def buildModel(ind):
# model
model = nef.Network('model')
# input
inputs = model.add(routing.twoDimsRouter('inputs')); # TODO: general IO dimensions
# output
model.make('output', neurons=OUTdim, dimensions=OUTdim, mode='spike',
max_rate= (mr,mr), intercept=(ii,ii),encoders=[[1],[1]],decoder_sign=+1);
# spiking ANN
model.make('ANN', neurons=N, dimensions=INdim, mode='spike',
encoders = ([1,0], [0,1]), intercept= (ii, ii), max_rate = (mr,mr))
# wire it:
model.connect('ANN','ANN',weight_func=setReccurent); # recurrent conencitons
model.connect('ANN','output',weight_func=setOutW); # ANN to output
# for each input, make one neuron and connect it
for d in range(INdim):
# choose factory for input neurons:
#inFactory.makeLinearNeuron(mr,d,model,'spike');
#inFactory.makeBioLinearNeuron(mr,d,model,'spike');
inFactory.makeSigmoidNeuron(mr, d, model,'spike');
#inFactory.makeBioSigmoidNeuron(mr,d,model,'spike');
# connect shared input to input neuron and neuron to ANN with weights
actualIn=d;
model.connect(inputs.getOrigin('out'), 'in_neuron_%d'%d, index_pre=d)
model.connect('in_neuron_%d'%d, 'ANN',weight_func=setInW)
return model;
# input
model.make('input', neurons=INdim, dimensions=INdim, mode='direct',
max_rate= (mr,mr), intercept=(ii,ii), encoders=[[1,0],[0,1]])
# spiking ANN
model.make('ANN', neurons=N, dimensions=INdim, mode='rate',
encoders = ([1,0], [0,1]), intercept= (ii, ii), max_rate = (mr,mr))
# for each input, make one neuron and connect it
for d in range(INdim):
#inFactory.makeLinearNeuron(100,d,model,'spike');
#inFactory.makeBioLinearNeuron(100,d,model,'spike');
inFactory.makeSigmoidNeuron(100, d, model,'spike');
#inFactory.makeBioSigmoidNeuron(100,d,model,'spike');
model.connect('input', 'in_neuron_%d'%d, index_pre=d)
if d==1:
model.connect('in_neuron_%d'%d, 'ANN',weight_func=ident) # todo better funcitons here
else:
model.connect('in_neuron_%d'%d, 'ANN',weight_func=ident)
net.add(model.network);
# wiring
net.connect(generator,'model.input',weight=1) # feed signal to both, plant and model
net.add_to_nengo();
