linlin
GitHub
5 years ago
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
10 changed files with 655 additions and 400 deletions
Unified View
Diff Options
-
+4 -2gklearn/dataset/data_fetcher.py
-
+80 -87gklearn/dataset/dataset.py
-
+20 -5gklearn/dataset/file_managers.py
-
+6 -6gklearn/dataset/metadata.py
-
+54 -53gklearn/kernels/graph_kernel.py
-
+1 -1gklearn/kernels/metadata.py
-
+121 -43gklearn/kernels/shortest_path.py
-
+182 -54gklearn/kernels/structural_sp.py
-
+97 -97gklearn/kernels/structuralspKernel.py
-
+90 -52gklearn/tests/test_graph_kernels.py
+ 4
- 2
gklearn/dataset/data_fetcher.py
View File
| @@ -74,6 +74,8 @@ class DataFetcher(): | |||||
| message = 'Invalid Dataset name "' + self._name + '".' | message = 'Invalid Dataset name "' + self._name + '".' | ||||
| message += '\nAvailable datasets are as follows: \n\n' | message += '\nAvailable datasets are as follows: \n\n' | ||||
| message += '\n'.join(ds for ds in sorted(DATASET_META)) | message += '\n'.join(ds for ds in sorted(DATASET_META)) | ||||
| message += '\n\nFollowing special suffices can be added to the name:' | |||||
| message += '\n\n' + '\n'.join(['_unlabeled']) | |||||
| raise ValueError(message) | raise ValueError(message) | ||||
| else: | else: | ||||
| self.write_archive_file(self._name) | self.write_archive_file(self._name) | ||||
| @@ -127,9 +129,9 @@ class DataFetcher(): | |||||
| def write_archive_file(self, ds_name): | def write_archive_file(self, ds_name): | ||||
| path = osp.join(self._root, ds_name) | path = osp.join(self._root, ds_name) | ||||
| url = DATASET_META[ds_name]['url'] | |||||
| # filename_dir = osp.join(path,filename) | # filename_dir = osp.join(path,filename) | ||||
| if not osp.exists(path) or self._reload: | if not osp.exists(path) or self._reload: | ||||
| url = DATASET_META[ds_name]['url'] | |||||
| response = self.download_file(url) | response = self.download_file(url) | ||||
| if response is None: | if response is None: | ||||
| return False | return False | ||||
| @@ -152,7 +154,7 @@ class DataFetcher(): | |||||
| with tarfile.open(filename_archive, 'r:gz') as tar: | with tarfile.open(filename_archive, 'r:gz') as tar: | ||||
| if self._reload and self._verbose: | if self._reload and self._verbose: | ||||
| print(filename + ' Downloaded.') | print(filename + ' Downloaded.') | ||||
| subpath = os.path.join(path, tar.getnames()[0]) | |||||
| subpath = os.path.join(path, tar.getnames()[0].split('/')[0]) | |||||
| if not osp.exists(subpath) or self._reload: | if not osp.exists(subpath) or self._reload: | ||||
| tar.extractall(path = path) | tar.extractall(path = path) | ||||
| return subpath | return subpath | ||||
+ 80
- 87
gklearn/dataset/dataset.py
View File
| @@ -7,24 +7,14 @@ Created on Thu Mar 26 18:48:27 2020 | |||||
| """ | """ | ||||
| import numpy as np | import numpy as np | ||||
| import networkx as nx | import networkx as nx | ||||
| from gklearn.utils.graph_files import load_dataset | |||||
| import os | import os | ||||
| from gklearn.dataset import DATASET_META, DataFetcher, DataLoader | |||||
| class Dataset(object): | class Dataset(object): | ||||
| def __init__(self, filename=None, filename_targets=None, **kwargs): | |||||
| if filename is None: | |||||
| self._graphs = None | |||||
| self._targets = None | |||||
| self._node_labels = None | |||||
| self._edge_labels = None | |||||
| self._node_attrs = None | |||||
| self._edge_attrs = None | |||||
| else: | |||||
| self.load_dataset(filename, filename_targets=filename_targets, **kwargs) | |||||
| def __init__(self, inputs=None, root='datasets', filename_targets=None, targets=None, mode='networkx', clean_labels=True, reload=False, verbose=False, **kwargs): | |||||
| self._substructures = None | self._substructures = None | ||||
| self._node_label_dim = None | self._node_label_dim = None | ||||
| self._edge_label_dim = None | self._edge_label_dim = None | ||||
| @@ -49,15 +39,61 @@ class Dataset(object): | |||||
| self._node_attr_dim = None | self._node_attr_dim = None | ||||
