So neurons tuning curves can be linearly weighted to approximate input variables, that I understand. However where does these weighting calculations occur in Nengo?
Thanks
So neurons tuning curves can be linearly weighted to approximate input variables, that I understand. However where does these weighting calculations occur in Nengo?
Thanks
You mentioned the code specifically, so I’m assuming you already know the theory behind solving for weights/decoders. However, if I’m wrong, feel free to ask questions about the theory too.
These weighting calculations occur using an instance of a nengo.Solver
passes as an argument into nengo.Connection(pre, post, solver=nengo.Solver)
. The default solver is the Least-Squares solver with L2 regularization.
Yes thanks yes i’m aware of the theory - min squared difference etc.