def connect(self):
# connect populations
self.injection = pynn.Projection(self.sources,
self.columns,
pynn.FixedProbabilityConnector(
0.02, weights=0.0033 * 2),
target='excitatory',
rng=pynn.random.NumpyRNG(seed=5337))
self.recurrent_excitation = pynn.Projection(
self.columns,
self.columns,
pynn.OneToOneConnector(weights=0.01 * 2),
target='excitatory')
self.lateral_inhibition = pynn.Projection(
self.columns,
self.columns,
pynn.DistanceDependentProbabilityConnector('d < 10',
weights=0.004 * 2),
target='inhibitory',
rng=pynn.random.NumpyRNG(seed=5337))
self.global_inhibition = pynn.Projection(
self.inhibitory_pool,
self.columns,
pynn.FixedProbabilityConnector(0.8, weights=0.0012 * 2),
target='inhibitory',
rng=pynn.random.NumpyRNG(seed=5337))
self.forward_inhibition = pynn.Projection(
self.sources,
self.inhibitory_pool,
pynn.FixedProbabilityConnector(0.8, weights=0.0035),
target='excitatory',
rng=pynn.random.NumpyRNG(seed=5337))
def create_local_inhibition(layers_dict):
"""
Creates local inhibitory connections from a neuron to its neighbors in an
area of a fixed distance. The latency of its neighboring neurons decreases
linearly with the distance from the spike from 15% to 5%, as described in
Masquelier's paper. Here we assumed that a weight of -10 inhibits the
neuron completely and took that as a starting point.
"""
for size, layers in layers_dict.items():
print('Create local inhibition for size', size)
for layer in layers:
sim.Projection(layer.population,
layer.population,
sim.DistanceDependentProbabilityConnector(
'd < 5', allow_self_connections=False),
sim.StaticSynapse(weight='.25 * d - 1.75'),
space=space.Space(axes='xy'))
def connect(self):
n_cells = self.parameters.config.n_cells
n_segments = self.parameters.config.n_segments
# connect populations
pynn.Projection(self.distal_input, self.distal,
pynn.OneToOneConnector(weights=0.025))
for i in range(self.parameters.config.n_columns):
# get "compartments" for all cells in this column
inhibitions = self.inhibitory[i * n_cells:(i + 1) * n_cells]
somas = self.soma[i * n_cells:(i + 1) * n_cells]
proximal_input = pynn.PopulationView(self.proximal_input, [i])
# set up connections with columnar symmetry
pynn.Projection(inhibitions,
somas,
pynn.DistanceDependentProbabilityConnector(
'd>=1', weights=0.2),
target='SYN_2') # 4
pynn.Projection(proximal_input,
somas,
pynn.AllToAllConnector(weights=0.08),
target='SYN_1') # 1
pynn.Projection(proximal_input, inhibitions,
pynn.AllToAllConnector(weights=0.042)) # 2
for j in range(self.parameters.config.n_cells):
# get "compartments" for this specific cell
segments = self.distal[i * n_cells * n_segments + j *
n_segments:i * n_cells * n_segments +
(j + 1) * n_segments]
inhibition = pynn.PopulationView(self.inhibitory,
[i * n_cells + j])
soma = pynn.PopulationView(self.soma, [i * n_cells + j])
# set up connections with cellular symmetry
pynn.Projection(segments,
inhibition,
pynn.AllToAllConnector(weights=0.15),
target='inhibitory') # 3
pynn.Projection(segments,
soma,
pynn.AllToAllConnector(weights=0.15),
target='SYN_3') # 3