def ann_to_snn(
ann: Union[nn.Module, str],
input_shape: Sequence[int],
data: Optional[torch.Tensor] = None,
percentile: float = 99.9,
node_type: Optional[nodes.Nodes] = SubtractiveResetIFNodes,
**kwargs,
) -> Network:
# language=rst
"""
Converts an artificial neural network (ANN) written as a
``torch.nn.Module`` into a near-equivalent spiking neural network.
:param ann: Artificial neural network implemented in PyTorch. Accepts
either ``torch.nn.Module`` or path to network saved using
``torch.save()``.
:param input_shape: Shape of input data.
:param data: Data to use to perform data-based weight normalization of
shape ``[n_examples, ...]``.
:param percentile: Percentile (in ``[0, 100]``) of activations to scale by
in data-based normalization scheme.
:param node_type: Class of ``Nodes`` to use in replacing
``torch.nn.Linear`` layers in original ANN.
:return: Spiking neural network implemented in PyTorch.
"""
if isinstance(ann, str):
ann = torch.load(ann)
else:
ann = deepcopy(ann)
assert isinstance(ann, nn.Module)
if data is None:
import warnings
warnings.warn("Data is None. Weights will not be scaled.",
RuntimeWarning)
else:
ann = data_based_normalization(ann=ann,
data=data.detach(),
percentile=percentile)
snn = Network()
input_layer = nodes.Input(shape=input_shape)
snn.add_layer(input_layer, name="Input")
children = []
for c in ann.children():
if isinstance(c, nn.Sequential):
for c2 in list(c.children()):
children.append(c2)
else:
children.append(c)
i = 0
prev = input_layer
while i < len(children) - 1:
current, nxt = children[i:i + 2]
layer, connection = _ann_to_snn_helper(prev, current, node_type,
**kwargs)
i += 1
if layer is None or connection is None:
continue
snn.add_layer(layer, name=str(i))
snn.add_connection(connection, source=str(i - 1), target=str(i))
prev = layer
current = children[-1]
layer, connection = _ann_to_snn_helper(prev,
current,
node_type,
last=True,
**kwargs)
i += 1
if layer is not None or connection is not None:
snn.add_layer(layer, name=str(i))
snn.add_connection(connection, source=str(i - 1), target=str(i))
return snn
def Translate_Into_Networks(input_N, Shape, Output_N, Weight):
network_list = []
path = "gene/"
file_list = os.listdir(path)
gene_file_check = [file for file in file_list if file.endswith(".txt")]
if len(gene_file_check) == 0:
import startup
Gene_List = Genetic.Read_Gene()
for i in range(len(Gene_List)):
network = Network()
Decoded_List = []
Decoded_DNA_List = []
for j in range(len(Gene_List[i])):
Decoded_Gene = Gene_List[i][j].split('-')
if (Decoded_Gene[3] == 'F'):
pass
else:
if Decoded_Gene[1] == '~':
Decoded_List.append(
[int(Decoded_Gene[0]),
int(Decoded_Gene[2]), 0])
elif Decoded_Gene[1] == '!':
Decoded_List.append(
[int(Decoded_Gene[0]),
int(Decoded_Gene[2]), 1])
elif Decoded_Gene[1] == '@':
Decoded_List.append(
[int(Decoded_Gene[0]),
int(Decoded_Gene[2]), 2])
elif Decoded_Gene[1] == '#':
Decoded_List.append(
[int(Decoded_Gene[0]),
int(Decoded_Gene[2]), 3])
elif Decoded_Gene[1] == '$':
Decoded_List.append(
[int(Decoded_Gene[0]),
int(Decoded_Gene[2]), 4])
else:
print("THE GENOTYPE VALUE IS UNVALID")
raise ValueError
Decoded_DNA_List.append(Decoded_List)
Decoded_RNA_List: list = Decoded_DNA_List.copy()
for decoded_dna_idx, decoded_dna in enumerate(Decoded_DNA_List):
Gene_NUM = len(decoded_dna)
for k in range(Gene_NUM):
a = Decoded_DNA_List[decoded_dna_idx][k]
for l in range(k, Gene_NUM):
b = Decoded_DNA_List[decoded_dna_idx][l]
if a and b == 1:
if decoded_dna[k][2] == 0:
Decoded_RNA_List[decoded_dna_idx].remove(
decoded_dna[l])
elif decoded_dna[k][2] == 1:
if decoded_dna[l][2] < 1:
Decoded_RNA_List[decoded_dna_idx].remove(
decoded_dna[k])
else:
Decoded_RNA_List[decoded_dna_idx].remove(
decoded_dna[l])
elif decoded_dna[k][2] == 2:
if decoded_dna[l][2] < 2:
Decoded_RNA_List[decoded_dna_idx].remove(
decoded_dna[k])
else:
Decoded_RNA_List[decoded_dna_idx].remove(
decoded_dna[l])
elif decoded_dna[k][2] == 3:
if decoded_dna[l][2] < 3:
Decoded_RNA_List[decoded_dna_idx].remove(
decoded_dna[k])
else:
Decoded_RNA_List[decoded_dna_idx].remove(
decoded_dna[l])
elif decoded_dna[k][2] == 4:
