"""SGCN runner."""
import time
import torch
import random
import numpy as np
import pandas as pd
from tqdm import trange
import torch.nn.init as init
from torch.nn import Parameter
import torch.nn.functional as F
from utils import calculate_auc, setup_features
from sklearn.model_selection import train_test_split
from signedsageconvolution import SignedSAGEConvolutionBase, SignedSAGEConvolutionDeep
from signedsageconvolution import ListModule
class SignedGraphConvolutionalNetwork(torch.nn.Module):
"""
Signed Graph Convolutional Network Class.
For details see: Signed Graph Convolutional Network.
Tyler Derr, Yao Ma, and Jiliang Tang ICDM, 2018.
https://arxiv.org/abs/1808.06354
"""
def __init__(self, device, args, X):
super(SignedGraphConvolutionalNetwork, self).__init__()
"""
SGCN Initialization.
:param device: Device for calculations.
:param args: Arguments object.
:param X: Node features.
"""
self.args = args
torch.manual_seed(self.args.seed)
self.device = device
self.X = X
self.setup_layers()
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 calculate_regression_loss(self, z, target):
"""
Calculating the regression loss for all pairs of nodes.
:param z: Hidden vertex representations.
:param target: Target vector.
:return loss_term: Regression loss.
:return predictions_soft: Predictions for each vertex pair.
"""
pos = torch.cat((self.positive_z_i, self.positive_z_j), 1)
neg = torch.cat((self.negative_z_i, self.negative_z_j), 1)
surr_neg_i = torch.cat((self.negative_z_i, self.negative_z_k), 1)
surr_neg_j = torch.cat((self.negative_z_j, self.negative_z_k), 1)
surr_pos_i = torch.cat((self.positive_z_i, self.positive_z_k), 1)
surr_pos_j = torch.cat((self.positive_z_j, self.positive_z_k), 1)
features = torch.cat((pos, neg, surr_neg_i, surr_neg_j, surr_pos_i, surr_pos_j))
predictions = torch.mm(features, self.regression_weights)
predictions_soft = F.log_softmax(predictions, dim=1)
loss_term = F.nll_loss(predictions_soft, target)
return loss_term, predictions_soft
def calculate_positive_embedding_loss(self, z, positive_edges):
"""
Calculating the loss on the positive edge embedding distances
:param z: Hidden vertex representation.
:param positive_edges: Positive training edges.
:return loss_term: Loss value on positive edge embedding.
"""
self.positive_surrogates = [random.choice(self.nodes) for node in range(positive_edges.shape[1])]
self.positive_surrogates = torch.from_numpy(np.array(self.positive_surrogates, dtype=np.int64).T)
self.positive_surrogates = self.positive_surrogates.type(torch.long).to(self.device)
positive_edges = torch.t(positive_edges)
self.positive_z_i = z[positive_edges[:, 0], :]
self.positive_z_j = z[positive_edges[:, 1], :]
self.positive_z_k = z[self.positive_surrogates, :]
norm_i_j = torch.norm(self.positive_z_i-self.positive_z_j, 2, 1, True).pow(2)
norm_i_k = torch.norm(self.positive_z_i-self.positive_z_k, 2, 1, True).pow(2)
term = norm_i_j-norm_i_k
term[term < 0] = 0
loss_term = term.mean()
return loss_term
def calculate_negative_embedding_loss(self, z, negative_edges):
"""
Calculating the loss on the negative edge embedding distances
:param z: Hidden vertex representation.
:param negative_edges: Negative training edges.
:return loss_term: Loss value on negative edge embedding.
"""
self.negative_surrogates = [random.choice(self.nodes) for node in range(negative_edges.shape[1])]
self.negative_surrogates = torch.from_numpy(np.array(self.negative_surrogates, dtype=np.int64).T)
self.negative_surrogates = self.negative_surrogates.type(torch.long).to(self.device)
negative_edges = torch.t(negative_edges)
self.negative_z_i = z[negative_edges[:, 0], :]
self.negative_z_j = z[negative_edges[:, 1], :]
self.negative_z_k = z[self.negative_surrogates, :]
norm_i_j = torch.norm(self.negative_z_i-self.negative_z_j, 2, 1, True).pow(2)
norm_i_k = torch.norm(self.negative_z_i-self.negative_z_k, 2, 1, True).pow(2)
term = norm_i_k-norm_i_j
term[term < 0] = 0
loss_term = term.mean()
return loss_term
def calculate_loss_function(self, z, positive_edges, negative_edges, target):
"""
Calculating the embedding losses, regression loss and weight regularization loss.
:param z: Node embedding.
:param positive_edges: Positive edge pairs.
:param negative_edges: Negative edge pairs.
:param target: Target vector.
:return loss: Value of loss.
"""
loss_term_1 = self.calculate_positive_embedding_loss(z, positive_edges)
loss_term_2 = self.calculate_negative_embedding_loss(z, negative_edges)
regression_loss, self.predictions = self.calculate_regression_loss(z, target)
loss_term = regression_loss+self.args.lamb*(loss_term_1+loss_term_2)
return loss_term
def forward(self, positive_edges, negative_edges, target):
"""
Model forward propagation pass. Can fit deep and single layer SGCN models.
:param positive_edges: Positive edges.
:param negative_edges: Negative edges.
:param target: Target vectors.
:return loss: Loss value.
:return self.z: Hidden vertex representations.