| self._edge_attr_dim = None | self._edge_attr_dim = None | ||||
| self._class_number = None | self._class_number = None | ||||
| self._ds_name = None | |||||
| if inputs is None: | |||||
| self._graphs = None | |||||
| self._targets = None | |||||
| self._node_labels = None | |||||
| self._edge_labels = None | |||||
| self._node_attrs = None | |||||
| self._edge_attrs = None | |||||
| # If inputs is a list of graphs. | |||||
| elif isinstance(inputs, list): | |||||
| node_labels = kwargs.get('node_labels', None) | |||||
| node_attrs = kwargs.get('node_attrs', None) | |||||
| edge_labels = kwargs.get('edge_labels', None) | |||||
| edge_attrs = kwargs.get('edge_attrs', None) | |||||
| self.load_graphs(inputs, targets=targets) | |||||
| self.set_labels(node_labels=node_labels, node_attrs=node_attrs, edge_labels=edge_labels, edge_attrs=edge_attrs) | |||||
| if clean_labels: | |||||
| self.clean_labels() | |||||
| elif isinstance(inputs, str): | |||||
| # If inputs is predefined dataset name. | |||||
| if inputs in DATASET_META: | |||||
| self.load_predefined_dataset(inputs, root=root, clean_labels=clean_labels, reload=reload, verbose=verbose) | |||||
| self._ds_name = inputs | |||||
| elif inputs.endswith('_unlabeled'): | |||||
| self.load_predefined_dataset(inputs[:len(inputs) - 10], root=root, clean_labels=clean_labels, reload=reload, verbose=verbose) | |||||
| self._ds_name = inputs | |||||
| # Deal with special suffices. | |||||
| self.check_special_suffices() | |||||
| # If inputs is a file name. | |||||
| elif os.path.isfile(inputs): | |||||
| self.load_dataset(inputs, filename_targets=filename_targets, clean_labels=clean_labels, **kwargs) | |||||
| # If inputs is a file name. | |||||
| else: | |||||
| raise ValueError('The "inputs" argument "' + inputs + '" is not a valid dataset name or file name.') | |||||
| else: | |||||
| raise TypeError('The "inputs" argument cannot be recognized. "Inputs" can be a list of graphs, a predefined dataset name, or a file name of a dataset.') | |||||
| def load_dataset(self, filename, filename_targets=None, **kwargs): | |||||
| self._graphs, self._targets, label_names = load_dataset(filename, filename_targets=filename_targets, **kwargs) | |||||
| def load_dataset(self, filename, filename_targets=None, clean_labels=True, **kwargs): | |||||
| self._graphs, self._targets, label_names = DataLoader(filename, filename_targets=filename_targets, **kwargs).data | |||||
| self._node_labels = label_names['node_labels'] | self._node_labels = label_names['node_labels'] | ||||
| self._node_attrs = label_names['node_attrs'] | self._node_attrs = label_names['node_attrs'] | ||||
| self._edge_labels = label_names['edge_labels'] | self._edge_labels = label_names['edge_labels'] | ||||
| self._edge_attrs = label_names['edge_attrs'] | self._edge_attrs = label_names['edge_attrs'] | ||||
| self.clean_labels() | |||||
| if clean_labels: | |||||
| self.clean_labels() | |||||
| def load_graphs(self, graphs, targets=None): | def load_graphs(self, graphs, targets=None): | ||||
| @@ -67,84 +103,33 @@ class Dataset(object): | |||||
| # self.set_labels_attrs() # @todo | # self.set_labels_attrs() # @todo | ||||
| def load_predefined_dataset(self, ds_name): | |||||
| current_path = os.path.dirname(os.path.realpath(__file__)) + '/' | |||||
| if ds_name == 'Acyclic': | |||||
| ds_file = current_path + '../../datasets/Acyclic/dataset_bps.ds' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'AIDS': | |||||
| ds_file = current_path + '../../datasets/AIDS/AIDS_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'Alkane': | |||||
| ds_file = current_path + '../../datasets/Alkane/dataset.ds' | |||||
| fn_targets = current_path + '../../datasets/Alkane/dataset_boiling_point_names.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file, filename_targets=fn_targets) | |||||
| elif ds_name == 'COIL-DEL': | |||||
| ds_file = current_path + '../../datasets/COIL-DEL/COIL-DEL_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'COIL-RAG': | |||||
