AutoGL/test/performance/graph_classification/dgl/base.py

"""
Performance check of DGL original dataset, model, trainer setting

Borrowed from DGL official examples: https://github.com/dmlc/dgl/tree/master/examples/pytorch/gin

TopkPool is not supported currently
"""

# from dgl.dataloading.pytorch.dataloader import GraphDataLoader
import pickle
from dgl.dataloading import GraphDataLoader
import numpy as np
from tqdm import tqdm

import random

import torch
import torch.nn as nn
import torch.optim as optim

from dgl.data import GINDataset

import torch
import torch.nn as nn
import torch.nn.functional as F
from dgl.nn.pytorch.conv import GINConv
from dgl.nn.pytorch.glob import SumPooling, AvgPooling, MaxPooling

def set_seed(seed=None):
    """
    Set seed of whole process
    """
    if seed is None:
        seed = random.randint(0, 5000)

    random.seed(seed)
    np.random.seed(seed)
    torch.manual_seed(seed)
    if torch.cuda.is_available():
        torch.cuda.manual_seed_all(seed)
        torch.backends.cudnn.deterministic = True
        torch.backends.cudnn.benchmark = False

class DatasetAbstraction():
    def __init__(self, graphs, labels):
        for g in graphs:
            g.ndata['feat'] = g.ndata['attr']
        self.graphs, self.labels = [], []
        for g, l in zip(graphs, labels):
            self.graphs.append(g)
            self.labels.append(l)
        self.gclasses = max(self.labels).item() + 1
        self.graph = self.graphs
    
    def __len__(self):
        return len(self.graphs)
    
    def __getitem__(self, idx):
        if isinstance(idx, int):
            return self.graphs[idx], self.labels[idx]
        elif isinstance(idx, torch.BoolTensor):
            idx = [i for i in range(len(idx)) if idx[i]]
        elif isinstance(idx, torch.Tensor) and idx.unique()[0].sum().item() == 1:
            idx = [i for i in range(len(idx)) if idx[i]]
        return DatasetAbstraction([self.graphs[i] for i in idx], [self.labels[i] for i in idx])

class ApplyNodeFunc(nn.Module):
    """Update the node feature hv with MLP, BN and ReLU."""
    def __init__(self, mlp):
        super(ApplyNodeFunc, self).__init__()
        self.mlp = mlp
        self.bn = nn.BatchNorm1d(self.mlp.output_dim)

    def forward(self, h):
        h = self.mlp(h)
        h = self.bn(h)
        h = F.relu(h)
        return h


class MLP(nn.Module):
    """MLP with linear output"""
    def __init__(self, num_layers, input_dim, hidden_dim, output_dim):
        """MLP layers construction
        Paramters
        ---------
        num_layers: int
            The number of linear layers
        input_dim: int
            The dimensionality of input features
        hidden_dim: int
            The dimensionality of hidden units at ALL layers
        output_dim: int
            The number of classes for prediction
        """
        super(MLP, self).__init__()
        self.linear_or_not = True  # default is linear model
        self.num_layers = num_layers
        self.output_dim = output_dim

        if num_layers < 1:
            raise ValueError("number of layers should be positive!")
        elif num_layers == 1:
            # Linear model
            self.linear = nn.Linear(input_dim, output_dim)
        else:
            # Multi-layer model
            self.linear_or_not = False
            self.linears = torch.nn.ModuleList()
            self.batch_norms = torch.nn.ModuleList()

            self.linears.append(nn.Linear(input_dim, hidden_dim))
            for layer in range(num_layers - 2):
                self.linears.append(nn.Linear(hidden_dim, hidden_dim))
            self.linears.append(nn.Linear(hidden_dim, output_dim))

            for layer in range(num_layers - 1):
                self.batch_norms.append(nn.BatchNorm1d((hidden_dim)))

    def forward(self, x):
        if self.linear_or_not:
            # If linear model
            return self.linear(x)
        else:
            # If MLP
            h = x
            for i in range(self.num_layers - 1):
                h = F.relu(self.batch_norms[i](self.linears[i](h)))
            return self.linears[-1](h)


class GIN(nn.Module):
    """GIN model"""
    def __init__(self, num_layers, num_mlp_layers, input_dim, hidden_dim,
                 output_dim, final_dropout, learn_eps, graph_pooling_type,
                 neighbor_pooling_type):
        """model parameters setting
        Paramters
        ---------
        num_layers: int
            The number of linear layers in the neural network
        num_mlp_layers: int
            The number of linear layers in mlps
        input_dim: int
            The dimensionality of input features
        hidden_dim: int
            The dimensionality of hidden units at ALL layers
        output_dim: int
            The number of classes for prediction
        final_dropout: float
            dropout ratio on the final linear layer
        learn_eps: boolean
            If True, learn epsilon to distinguish center nodes from neighbors
            If False, aggregate neighbors and center nodes altogether.
        neighbor_pooling_type: str
            how to aggregate neighbors (sum, mean, or max)
        graph_pooling_type: str
            how to aggregate entire nodes in a graph (sum, mean or max)
        """
        super(GIN, self).__init__()
        self.num_layers = num_layers
        self.learn_eps = learn_eps

