Optimize learning and connections in SNN code

Xmodal / autonomX

Experiment with life-like processes to generate dynamic, emergent and self-organizing patterns and output these patterns via OSC.

10 stars 4 forks source link

Optimize learning and connections in SNN code #198

Open alx-s opened 4 years ago

alx-s commented 4 years ago

Optimize the connection array so that no unnecessary computing is done (eg: no need to compute a connection that does not actually exist).
Question: how does it relate to learning, and how do we approach efficiency and size of networks versus ability of the network to learn
have a number box or slider to select the maximum number of "learning connections"

alx-s commented 4 years ago

@simondemeule Need precision and specification

alx-s commented 4 years ago

@simondemeule We need to clarify the distinciton between "inactive connection" and "active connection with have a value of zero" (does it mean they became inactive?). The big question behind is: in the SNN network model, can a connection initially at zero eventually develop and grow over time? (the code implementation allows it but does it make sense in term of model)