| ds_file = current_path + '../../datasets/COIL-RAG/COIL-RAG_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'COLORS-3': | |||||
| ds_file = current_path + '../../datasets/COLORS-3/COLORS-3_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'Cuneiform': | |||||
| ds_file = current_path + '../../datasets/Cuneiform/Cuneiform_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'DD': | |||||
| ds_file = current_path + '../../datasets/DD/DD_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'ENZYMES': | |||||
| ds_file = current_path + '../../datasets/ENZYMES_txt/ENZYMES_A_sparse.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'Fingerprint': | |||||
| ds_file = current_path + '../../datasets/Fingerprint/Fingerprint_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'FRANKENSTEIN': | |||||
| ds_file = current_path + '../../datasets/FRANKENSTEIN/FRANKENSTEIN_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'Letter-high': # node non-symb | |||||
| ds_file = current_path + '../../datasets/Letter-high/Letter-high_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'Letter-low': # node non-symb | |||||
| ds_file = current_path + '../../datasets/Letter-low/Letter-low_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'Letter-med': # node non-symb | |||||
| ds_file = current_path + '../../datasets/Letter-med/Letter-med_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'MAO': | |||||
| ds_file = current_path + '../../datasets/MAO/dataset.ds' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'Monoterpenoides': | |||||
| ds_file = current_path + '../../datasets/Monoterpenoides/dataset_10+.ds' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'MUTAG': | |||||
| ds_file = current_path + '../../datasets/MUTAG/MUTAG_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'NCI1': | |||||
| ds_file = current_path + '../../datasets/NCI1/NCI1_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'NCI109': | |||||
| ds_file = current_path + '../../datasets/NCI109/NCI109_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'PAH': | |||||
| ds_file = current_path + '../../datasets/PAH/dataset.ds' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'SYNTHETIC': | |||||
| pass | |||||
| elif ds_name == 'SYNTHETICnew': | |||||
| ds_file = current_path + '../../datasets/SYNTHETICnew/SYNTHETICnew_A.txt' | |||||
| self._graphs, self._targets, label_names = load_dataset(ds_file) | |||||
| elif ds_name == 'Synthie': | |||||
| pass | |||||
| def load_predefined_dataset(self, ds_name, root='datasets', clean_labels=True, reload=False, verbose=False): | |||||
| path = DataFetcher(name=ds_name, root=root, reload=reload, verbose=verbose).path | |||||
| if DATASET_META[ds_name]['database'] == 'tudataset': | |||||
| ds_file = os.path.join(path, ds_name + '_A.txt') | |||||
| fn_targets = None | |||||
| else: | else: | ||||
| raise Exception('The dataset name "', ds_name, '" is not pre-defined.') | |||||
| load_files = DATASET_META[ds_name]['load_files'] | |||||
| if isinstance(load_files[0], str): | |||||
| ds_file = os.path.join(path, load_files[0]) | |||||
| else: # load_files[0] is a list of files. | |||||
| ds_file = [os.path.join(path, fn) for fn in load_files[0]] | |||||
| fn_targets = os.path.join(path, load_files[1]) if len(load_files) == 2 else None | |||||
| self._graphs, self._targets, label_names = DataLoader(ds_file, filename_targets=fn_targets).data | |||||
| self._node_labels = label_names['node_labels'] | self._node_labels = label_names['node_labels'] | ||||
| self._node_attrs = label_names['node_attrs'] | self._node_attrs = label_names['node_attrs'] | ||||
| self._edge_labels = label_names['edge_labels'] | self._edge_labels = label_names['edge_labels'] | ||||
| self._edge_attrs = label_names['edge_attrs'] | self._edge_attrs = label_names['edge_attrs'] | ||||
| self.clean_labels() | |||||
| if clean_labels: | |||||
| self.clean_labels() | |||||
| # Deal with specific datasets. | |||||
| if ds_name == 'Alkane': | |||||
| self.trim_dataset(edge_required=True) | |||||
| self.remove_labels(node_labels=['atom_symbol']) | |||||
