def setup_layers(self):
"""
Adding Base Layers, Deep Signed GraphSAGE layers.
Assing Regression Parameters if the model is not a single layer model.
"""
self.nodes = range(self.X.shape[0])
self.neurons = self.args.layers
self.layers = len(self.neurons)
self.positive_base_aggregator = SignedSAGEConvolutionBase(self.X.shape[1]*2,
self.neurons[0]).to(self.device)
self.negative_base_aggregator = SignedSAGEConvolutionBase(self.X.shape[1]*2,
self.neurons[0]).to(self.device)
self.positive_aggregators = []
self.negative_aggregators = []
for i in range(1, self.layers):
self.positive_aggregators.append(SignedSAGEConvolutionDeep(3*self.neurons[i-1],
self.neurons[i]).to(self.device))
self.negative_aggregators.append(SignedSAGEConvolutionDeep(3*self.neurons[i-1],
self.neurons[i]).to(self.device))
self.positive_aggregators = ListModule(*self.positive_aggregators)
self.negative_aggregators = ListModule(*self.negative_aggregators)
self.regression_weights = Parameter(torch.Tensor(4*self.neurons[-1], 3))
init.xavier_normal_(self.regression_weights)
def setup_layers(self):
"""
Adding Base Layers, Deep Signed GraphSAGE layers.
Assing Regression Parameters if the model is not a single layer model.
"""
self.nodes = range(self.X.shape[0])
self.neurons = self.args.layers
self.layers = len(self.neurons)
self.positive_base_aggregator = SignedSAGEConvolutionBase(
(self.X.shape[1] * 2) * 1, self.neurons[0]).to(self.device)
self.negative_base_aggregator = SignedSAGEConvolutionBase(
(self.X.shape[1] * 2) * 1, self.neurons[0]).to(self.device)
self.positive_aggregators = []
self.negative_aggregators = []
for i in range(
1, self.layers
): # different for deep and base - effects weight params
self.positive_aggregators.append(
SignedSAGEConvolutionDeep(
(
4 * self.neurons[i - 1]
), # 3 beacuse original, positive and negative and concatenated, decides the shape of weight matrix
self.neurons[i]).to(self.device))
self.negative_aggregators.append(
SignedSAGEConvolutionDeep((4 * self.neurons[i - 1]),
self.neurons[i]).to(self.device))
self.positive_aggregators = ListModule(*self.positive_aggregators)
self.negative_aggregators = ListModule(*self.negative_aggregators)
self.regression_weights = Parameter(
torch.Tensor(4 * self.neurons[-1], self.neurons[-1]))
self.fc_weights = Parameter(torch.Tensor(self.neurons[-1], 3))
if (self.args.edge_features_incls):
self.regression_weights = Parameter(
torch.Tensor(
4 * (self.neurons[-1] + (edge_features_count - 2) * 4),
self.neurons[-1])) # excluding users
self.fc_weights = Parameter(torch.Tensor(self.neurons[-1], 3))
init.xavier_normal_(self.regression_weights)
init.xavier_normal_(self.fc_weights)
def setup_layers(self):
"""
Adding Base Layers, Deep Signed GraphSAGE layers and Regression Parameters if the model is not a single layer model.
"""
self.nodes = range(self.X.shape[0])
self.neurons = self.args.layers
self.layers = len(self.neurons)
self.positive_base_aggregator_out = SignedSAGEConvolutionBase(
self.X.shape[1] * 2, self.neurons[0], ['out']).to(self.device)
self.negative_base_aggregator_out = SignedSAGEConvolutionBase(
self.X.shape[1] * 2, self.neurons[0], ['out']).to(self.device)
self.positive_base_aggregator_in = SignedSAGEConvolutionBase(
self.X.shape[1] * 2, self.neurons[0], ['in']).to(self.device)
self.negative_base_aggregator_in = SignedSAGEConvolutionBase(
self.X.shape[1] * 2, self.neurons[0], ['in']).to(self.device)
self.outoutFriends_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1],
['out', 'out']).to(self.device)
self.outinFriends_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1],
['out', 'in']).to(self.device)
self.inoutFriends_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1],
['in', 'out']).to(self.device)
self.ininFriends_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1], ['in', 'in']).to(self.device)
self.outoutEnemies_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1],
['out', 'out']).to(self.device)
self.outinEnemies_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1],
['out', 'in']).to(self.device)
self.inoutEnemies_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1],
['in', 'out']).to(self.device)
self.ininEnemies_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1], ['in', 'in']).to(self.device)
self.regression_weights1 = Parameter(
torch.Tensor(8 * self.neurons[-1], 64))
self.regression_weights2 = Parameter(torch.Tensor(64, self.X.shape[1]))
init.xavier_normal_(self.regression_weights1)
init.xavier_normal_(self.regression_weights2)
def setup_layers(self):
"""
Adding Base Layers, Deep Signed GraphSAGE layers and Regression Parameters if the model is not a single layer model.
