import os import os.path as osp import shutil import torch from torch_geometric.data import InMemoryDataset, download_url, extract_zip from torch_geometric.io import read_tu_data from itertools import repeat, product import numpy as np from copy import deepcopy import pdb class TUDataset_aug(InMemoryDataset): r"""A variety of graph kernel benchmark datasets, *.e.g.* "IMDB-BINARY", "REDDIT-BINARY" or "PROTEINS", collected from the `TU Dortmund University `_. `_. In addition, this dataset wrapper provides `cleaned dataset versions `_ as motivated by the `"Understanding Isomorphism Bias in Graph Data Sets" `_ paper, containing only non-isomorphic graphs. .. note:: Some datasets may not come with any node labels. You can then either make use of the argument :obj:`use_node_attr` to load additional continuous node attributes (if present) or provide synthetic node features using transforms such as like :class:`torch_geometric.transforms.Constant` or :class:`torch_geometric.transforms.OneHotDegree`. Args: root (string): Root directory where the dataset should be saved. name (string): The `name `_ of the dataset. transform (callable, optional): A function/transform that takes in an :obj:`torch_geometric.data.Data` object and returns a transformed version. The data object will be transformed before every access. (default: :obj:`None`) pre_transform (callable, optional): A function/transform that takes in an :obj:`torch_geometric.data.Data` object and returns a transformed version. The data object will be transformed before being saved to disk. (default: :obj:`None`) pre_filter (callable, optional): A function that takes in an :obj:`torch_geometric.data.Data` object and returns a boolean value, indicating whether the data object should be included in the final dataset. (default: :obj:`None`) use_node_attr (bool, optional): If :obj:`True`, the dataset will contain additional continuous node attributes (if present). (default: :obj:`False`) use_edge_attr (bool, optional): If :obj:`True`, the dataset will contain additional continuous edge attributes (if present). (default: :obj:`False`) cleaned: (bool, optional): If :obj:`True`, the dataset will contain only non-isomorphic graphs. (default: :obj:`False`) """ url = ('http://ls11-www.cs.tu-dortmund.de/people/morris/' 'graphkerneldatasets') cleaned_url = ('https://raw.githubusercontent.com/nd7141/' 'graph_datasets/master/datasets') def __init__(self, root, name, transform=None, pre_transform=None, pre_filter=None, use_node_attr=False, use_edge_attr=False, cleaned=False, aug=None): self.name = name self.cleaned = cleaned super(TUDataset_aug, self).__init__(root, transform, pre_transform, pre_filter) self.data, self.slices = torch.load(self.processed_paths[0]) if self.data.x is not None and not use_node_attr: num_node_attributes = self.num_node_attributes self.data.x = self.data.x[:, num_node_attributes:] if self.data.edge_attr is not None and not use_edge_attr: num_edge_attributes = self.num_edge_attributes self.data.edge_attr = self.data.edge_attr[:, num_edge_attributes:] if not (self.name == 'MUTAG' or self.name == 'PTC_MR' or self.name == 'DD' or self.name == 'PROTEINS' or self.name == 'NCI1' or self.name == 'NCI109'): edge_index = self.data.edge_index[0, :].numpy() _, num_edge = self.data.edge_index.size() nlist = [edge_index[n] + 1 for n in range(num_edge - 1) if edge_index[n] > edge_index[n + 1]] nlist.append(edge_index[-1] + 1) num_node = np.array(nlist).sum() self.data.x = torch.ones((num_node, 1)) edge_slice = [0] k = 0 for n in nlist: k = k + n edge_slice.append(k) self.slices['x'] = torch.tensor(edge_slice) ''' print(self.data.x.size()) print(self.slices['x']) print(self.slices['x'].size()) assert False ''' self.aug = aug @property def raw_dir(self): name = 'raw{}'.format('_cleaned' if self.cleaned else '') return osp.join(self.root, self.name, name) @property def processed_dir(self): name = 'processed{}'.format('_cleaned' if self.cleaned else '') return osp.join(self.root, self.name, name) @property def num_node_labels(self): if self.data.x is None: return 0 