| def set_labels(self, node_labels=[], node_attrs=[], edge_labels=[], edge_attrs=[]): | def set_labels(self, node_labels=[], node_attrs=[], edge_labels=[], edge_attrs=[]): | ||||
| @@ -573,6 +558,14 @@ class Dataset(object): | |||||
| return dataset | return dataset | ||||
| def check_special_suffices(self): | |||||
| if self._ds_name.endswith('_unlabeled'): | |||||
| self.remove_labels(node_labels=self._node_labels, | |||||
| edge_labels=self._edge_labels, | |||||
| node_attrs=self._node_attrs, | |||||
| edge_attrs=self._edge_attrs) | |||||
| def get_all_node_labels(self): | def get_all_node_labels(self): | ||||
| node_labels = [] | node_labels = [] | ||||
| for g in self._graphs: | for g in self._graphs: | ||||
+ 20
- 5
gklearn/dataset/file_managers.py
View File
| @@ -38,7 +38,11 @@ class DataLoader(): | |||||
| for details. Note here filename is the name of either .txt file in | for details. Note here filename is the name of either .txt file in | ||||
| the dataset directory. | the dataset directory. | ||||
| """ | """ | ||||
| extension = splitext(filename)[1][1:] | |||||
| if isinstance(filename, str): | |||||
| extension = splitext(filename)[1][1:] | |||||
| else: # filename is a list of files. | |||||
| extension = splitext(filename[0])[1][1:] | |||||
| if extension == "ds": | if extension == "ds": | ||||
| self._graphs, self._targets, self._label_names = self.load_from_ds(filename, filename_targets) | self._graphs, self._targets, self._label_names = self.load_from_ds(filename, filename_targets) | ||||
| elif extension == "cxl": | elif extension == "cxl": | ||||
| @@ -67,13 +71,24 @@ class DataLoader(): | |||||
| Note these graph formats are checked automatically by the extensions of | Note these graph formats are checked automatically by the extensions of | ||||
| graph files. | graph files. | ||||
| """ | |||||
| dirname_dataset = dirname(filename) | |||||
| """ | |||||
| if isinstance(filename, str): | |||||
| dirname_dataset = dirname(filename) | |||||
| with open(filename) as f: | |||||
| content = f.read().splitlines() | |||||
| else: # filename is a list of files. | |||||
| dirname_dataset = dirname(filename[0]) | |||||
| content = [] | |||||
| for fn in filename: | |||||
| with open(fn) as f: | |||||
| content += f.read().splitlines() | |||||
| # to remove duplicate file names. | |||||
| data = [] | data = [] | ||||
| y = [] | y = [] | ||||
| label_names = {'node_labels': [], 'edge_labels': [], 'node_attrs': [], 'edge_attrs': []} | label_names = {'node_labels': [], 'edge_labels': [], 'node_attrs': [], 'edge_attrs': []} | ||||
| with open(filename) as fn: | |||||
| content = fn.read().splitlines() | |||||
| content = [line for line in content if not line.endswith('.ds')] # Alkane | |||||
| content = [line for line in content if not line.startswith('#')] # Acyclic | |||||
| extension = splitext(content[0].split(' ')[0])[1][1:] | extension = splitext(content[0].split(' ')[0])[1][1:] | ||||
| if extension == 'ct': | if extension == 'ct': | ||||
| load_file_fun = self.load_ct | load_file_fun = self.load_ct | ||||
+ 6
- 6
gklearn/dataset/metadata.py
View File
| @@ -32,7 +32,7 @@ GREYC_META = { | |||||
| 'domain': 'small molecules', | 'domain': 'small molecules', | ||||
| 'train_valid_test': [], | 'train_valid_test': [], | ||||
| 'stereoisomerism': True, | 'stereoisomerism': True, | ||||
| 'load_files': [], | |||||
| 'load_files': ['data.ds'], | |||||
| }, | }, | ||||
| 'Acyclic': { | 'Acyclic': { | ||||
| 'database': 'greyc', | 'database': 'greyc', | ||||
| @@ -165,7 +165,7 @@ GREYC_META = { | |||||
| 'domain': 'small molecules', | 'domain': 'small molecules', | ||||
| 'train_valid_test': ['trainset_0.ds', None, 'testset_0.ds'], | 'train_valid_test': ['trainset_0.ds', None, 'testset_0.ds'], | ||||
| 'stereoisomerism': False, | 'stereoisomerism': False, | ||||
| 'load_files': [], | |||||
| 'load_files': [['trainset_0.ds', 'testset_0.ds']], | |||||
| }, | }, | ||||
| 'PTC': { | 'PTC': { | ||||
| 'database': 'greyc', | 'database': 'greyc', | ||||