"""
self.h_pos, self.h_neg = [], []
self.h_pos.append(torch.tanh(self.positive_base_aggregator(self.X, positive_edges)))
self.h_neg.append(torch.tanh(self.negative_base_aggregator(self.X, negative_edges)))
for i in range(1, self.layers):
self.h_pos.append(torch.tanh(self.positive_aggregators[i-1](self.h_pos[i-1], self.h_neg[i-1], positive_edges, negative_edges)))
self.h_neg.append(torch.tanh(self.negative_aggregators[i-1](self.h_neg[i-1], self.h_pos[i-1], positive_edges, negative_edges)))
self.z = torch.cat((self.h_pos[-1], self.h_neg[-1]), 1)
loss = self.calculate_loss_function(self.z, positive_edges, negative_edges, target)
return loss, self.z
class SignedGCNTrainer(object):
"""
Object to train and score the SGCN, log the model behaviour and save the output.
"""
def __init__(self, args, edges):
"""
Constructing the trainer instance and setting up logs.
:param args: Arguments object.
:param edges: Edge data structure with positive and negative edges separated.
"""
self.args = args
self.edges = edges
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
self.setup_logs()
def setup_logs(self):
"""
Creating a log dictionary.
"""
self.logs = {}
self.logs["parameters"] = vars(self.args)
self.logs["performance"] = [["Epoch", "AUC", "F1"]]
self.logs["training_time"] = [["Epoch", "Seconds"]]
def setup_dataset(self):
"""
Creating train and test split.
"""
self.positive_edges, self.test_positive_edges = train_test_split(self.edges["positive_edges"],
test_size=self.args.test_size)
self.negative_edges, self.test_negative_edges = train_test_split(self.edges["negative_edges"],
test_size=self.args.test_size)
self.ecount = len(self.positive_edges + self.negative_edges)
self.X = setup_features(self.args,
self.positive_edges,
self.negative_edges,
self.edges["ncount"])
self.positive_edges = torch.from_numpy(np.array(self.positive_edges,
dtype=np.int64).T).type(torch.long).to(self.device)
self.negative_edges = torch.from_numpy(np.array(self.negative_edges,
dtype=np.int64).T).type(torch.long).to(self.device)
self.y = np.array([0 if i < int(self.ecount/2) else 1 for i in range(self.ecount)]+[2]*(self.ecount*2))
self.y = torch.from_numpy(self.y).type(torch.LongTensor).to(self.device)
self.X = torch.from_numpy(self.X).float().to(self.device)
def score_model(self, epoch):
"""
Score the model on the test set edges in each epoch.
:param epoch: Epoch number.
"""
loss, self.train_z = self.model(self.positive_edges, self.negative_edges, self.y)
score_positive_edges = torch.from_numpy(np.array(self.test_positive_edges, dtype=np.int64).T).type(torch.long).to(self.device)
score_negative_edges = torch.from_numpy(np.array(self.test_negative_edges, dtype=np.int64).T).type(torch.long).to(self.device)
test_positive_z = torch.cat((self.train_z[score_positive_edges[0, :], :], self.train_z[score_positive_edges[1, :], :]), 1)
test_negative_z = torch.cat((self.train_z[score_negative_edges[0, :], :], self.train_z[score_negative_edges[1, :], :]), 1)
scores = torch.mm(torch.cat((test_positive_z, test_negative_z), 0), self.model.regression_weights.to(self.device))
probability_scores = torch.exp(F.softmax(scores, dim=1))
predictions = probability_scores[:, 0]/probability_scores[:, 0:2].sum(1)
predictions = predictions.cpu().detach().numpy()
targets = [0]*len(self.test_positive_edges) + [1]*len(self.test_negative_edges)
auc, f1 = calculate_auc(targets, predictions, self.edges)
self.logs["performance"].append([epoch+1, auc, f1])
def create_and_train_model(self):
"""
Model training and scoring.
"""
print("\nTraining started.\n")
self.model = SignedGraphConvolutionalNetwork(self.device, self.args, self.X).to(self.device)
self.optimizer = torch.optim.Adam(self.model.parameters(),
lr=self.args.learning_rate,
weight_decay=self.args.weight_decay)
self.model.train()
self.epochs = trange(self.args.epochs, desc="Loss")
for epoch in self.epochs:
start_time = time.time()
self.optimizer.zero_grad()
loss, _ = self.model(self.positive_edges, self.negative_edges, self.y)
loss.backward()
self.epochs.set_description("SGCN (Loss=%g)" % round(loss.item(), 4))
self.optimizer.step()
self.logs["training_time"].append([epoch+1, time.time()-start_time])
if self.args.test_size > 0:
self.score_model(epoch)
def save_model(self):
"""
Saving the embedding and model weights.
"""
print("\nEmbedding is saved.\n")
self.train_z = self.train_z.cpu().detach().numpy()
embedding_header = ["id"] + ["x_"+str(x) for x in range(self.train_z.shape[1])]
self.train_z = np.concatenate([np.array(range(self.train_z.shape[0])).reshape(-1, 1), self.train_z], axis=1)
self.train_z = pd.DataFrame(self.train_z, columns=embedding_header)
self.train_z.to_csv(self.args.embedding_path, index=None)
print("\nRegression weights are saved.\n")
self.regression_weights = self.model.regression_weights.cpu().detach().numpy().T
regression_header = ["x_" + str(x) for x in range(self.regression_weights.shape[1])]
self.regression_weights = pd.DataFrame(self.regression_weights, columns=regression_header)
self.regression_weights.to_csv(self.args.regression_weights_path, index=None)
