPD processing", and refer instead to the figure in the main text.

Maybe would be good to also include (one or two subplots of) figure 2 (the spike train examples) in the main text?

tfiers commented 3 months ago

Were we thinking of putting this diagram in the main text (input only): arch-stimuli-4584fed55f158904200019e64c66bd5a

or this one (input plus trained model)? arch-full-63562c0fd6f294b5e214c72df88d84e0

thesamovar commented 3 months ago

Second one! Thanks Tomas.

tfiers commented 2 months ago

I tried to put some actual example spike trains in the figure (corresponding notebook: https://github.com/comob-project/snn-sound-localization/blob/b11c474a739cdd6013a34123c5707c55faad5dbf/research/figure_input-and-architecture.ipynb), but that makes the spikes are too small in comparison to the rest of the figure: image

So abandoning this idea.

ghoshm commented 2 months ago

Looks great - I've added this to the paper with a caption.

thesamovar commented 2 months ago

Thanks Tomas, your updated version of this figure looks great!