| @@ -654,7 +654,7 @@ TUDataset_META = { | |||||
| 'node_attr_dim': 0, | 'node_attr_dim': 0, | ||||
| 'geometry': None, | 'geometry': None, | ||||
| 'edge_attr_dim': 0, | 'edge_attr_dim': 0, | ||||
| 'url': 'https://www.chrsmrrs.com/graphkerneldatasets/NCI-H23.zip-H23', | |||||
| 'url': 'https://www.chrsmrrs.com/graphkerneldatasets/NCI-H23.zip', | |||||
| 'domain': 'small molecules', | 'domain': 'small molecules', | ||||
| }, | }, | ||||
| 'NCI-H23H': { | 'NCI-H23H': { | ||||
| @@ -670,7 +670,7 @@ TUDataset_META = { | |||||
| 'node_attr_dim': 0, | 'node_attr_dim': 0, | ||||
| 'geometry': None, | 'geometry': None, | ||||
| 'edge_attr_dim': 0, | 'edge_attr_dim': 0, | ||||
| 'url': 'https://www.chrsmrrs.com/graphkerneldatasets/NCI-H23H.zip-H23H', | |||||
| 'url': 'https://www.chrsmrrs.com/graphkerneldatasets/NCI-H23H.zip', | |||||
| 'domain': 'small molecules', | 'domain': 'small molecules', | ||||
| }, | }, | ||||
| 'OVCAR-8': { | 'OVCAR-8': { | ||||
| @@ -686,7 +686,7 @@ TUDataset_META = { | |||||
| 'node_attr_dim': 0, | 'node_attr_dim': 0, | ||||
| 'geometry': None, | 'geometry': None, | ||||
| 'edge_attr_dim': 0, | 'edge_attr_dim': 0, | ||||
| 'url': 'https://www.chrsmrrs.com/graphkerneldatasets/OVCAR-8.zip-8', | |||||
| 'url': 'https://www.chrsmrrs.com/graphkerneldatasets/OVCAR-8.zip', | |||||
| 'domain': 'small molecules', | 'domain': 'small molecules', | ||||
| }, | }, | ||||
| 'OVCAR-8H': { | 'OVCAR-8H': { | ||||
| @@ -702,7 +702,7 @@ TUDataset_META = { | |||||
| 'node_attr_dim': 0, | 'node_attr_dim': 0, | ||||
| 'geometry': None, | 'geometry': None, | ||||
| 'edge_attr_dim': 0, | 'edge_attr_dim': 0, | ||||
| 'url': 'https://www.chrsmrrs.com/graphkerneldatasets/OVCAR-8H.zip-8H', | |||||
| 'url': 'https://www.chrsmrrs.com/graphkerneldatasets/OVCAR-8H.zip', | |||||
| 'domain': 'small molecules', | 'domain': 'small molecules', | ||||
| }, | }, | ||||
| 'P388': { | 'P388': { | ||||
+ 54
- 53
gklearn/kernels/graph_kernel.py
View File
| @@ -9,10 +9,11 @@ import numpy as np | |||||
| import networkx as nx | import networkx as nx | ||||
| import multiprocessing | import multiprocessing | ||||
| import time | import time | ||||
| from gklearn.utils import normalize_gram_matrix | |||||
| class GraphKernel(object): | class GraphKernel(object): | ||||
| def __init__(self): | def __init__(self): | ||||
| self._graphs = None | self._graphs = None | ||||
| self._parallel = '' | self._parallel = '' | ||||
| @@ -22,14 +23,14 @@ class GraphKernel(object): | |||||
| self._run_time = 0 | self._run_time = 0 | ||||
| self._gram_matrix = None | self._gram_matrix = None | ||||
| self._gram_matrix_unnorm = None | self._gram_matrix_unnorm = None | ||||
| def compute(self, *graphs, **kwargs): | def compute(self, *graphs, **kwargs): | ||||
| self._parallel = kwargs.get('parallel', 'imap_unordered') | self._parallel = kwargs.get('parallel', 'imap_unordered') | ||||
| self._n_jobs = kwargs.get('n_jobs', multiprocessing.cpu_count()) | self._n_jobs = kwargs.get('n_jobs', multiprocessing.cpu_count()) | ||||
| self._normalize = kwargs.get('normalize', True) | self._normalize = kwargs.get('normalize', True) | ||||
| self._verbose = kwargs.get('verbose', 2) | self._verbose = kwargs.get('verbose', 2) | ||||
| if len(graphs) == 1: | if len(graphs) == 1: | ||||
| if not isinstance(graphs[0], list): | if not isinstance(graphs[0], list): | ||||
| raise Exception('Cannot detect graphs.') | raise Exception('Cannot detect graphs.') | ||||
| @@ -40,9 +41,9 @@ class GraphKernel(object): | |||||
| self._gram_matrix = self._compute_gram_matrix() | self._gram_matrix = self._compute_gram_matrix() | ||||
| self._gram_matrix_unnorm = np.copy(self._gram_matrix) | self._gram_matrix_unnorm = np.copy(self._gram_matrix) | ||||
| if self._normalize: | if self._normalize: | ||||
| self._gram_matrix = self.normalize_gm(self._gram_matrix) | |||||
| self._gram_matrix = normalize_gram_matrix(self._gram_matrix) | |||||
| return self._gram_matrix, self._run_time | return self._gram_matrix, self._run_time | ||||
| elif len(graphs) == 2: | elif len(graphs) == 2: | ||||
