gcn_spmv.py

"""
Semi-Supervised Classification with Graph Convolutional Networks
Paper: https://arxiv.org/abs/1609.02907
Code: https://github.com/tkipf/gcn

GCN with SPMV specialization.
"""
import argparse
import numpy as np
import time
import torch
import torch.nn as nn
import torch.nn.functional as F
import dgl.function as fn
from dgl import DGLGraph
from dgl.data import register_data_args, load_data

class NodeApplyModule(nn.Module):
    def __init__(self, in_feats, out_feats, activation=None):
        super(NodeApplyModule, self).__init__()
        self.linear = nn.Linear(in_feats, out_feats)
        nn.init.xavier_normal_(self.linear.weight)
        self.activation = activation

    def forward(self, nodes):
        # normalization by square root of dst degree
        h = nodes.data['h'] * nodes.data['norm']
        h = self.linear(h)
        if self.activation:
            h = self.activation(h)
        return {'h': h}

class GCN(nn.Module):
    def __init__(self,
                 g,
                 in_feats,
                 n_hidden,
                 n_classes,
                 n_layers,
                 activation,
                 dropout):
        super(GCN, self).__init__()
        self.g = g

        if dropout:
            self.dropout = nn.Dropout(p=dropout)
        else:
            self.dropout = 0.

        self.layers = nn.ModuleList()

        # input layer
        self.layers.append(NodeApplyModule(in_feats, n_hidden, activation))

        # hidden layers
        for i in range(n_layers - 1):
            self.layers.append(NodeApplyModule(n_hidden, n_hidden, activation))

        # output layer
        self.layers.append(NodeApplyModule(n_hidden, n_classes))

    def forward(self, features):
        self.g.ndata['h'] = features

        for idx, layer in enumerate(self.layers):
            # apply dropout
            if idx > 0 and self.dropout:
                self.g.ndata['h'] = self.dropout(self.g.ndata['h'])
            # normalization by square root of src degree
            self.g.ndata['h'] = self.g.ndata['h'] * self.g.ndata['norm']
            self.g.update_all(fn.copy_src(src='h', out='m'),
                              fn.sum(msg='m', out='h'),
                              layer)
        return self.g.pop_n_repr('h')

def evaluate(model, features, labels, mask):
    model.eval()
    with torch.no_grad():
        logits = model(features)
        logits = logits[mask]
        labels = labels[mask]
        _, indices = torch.max(logits, dim=1)
        correct = torch.sum(indices == labels)
        return correct.item() * 1.0 / len(labels)

def main(args):
    # load and preprocess dataset
    data = load_data(args)
    features = torch.FloatTensor(data.features)
    labels = torch.LongTensor(data.labels)
    train_mask = torch.ByteTensor(data.train_mask)
    val_mask = torch.ByteTensor(data.val_mask)
    test_mask = torch.ByteTensor(data.test_mask)
    in_feats = features.shape[1]
    n_classes = data.num_labels
    n_edges = data.graph.number_of_edges()

    if args.gpu < 0:
        cuda = False
    else:
        cuda = True
        torch.cuda.set_device(args.gpu)
        features = features.cuda()
        labels = labels.cuda()
        train_mask = train_mask.cuda()
        val_mask = val_mask.cuda()
        test_mask = test_mask.cuda()

    # graph preprocess and calculate normalization factor
    g = DGLGraph(data.graph)
    n_edges = g.number_of_edges()
    # add self loop
    g.add_edges(g.nodes(), g.nodes())
    # normalization
    degs = g.in_degrees().float()
    norm = torch.pow(degs, -0.5)
    norm[torch.isinf(norm)] = 0
    if cuda:
        norm = norm.cuda()
    g.ndata['norm'] = norm.unsqueeze(1)

    # create GCN model
    model = GCN(g,
                in_feats,
                args.n_hidden,
                n_classes,
                args.n_layers,
                F.relu,
                args.dropout)

    if cuda:
        model.cuda()

    # use optimizer
    optimizer = torch.optim.Adam(model.parameters(),
                                 lr=args.lr,
                                 weight_decay=args.weight_decay)

    # initialize graph
    dur = []
    for epoch in range(args.n_epochs):
        model.train()
        if epoch >= 3:
            t0 = time.time()
        # forward
        logits = model(features)
        logp = F.log_softmax(logits, 1)
        loss = F.nll_loss(logp[train_mask], labels[train_mask])

        optimizer.zero_grad()
        loss.backward()
        optimizer.step()

        if epoch >= 3:
            dur.append(time.time() - t0)

        acc = evaluate(model, features, labels, val_mask)
        print("Epoch {:05d} | Time(s) {:.4f} | Loss {:.4f} | Accuracy {:.4f} | "
              "ETputs(KTEPS) {:.2f}". format(epoch, np.mean(dur), loss.item(),
                                             acc, n_edges / np.mean(dur) / 1000))

    print()
    acc = evaluate(model, features, labels, test_mask)
    print("Test Accuracy {:.4f}".format(acc))


if __name__ == '__main__':
    parser = argparse.ArgumentParser(description='GCN')
    register_data_args(parser)
    parser.add_argument("--dropout", type=float, default=0.5,
            help="dropout probability")
    parser.add_argument("--gpu", type=int, default=-1,
            help="gpu")
    parser.add_argument("--lr", type=float, default=1e-2,
            help="learning rate")
    parser.add_argument("--n-epochs", type=int, default=200,
            help="number of training epochs")
    parser.add_argument("--n-hidden", type=int, default=16,
            help="number of hidden gcn units")
    parser.add_argument("--n-layers", type=int, default=1,
            help="number of hidden gcn layers")
    parser.add_argument("--weight-decay", type=float, default=5e-4,
            help="Weight for L2 loss")
    args = parser.parse_args()
    print(args)

    main(args)