        # List of MLPs
        self.ginlayers = torch.nn.ModuleList()
        self.batch_norms = torch.nn.ModuleList()

        for layer in range(self.num_layers - 1):
            if layer == 0:
                mlp = MLP(num_mlp_layers, input_dim, hidden_dim, hidden_dim)
            else:
                mlp = MLP(num_mlp_layers, hidden_dim, hidden_dim, hidden_dim)

            self.ginlayers.append(
                GINConv(ApplyNodeFunc(mlp), neighbor_pooling_type, 0, self.learn_eps))
            self.batch_norms.append(nn.BatchNorm1d(hidden_dim))

        # Linear function for graph poolings of output of each layer
        # which maps the output of different layers into a prediction score
        self.linears_prediction = torch.nn.ModuleList()

        for layer in range(num_layers):
            if layer == 0:
                self.linears_prediction.append(
                    nn.Linear(input_dim, output_dim))
            else:
                self.linears_prediction.append(
                    nn.Linear(hidden_dim, output_dim))

        self.drop = nn.Dropout(final_dropout)

        if graph_pooling_type == 'sum':
            self.pool = SumPooling()
        elif graph_pooling_type == 'mean':
            self.pool = AvgPooling()
        elif graph_pooling_type == 'max':
            self.pool = MaxPooling()
        else:
            raise NotImplementedError

    def forward(self, g, h):
        # list of hidden representation at each layer (including input)
        hidden_rep = [h]

        for i in range(self.num_layers - 1):
            h = self.ginlayers[i](g, h)
            h = self.batch_norms[i](h)
            h = F.relu(h)
            hidden_rep.append(h)

        score_over_layer = 0

        # perform pooling over all nodes in each graph in every layer
        for i, h in enumerate(hidden_rep):
            pooled_h = self.pool(g, h)
            score_over_layer += self.drop(self.linears_prediction[i](pooled_h))

        return score_over_layer


def train(net, trainloader, validloader, optimizer, criterion, epoch, device):
    best_model = pickle.dumps(net.state_dict())
    
    best_acc = 0.
    for e in range(epoch):
        net.train()
        for graphs, labels in trainloader:

            labels = labels.to(device)
            graphs = graphs.to(device)
            feat = graphs.ndata.pop('attr')
            outputs = net(graphs, feat)

            loss = criterion(outputs, labels)

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

        
        gt = []
        pr = []
        net.eval()
        for graphs, labels in validloader:
            labels = labels.to(device)
            graphs = graphs.to(device)
            gt.append(labels)
            feat = graphs.ndata.pop('attr')
            outputs = net(graphs, feat)
            pr.append(outputs.argmax(1))
        gt = torch.cat(gt, dim=0)
        pr = torch.cat(pr, dim=0)
        acc = (gt == pr).float().mean().item()
        if acc > best_acc:
            best_acc = acc
            best_model = pickle.dumps(net.state_dict())
    
    net.load_state_dict(pickle.loads(best_model))

    return net

def eval_net(net, dataloader, device):
    net.eval()

    total = 0
    total_correct = 0

    for data in dataloader:
        graphs, labels = data
        graphs = graphs.to(device)
        labels = labels.to(device)
        feat = graphs.ndata.pop('attr')
        total += len(labels)
        outputs = net(graphs, feat)
        _, predicted = torch.max(outputs.data, 1)

        total_correct += (predicted == labels.data).sum().item()

    acc = 1.0 * total_correct / total

    net.train()

    return acc


def main():

    import argparse
    parser = argparse.ArgumentParser()
    parser.add_argument("--repeat", type=int, default=10)
    parser.add_argument('--dataset', type=str, choices=['MUTAG', 'COLLAB', 'IMDBBINARY', 'IMDBMULTI', 'NCI1', 'PROTEINS', 'PTC', 'REDDITBINARY', 'REDDITMULTI5K'], default='MUTAG')

    args = parser.parse_args()

    device = torch.device('cuda')
    dataset_ = GINDataset(args.dataset, False)
    dataset = DatasetAbstraction([g[0] for g in dataset_], [g[1] for g in dataset_])
    
    # 1. split dataset [fix split]
    dataids = list(range(len(dataset)))
    random.seed(2021)
    random.shuffle(dataids)
    
    fold = int(len(dataset) * 0.1)
    train_dataset = dataset[dataids[:fold * 8]]
    val_dataset = dataset[dataids[fold * 8: fold * 9]]
    test_dataset = dataset[dataids[fold * 9: ]]

    trainloader = GraphDataLoader(train_dataset, batch_size=32, shuffle=True)
    valloader = GraphDataLoader(val_dataset, batch_size=32, shuffle=False)
    testloader = GraphDataLoader(test_dataset, batch_size=32, shuffle=False)

    accs = []
    for seed in tqdm(range(args.repeat)):
        # set up seeds, args.seed supported
        set_seed(seed)

        model = GIN(
            5, 2, dataset_.dim_nfeats, 64, dataset_.gclasses, 0.5, False,
            "sum", "sum").to(device)

        criterion = nn.CrossEntropyLoss()  # defaul reduce is true
        optimizer = optim.Adam(model.parameters(), lr=0.0001)

        model = train(model, trainloader, valloader, optimizer, criterion, 100, device)
        acc = eval_net(model, testloader, device)
        accs.append(acc)

    print('{:.2f} ~ {:.2f}'.format(np.mean(accs) * 100, np.std(accs) * 100))

if __name__ == '__main__':
    main()