| if self.is_graph(graphs[0]) and self.is_graph(graphs[1]): | if self.is_graph(graphs[0]) and self.is_graph(graphs[1]): | ||||
| kernel = self._compute_single_kernel(graphs[0].copy(), graphs[1].copy()) | kernel = self._compute_single_kernel(graphs[0].copy(), graphs[1].copy()) | ||||
| @@ -59,14 +60,14 @@ class GraphKernel(object): | |||||
| return kernel_list, self._run_time | return kernel_list, self._run_time | ||||
| else: | else: | ||||
| raise Exception('Cannot detect graphs.') | raise Exception('Cannot detect graphs.') | ||||
| elif len(graphs) == 0 and self._graphs is None: | elif len(graphs) == 0 and self._graphs is None: | ||||
| raise Exception('Please add graphs before computing.') | raise Exception('Please add graphs before computing.') | ||||
| else: | else: | ||||
| raise Exception('Cannot detect graphs.') | raise Exception('Cannot detect graphs.') | ||||
| def normalize_gm(self, gram_matrix): | def normalize_gm(self, gram_matrix): | ||||
| import warnings | import warnings | ||||
| warnings.warn('gklearn.kernels.graph_kernel.normalize_gm will be deprecated, use gklearn.utils.normalize_gram_matrix instead', DeprecationWarning) | warnings.warn('gklearn.kernels.graph_kernel.normalize_gm will be deprecated, use gklearn.utils.normalize_gram_matrix instead', DeprecationWarning) | ||||
| @@ -77,8 +78,8 @@ class GraphKernel(object): | |||||
| gram_matrix[i][j] /= np.sqrt(diag[i] * diag[j]) | gram_matrix[i][j] /= np.sqrt(diag[i] * diag[j]) | ||||
| gram_matrix[j][i] = gram_matrix[i][j] | gram_matrix[j][i] = gram_matrix[i][j] | ||||
| return gram_matrix | return gram_matrix | ||||
| def compute_distance_matrix(self): | def compute_distance_matrix(self): | ||||
| if self._gram_matrix is None: | if self._gram_matrix is None: | ||||
| raise Exception('Please compute the Gram matrix before computing distance matrix.') | raise Exception('Please compute the Gram matrix before computing distance matrix.') | ||||
| @@ -97,98 +98,98 @@ class GraphKernel(object): | |||||
| dis_min = np.min(np.min(dis_mat[dis_mat != 0])) | dis_min = np.min(np.min(dis_mat[dis_mat != 0])) | ||||
| dis_mean = np.mean(np.mean(dis_mat)) | dis_mean = np.mean(np.mean(dis_mat)) | ||||
| return dis_mat, dis_max, dis_min, dis_mean | return dis_mat, dis_max, dis_min, dis_mean | ||||
| def _compute_gram_matrix(self): | def _compute_gram_matrix(self): | ||||
| start_time = time.time() | start_time = time.time() | ||||
| if self._parallel == 'imap_unordered': | if self._parallel == 'imap_unordered': | ||||
| gram_matrix = self._compute_gm_imap_unordered() | gram_matrix = self._compute_gm_imap_unordered() | ||||
| elif self._parallel is None: | elif self._parallel is None: | ||||
| gram_matrix = self._compute_gm_series() | gram_matrix = self._compute_gm_series() | ||||
| else: | else: | ||||
| raise Exception('Parallel mode is not set correctly.') | raise Exception('Parallel mode is not set correctly.') | ||||
| self._run_time = time.time() - start_time | self._run_time = time.time() - start_time | ||||
| if self._verbose: | if self._verbose: | ||||
| print('Gram matrix of size %d built in %s seconds.' | print('Gram matrix of size %d built in %s seconds.' | ||||
| % (len(self._graphs), self._run_time)) | % (len(self._graphs), self._run_time)) | ||||
| return gram_matrix | return gram_matrix | ||||
| def _compute_gm_series(self): | def _compute_gm_series(self): | ||||
| pass | pass | ||||
| def _compute_gm_imap_unordered(self): | def _compute_gm_imap_unordered(self): | ||||
| pass | pass | ||||
| def _compute_kernel_list(self, g1, g_list): | def _compute_kernel_list(self, g1, g_list): | ||||
| start_time = time.time() | start_time = time.time() | ||||
| if self._parallel == 'imap_unordered': | if self._parallel == 'imap_unordered': | ||||
| kernel_list = self._compute_kernel_list_imap_unordered(g1, g_list) | kernel_list = self._compute_kernel_list_imap_unordered(g1, g_list) | ||||
| elif self._parallel is None: | elif self._parallel is None: | ||||