for i in range(self.data.x.size(1)): x = self.data.x[:, i:] if ((x == 0) | (x == 1)).all() and (x.sum(dim=1) == 1).all(): return self.data.x.size(1) - i return 0 @property def num_node_attributes(self): if self.data.x is None: return 0 return self.data.x.size(1) - self.num_node_labels @property def num_edge_labels(self): if self.data.edge_attr is None: return 0 for i in range(self.data.edge_attr.size(1)): if self.data.edge_attr[:, i:].sum() == self.data.edge_attr.size(0): return self.data.edge_attr.size(1) - i return 0 @property def num_edge_attributes(self): if self.data.edge_attr is None: return 0 return self.data.edge_attr.size(1) - self.num_edge_labels @property def raw_file_names(self): names = ['A', 'graph_indicator'] return ['{}_{}.txt'.format(self.name, name) for name in names] @property def processed_file_names(self): return 'data.pt' def download(self): url = self.cleaned_url if self.cleaned else self.url folder = osp.join(self.root, self.name) path = download_url('{}/{}.zip'.format(url, self.name), folder) extract_zip(path, folder) os.unlink(path) shutil.rmtree(self.raw_dir) os.rename(osp.join(folder, self.name), self.raw_dir) def process(self): self.data, self.slices,_ = read_tu_data(self.raw_dir, self.name) if self.pre_filter is not None: data_list = [self.get(idx) for idx in range(len(self))] data_list = [data for data in data_list if self.pre_filter(data)] self.data, self.slices = self.collate(data_list) if self.pre_transform is not None: data_list = [self.get(idx) for idx in range(len(self))] data_list = [self.pre_transform(data) for data in data_list] self.data, self.slices = self.collate(data_list) torch.save((self.data, self.slices), self.processed_paths[0]) def __repr__(self): return '{}({})'.format(self.name, len(self)) def get_num_feature(self): data = self.data.__class__() if hasattr(self.data, '__num_nodes__'): data.num_nodes = self.data.__num_nodes__[0] # for key in self.data.keys: for key in self.slices.keys(): item, slices = self.data[key], self.slices[key] if torch.is_tensor(item): s = list(repeat(slice(None), item.dim())) s[self.data.__cat_dim__(key, item)] = slice(slices[0], slices[0 + 1]) else: s = slice(slices[idx], slices[idx + 1]) data[key] = item[s] _, num_feature = data.x.size() return num_feature def get(self, idx): data = self.data.__class__() if hasattr(self.data, '__num_nodes__'): data.num_nodes = self.data.__num_nodes__[idx] # for key in self.data.keys: for key in self.slices.keys(): item, slices = self.data[key], self.slices[key] if torch.is_tensor(item): s = list(repeat(slice(None), item.dim())) s[self.data.__cat_dim__(key, item)] = slice(slices[idx], slices[idx + 1]) else: s = slice(slices[idx], slices[idx + 1]) data[key] = item[s] """ edge_index = data.edge_index node_num = data.x.size()[0] edge_num = data.edge_index.size()[1] data.edge_index = torch.tensor([[edge_index[0, n], edge_index[1, n]] for n in range(edge_num) if edge_index[0, n] < node_num and edge_index[1, n] < node_num] + [[n, n] for n in range(node_num)], dtype=torch.int64).t() """ node_num = data.edge_index.max() sl = torch.tensor([[n,n] for n in range(node_num)]).t() data.edge_index = torch.cat((data.edge_index, sl), dim=1) if self.aug == 'dnodes': data_aug = drop_nodes(deepcopy(data)) elif self.aug == 'pedges': data_aug = permute_edges(deepcopy(data)) elif self.aug == 'subgraph': data_aug = subgraph(deepcopy(data)) elif self.aug == 'mask_nodes': data_aug = mask_nodes(deepcopy(data)) elif self.aug == 'none': """ if data.edge_index.max() > data.x.size()[0]: print(data.edge_index) print(data.x.size()) assert False """ data_aug = deepcopy(data) data_aug.x = torch.ones((data.edge_index.max()+1, 1)) elif self.aug == 'random2': n = np.random.randint(2) if n == 0: data_aug = drop_nodes(deepcopy(data)) elif n == 1: data_aug = subgraph(deepcopy(data)) else: print('sample error') assert False elif self.aug == 'random3': n = np.random.randint(3) if n == 0: data_aug = drop_nodes(deepcopy(data)) elif n == 1: data_aug = permute_edges(deepcopy(data)) elif n == 2: data_aug = subgraph(deepcopy(data)) else: print('sample error') assert False elif self.aug == 'random4': n = np.random.randint(4) if n == 0: data_aug = drop_nodes(deepcopy(data)) elif n == 1: data_aug = permute_edges(deepcopy(data)) elif n == 2: data_aug = subgraph(deepcopy(data)) elif n == 3: data_aug = mask_nodes(deepcopy(data)) else: print('sample error') assert False else: print('augmentation error') assert False # print(data, data_aug) # assert False return data, data_aug def drop_nodes(data): node_num, _ = data.x.size() _, edge_num = data.edge_index.size() drop_num = int(node_num / 10) idx_drop = np.random.choice(node_num, drop_num, replace=False) idx_nondrop = [n for n in range(node_num) if not n in idx_drop] idx_dict = {idx_nondrop[n]:n for n in list(range(node_num - drop_num))} # data.x = data.x[idx_nondrop] edge_index = data.edge_index.numpy() adj = torch.zeros((node_num, node_num)) adj[edge_index[0], edge_index[1]] = 1 adj[idx_drop, :] = 0 adj[:, idx_drop] = 0 edge_index = adj.nonzero().t() data.edge_index = edge_index # edge_index = [[idx_dict[edge_index[0, n]], idx_dict[edge_index[1, n]]] for n in range(edge_num) if (not edge_index[0, n] in idx_drop) and (not edge_index[1, n] in idx_drop)] # edge_index = [[edge_index[0, n], edge_index[1, n]] for n in range(edge_num) if (not edge_index[0, n] in idx_drop) and (not edge_index[1, n] in idx_drop)] + [[n, n] for n in idx_nondrop] # data.edge_index = torch.tensor(edge_index).transpose_(0, 1) return data def permute_edges(data): node_num, _ = data.x.size() _, edge_num = data.edge_index.size() permute_num = int(edge_num / 10) edge_index = data.edge_index.transpose(0, 1).numpy() idx_add = np.random.choice(node_num, (permute_num, 2)) # idx_add = [[idx_add[0, n], idx_add[1, n]] for n in range(permute_num) if not (idx_add[0, n], idx_add[1, n]) in edge_index] # edge_index = np.concatenate((np.array([edge_index[n] for n in range(edge_num) if not n in np.random.choice(edge_num, permute_num, replace=False)]), idx_add), axis=0) # edge_index = np.concatenate((edge_index[np.random.choice(edge_num, edge_num-permute_num, replace=False)], idx_add), axis=0) edge_index = edge_index[np.random.choice(edge_num, edge_num-permute_num, replace=False)] # edge_index = [edge_index[n] for n in range(edge_num) if not n in np.random.choice(edge_num, permute_num, replace=False)] + idx_add data.edge_index = torch.tensor(edge_index).transpose_(0, 1) return data def subgraph(data): node_num, _ = data.x.size() _, edge_num = data.edge_index.size() sub_num = int(node_num * 0.2) edge_index = data.edge_index.numpy() idx_sub = [np.random.randint(node_num, size=1)[0]] idx_neigh = set([n for n in edge_index[1][edge_index[0]==idx_sub[0]]]) count = 0 while len(idx_sub) <= sub_num: count = count + 1 if count > node_num: break if len(idx_neigh) == 0: break sample_node = np.random.choice(list(idx_neigh)) if sample_node in idx_sub: continue idx_sub.append(sample_node) idx_neigh.union(set([n for n in edge_index[1][edge_index[0]==idx_sub[-1]]])) idx_drop = [n for n in range(node_num) if not n in idx_sub] idx_nondrop = idx_sub idx_dict = {idx_nondrop[n]:n for n in list(range(len(idx_nondrop)))} # data.x = data.x[idx_nondrop] edge_index = data.edge_index.numpy() adj = torch.zeros((node_num, node_num)) adj[edge_index[0], edge_index[1]] = 1 adj[idx_drop, :] = 0 adj[:, idx_drop] = 0 edge_index = adj.nonzero().t() data.edge_index = edge_index # edge_index = [[idx_dict[edge_index[0, n]], idx_dict[edge_index[1, n]]] for n in range(edge_num) if (not edge_index[0, n] in idx_drop) and (not edge_index[1, n] in idx_drop)] # edge_index = [[edge_index[0, n], edge_index[1, n]] for n in range(edge_num) if (not edge_index[0, n] in idx_drop) and (not edge_index[1, n] in idx_drop)] + [[n, n] for n in idx_nondrop] # data.edge_index = torch.tensor(edge_index).transpose_(0, 1) return data def mask_nodes(data): node_num, feat_dim = data.x.size() mask_num = int(node_num / 10) idx_mask = np.random.choice(node_num, mask_num, replace=False) data.x[idx_mask] = torch.tensor(np.random.normal(loc=0.5, scale=0.5, size=(mask_num, feat_dim)), dtype=torch.float32) return data