"""
self.nodes = range(self.X.shape[0])
self.neurons = self.args.layers
self.layers = len(self.neurons)
self.positive_base_aggregator_out = SignedSAGEConvolutionBase(
self.X.shape[1] * 2, self.neurons[0], ['out']).to(self.device)
self.negative_base_aggregator_out = SignedSAGEConvolutionBase(
self.X.shape[1] * 2, self.neurons[0], ['out']).to(self.device)
self.positive_base_aggregator_in = SignedSAGEConvolutionBase(
self.X.shape[1] * 2, self.neurons[0], ['in']).to(self.device)
self.negative_base_aggregator_in = SignedSAGEConvolutionBase(
self.X.shape[1] * 2, self.neurons[0], ['in']).to(self.device)
self.outoutFriends_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1],
['out', 'out']).to(self.device)
self.outinFriends_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1],
['out', 'in']).to(self.device)
self.inoutFriends_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1],
['in', 'out']).to(self.device)
self.ininFriends_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1], ['in', 'in']).to(self.device)
self.outoutEnemies_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1],
['out', 'out']).to(self.device)
self.outinEnemies_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1],
['out', 'in']).to(self.device)
self.inoutEnemies_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1],
['in', 'out']).to(self.device)
self.ininEnemies_aggregator = SignedSAGEConvolutionDeep(
4 * self.neurons[0], self.neurons[1], ['in', 'in']).to(self.device)
# self.positive_aggregators_out = []
# self.negative_aggregators_out = []
# self.positive_aggregators_in = []
# self.negative_aggregators_in = []
# for i in range(1,self.layers):
# self.positive_aggregators_out.append(SignedSAGEConvolutionDeep(6*self.neurons[i-1], self.neurons[i], 'out').to(self.device))
# self.negative_aggregators_out.append(SignedSAGEConvolutionDeep(6*self.neurons[i-1], self.neurons[i], 'out').to(self.device))
# self.positive_aggregators_in.append(SignedSAGEConvolutionDeep(6*self.neurons[i-1], self.neurons[i], 'in').to(self.device))
# self.negative_aggregators_in.append(SignedSAGEConvolutionDeep(6*self.neurons[i-1], self.neurons[i], 'in').to(self.device))
# positive_out = SignedSAGEConvolutionDeep(6*self.neurons[0], self.neurons[0], 'out').to(self.device)
# negative_out = SignedSAGEConvolutionDeep(6*self.neurons[0], self.neurons[0], 'out').to(self.device)
# positive_in = SignedSAGEConvolutionDeep(6*self.neurons[0], self.neurons[0], 'in').to(self.device)
# negative_in= SignedSAGEConvolutionDeep(6*self.neurons[0], self.neurons[0], 'in').to(self.device)
# for i in range(1,self.layers):
# self.positive_aggregators_out.append(positive_out)
# self.negative_aggregators_out.append(negative_out)
# self.positive_aggregators_in.append(positive_in)
# self.negative_aggregators_in.append(negative_in)
# self.positive_aggregators_out = ListModule(*self.positive_aggregators_out)
# self.negative_aggregators_out = ListModule(*self.negative_aggregators_out)
# self.positive_aggregators_in = ListModule(*self.positive_aggregators_in)
# self.negative_aggregators_in = ListModule(*self.negative_aggregators_in)
# self.regression_weights = Parameter(torch.Tensor(4*self.neurons[-1], 3))
# self.regression_weights = Parameter(torch.Tensor(8*self.neurons[-1], 2))
self.regression_weights = Parameter(
torch.Tensor(8 * self.neurons[-1], 1))
init.xavier_normal_(self.regression_weights)