| kernel_list = self._compute_kernel_list_series(g1, g_list) | kernel_list = self._compute_kernel_list_series(g1, g_list) | ||||
| else: | else: | ||||
| raise Exception('Parallel mode is not set correctly.') | raise Exception('Parallel mode is not set correctly.') | ||||
| self._run_time = time.time() - start_time | self._run_time = time.time() - start_time | ||||
| if self._verbose: | if self._verbose: | ||||
| print('Graph kernel bewteen a graph and a list of %d graphs built in %s seconds.' | print('Graph kernel bewteen a graph and a list of %d graphs built in %s seconds.' | ||||
| % (len(g_list), self._run_time)) | % (len(g_list), self._run_time)) | ||||
| return kernel_list | return kernel_list | ||||
| def _compute_kernel_list_series(self, g1, g_list): | def _compute_kernel_list_series(self, g1, g_list): | ||||
| pass | pass | ||||
| def _compute_kernel_list_imap_unordered(self, g1, g_list): | def _compute_kernel_list_imap_unordered(self, g1, g_list): | ||||
| pass | pass | ||||
| def _compute_single_kernel(self, g1, g2): | def _compute_single_kernel(self, g1, g2): | ||||
| start_time = time.time() | start_time = time.time() | ||||
| kernel = self._compute_single_kernel_series(g1, g2) | kernel = self._compute_single_kernel_series(g1, g2) | ||||
| self._run_time = time.time() - start_time | self._run_time = time.time() - start_time | ||||
| if self._verbose: | if self._verbose: | ||||
| print('Graph kernel bewteen two graphs built in %s seconds.' % (self._run_time)) | print('Graph kernel bewteen two graphs built in %s seconds.' % (self._run_time)) | ||||
| return kernel | return kernel | ||||
| def _compute_single_kernel_series(self, g1, g2): | def _compute_single_kernel_series(self, g1, g2): | ||||
| pass | pass | ||||
| def is_graph(self, graph): | def is_graph(self, graph): | ||||
| if isinstance(graph, nx.Graph): | if isinstance(graph, nx.Graph): | ||||
| return True | return True | ||||
| if isinstance(graph, nx.DiGraph): | if isinstance(graph, nx.DiGraph): | ||||
| return True | |||||
| return True | |||||
| if isinstance(graph, nx.MultiGraph): | if isinstance(graph, nx.MultiGraph): | ||||
| return True | |||||
| return True | |||||
| if isinstance(graph, nx.MultiDiGraph): | if isinstance(graph, nx.MultiDiGraph): | ||||
| return True | |||||
| return True | |||||
| return False | return False | ||||
| @property | @property | ||||
| def graphs(self): | def graphs(self): | ||||
| return self._graphs | return self._graphs | ||||
| @property | @property | ||||
| def parallel(self): | def parallel(self): | ||||
| return self._parallel | return self._parallel | ||||
| @property | @property | ||||
| def n_jobs(self): | def n_jobs(self): | ||||
| return self._n_jobs | return self._n_jobs | ||||
| @@ -197,30 +198,30 @@ class GraphKernel(object): | |||||
| @property | @property | ||||
| def verbose(self): | def verbose(self): | ||||
| return self._verbose | return self._verbose | ||||
| @property | @property | ||||
| def normalize(self): | def normalize(self): | ||||
| return self._normalize | return self._normalize | ||||
| @property | @property | ||||
| def run_time(self): | def run_time(self): | ||||
| return self._run_time | return self._run_time | ||||
| @property | @property | ||||
| def gram_matrix(self): | def gram_matrix(self): | ||||
| return self._gram_matrix | return self._gram_matrix | ||||
| @gram_matrix.setter | @gram_matrix.setter | ||||
| def gram_matrix(self, value): | def gram_matrix(self, value): | ||||
| self._gram_matrix = value | self._gram_matrix = value | ||||
| @property | @property | ||||
| def gram_matrix_unnorm(self): | def gram_matrix_unnorm(self): | ||||
| return self._gram_matrix_unnorm | |||||
| return self._gram_matrix_unnorm | |||||
| @gram_matrix_unnorm.setter | @gram_matrix_unnorm.setter | ||||
| def gram_matrix_unnorm(self, value): | def gram_matrix_unnorm(self, value): | ||||
+ 1
- 1
gklearn/kernels/metadata.py
View File
| @@ -12,7 +12,7 @@ GRAPH_KERNELS = { | |||||
| 'common walk': '', | 'common walk': '', | ||||
| 'marginalized': '', | 'marginalized': '', | ||||