if decoded_dna[l][2] >= 4:
Decoded_RNA_List[decoded_dna_idx].remove(
decoded_dna[l])
else:
Decoded_RNA_List[decoded_dna_idx].remove(
decoded_dna[k])
else:
pass
else:
pass
for Decoded_RNA in Decoded_RNA_List:
layer_list = {}
for m in range(len(Decoded_RNA)):
for n in range(m, len(Decoded_RNA)):
if Decoded_RNA[m][1] == Decoded_RNA[n][0]:
if Decoded_RNA[n][2] == 0:
layer_list[Decoded_RNA[m][0]] = nodes.IFNodes(
n=1, traces=True)
elif Decoded_RNA[n][2] == 1:
layer_list[Decoded_RNA[m][0]] = nodes.LIFNodes(
n=1, traces=True)
elif Decoded_RNA[n][2] == 2:
layer_list[Decoded_RNA[m]
[0]] = nodes.McCullochPitts(n=1,
traces=True)
elif Decoded_RNA[n][2] == 3:
layer_list[Decoded_RNA[m]
[0]] = nodes.IzhikevichNodes(
n=1, traces=True)
elif Decoded_RNA[n][2] == 4:
layer_list[Decoded_RNA[m][0]] = nodes.SRM0Nodes(
n=1, traces=True)
else:
print("UNVALID GENO_NEURON CODE")
raise ValueError
elif n == len(Decoded_List) - 1:
layer_list[Decoded_RNA[m][1]] = nodes.LIFNodes(n=1)
for l in range(len(Decoded_RNA)):
if not Decoded_RNA[l][0] in layer_list:
if Decoded_RNA[l][2] == 0:
layer_list[Decoded_RNA[l][0]] = nodes.IFNodes(
n=1, traces=True)
elif Decoded_RNA[l][2] == 1:
layer_list[Decoded_RNA[l][0]] = nodes.LIFNodes(
n=1, traces=True)
elif Decoded_RNA[l][2] == 2:
layer_list[Decoded_RNA[l][0]] = nodes.McCullochPitts(
n=1, traces=True)
elif Decoded_RNA[l][2] == 3:
layer_list[Decoded_RNA[l][0]] = nodes.IzhikevichNodes(
n=1, traces=True)
elif Decoded_RNA[l][2] == 4:
layer_list[Decoded_RNA[l][0]] = nodes.SRM0Nodes(
n=1, traces=True)
Input_Layer = nodes.Input(n=input_N, shape=Shape, traces=True)
out = nodes.LIFNodes(n=Output_N, refrac=0, traces=True)
network.add_layer(layer=Input_Layer, name="Input Layer")
for key_l in list(layer_list.keys()):
network.add_layer(layer=layer_list[key_l], name=str(key_l))
network.add_layer(layer=out, name="Output Layer")
if len(layer_list.keys()) == 0:
layer = nodes.LIFNodes(n=1, traces=True)
network.add_layer(layer=layer, name="mid layer")
inpt_connection = Connection(source=Input_Layer,
target=layer,
w=Weight * torch.ones(input_N))
opt_connection = Connection(source=layer,
target=out,
w=Weight * torch.ones(1))
network.add_connection(inpt_connection,
source="Input_Layer",
target="mid layer")
network.add_connection(opt_connection,
source="mid layer",
target="Output Layer")
else:
for key_ic in list(layer_list.keys()):
inpt_connection = Connection(source=Input_Layer,
target=layer_list[key_ic],
w=Weight * torch.ones(input_N))
network.add_connection(inpt_connection,
source="Input_Layer",
target=str(key_ic))
for key_op in list(layer_list.keys()):
output_connection = Connection(source=layer_list[key_op],
target=out,
w=Weight * torch.ones(1),
update_rule=MSTDP)
network.add_connection(output_connection,
source=str(key_op),
target="Output Layer")
for generating_protein in Decoded_RNA:
mid_connection = Connection(
source=layer_list[generating_protein[0]],
target=layer_list[generating_protein[1]],
w=Weight * torch.ones(1),
update_rule=MSTDP)
network.add_connection(mid_connection,
source=str(generating_protein[0]),
target=str(generating_protein[1]))
network_list.append(network)
network.save('Network/' + str(i) + '.pt')
return network_list
from bindsnet.network import Network from bindsnet.network import nodes, topology, monitors from bindsnet.analysis.plotting import plot_spikes # Parameters. n_input = 100 n_output = 100 time = 1000 # Create network object. network = Network() # Create input and output groups of neurons. input_group = nodes.Input(n=n_input) # 100 input nodes. output_group = nodes.LIFNodes(n=n_output) # 500 output nodes. network.add_layer(input_group, name='input') network.add_layer(output_group, name='output') # Input -> output connection. # Unit Gaussian feed-forward weights. w = torch.randn(n_input, n_output) forward_conn = topology.Connection(input_group, output_group, w=w) # Output -> output connection. # Random, inhibitory recurrent weights. w = torch.bernoulli(torch.rand(n_output, n_output)) - torch.diag(torch.ones(n_output)) recurrent_conn = topology.Connection(output_group, output_group, w=w)