| 'sylvester equation': '', | 'sylvester equation': '', | ||||
| 'fixed_point': '', | |||||
| 'fixed point': '', | |||||
| 'conjugate gradient': '', | 'conjugate gradient': '', | ||||
| 'spectral decomposition': '', | 'spectral decomposition': '', | ||||
| ### based on paths. | ### based on paths. | ||||
+ 121
- 43
gklearn/kernels/shortest_path.py
View File
| @@ -5,9 +5,9 @@ Created on Tue Apr 7 15:24:58 2020 | |||||
| @author: ljia | @author: ljia | ||||
| @references: | |||||
| [1] Borgwardt KM, Kriegel HP. Shortest-path kernels on graphs. InData | |||||
| @references: | |||||
| [1] Borgwardt KM, Kriegel HP. Shortest-path kernels on graphs. InData | |||||
| Mining, Fifth IEEE International Conference on 2005 Nov 27 (pp. 8-pp). IEEE. | Mining, Fifth IEEE International Conference on 2005 Nov 27 (pp. 8-pp). IEEE. | ||||
| """ | """ | ||||
| @@ -17,33 +17,36 @@ from itertools import product | |||||
| from multiprocessing import Pool | from multiprocessing import Pool | ||||
| from tqdm import tqdm | from tqdm import tqdm | ||||
| import numpy as np | import numpy as np | ||||
| import networkx as nx | |||||
| from gklearn.utils.parallel import parallel_gm, parallel_me | from gklearn.utils.parallel import parallel_gm, parallel_me | ||||
| from gklearn.utils.utils import getSPGraph | from gklearn.utils.utils import getSPGraph | ||||
| from gklearn.kernels import GraphKernel | from gklearn.kernels import GraphKernel | ||||
| class ShortestPath(GraphKernel): | class ShortestPath(GraphKernel): | ||||
| def __init__(self, **kwargs): | def __init__(self, **kwargs): | ||||
| GraphKernel.__init__(self) | GraphKernel.__init__(self) | ||||
| self._node_labels = kwargs.get('node_labels', []) | self._node_labels = kwargs.get('node_labels', []) | ||||
| self._node_attrs = kwargs.get('node_attrs', []) | self._node_attrs = kwargs.get('node_attrs', []) | ||||
| self._edge_weight = kwargs.get('edge_weight', None) | self._edge_weight = kwargs.get('edge_weight', None) | ||||
| self._node_kernels = kwargs.get('node_kernels', None) | self._node_kernels = kwargs.get('node_kernels', None) | ||||
| self._fcsp = kwargs.get('fcsp', True) | |||||
| self._ds_infos = kwargs.get('ds_infos', {}) | self._ds_infos = kwargs.get('ds_infos', {}) | ||||
| def _compute_gm_series(self): | def _compute_gm_series(self): | ||||
| self._all_graphs_have_edges(self._graphs) | |||||
| # get shortest path graph of each graph. | # get shortest path graph of each graph. | ||||
| if self._verbose >= 2: | if self._verbose >= 2: | ||||
| iterator = tqdm(self._graphs, desc='getting sp graphs', file=sys.stdout) | iterator = tqdm(self._graphs, desc='getting sp graphs', file=sys.stdout) | ||||
| else: | else: | ||||
| iterator = self._graphs | iterator = self._graphs | ||||
| self._graphs = [getSPGraph(g, edge_weight=self._edge_weight) for g in iterator] | self._graphs = [getSPGraph(g, edge_weight=self._edge_weight) for g in iterator] | ||||
| # compute Gram matrix. | # compute Gram matrix. | ||||
| gram_matrix = np.zeros((len(self._graphs), len(self._graphs))) | gram_matrix = np.zeros((len(self._graphs), len(self._graphs))) | ||||
| from itertools import combinations_with_replacement | from itertools import combinations_with_replacement | ||||
| itr = combinations_with_replacement(range(0, len(self._graphs)), 2) | itr = combinations_with_replacement(range(0, len(self._graphs)), 2) | ||||
| if self._verbose >= 2: | if self._verbose >= 2: | ||||
| @@ -54,11 +57,12 @@ class ShortestPath(GraphKernel): | |||||
| kernel = self._sp_do(self._graphs[i], self._graphs[j]) | kernel = self._sp_do(self._graphs[i], self._graphs[j]) | ||||
| gram_matrix[i][j] = kernel | gram_matrix[i][j] = kernel | ||||
| gram_matrix[j][i] = kernel | gram_matrix[j][i] = kernel | ||||
| return gram_matrix | return gram_matrix | ||||
| def _compute_gm_imap_unordered(self): | def _compute_gm_imap_unordered(self): | ||||
| self._all_graphs_have_edges(self._graphs) | |||||
| # get shortest path graph of each graph. | # get shortest path graph of each graph. | ||||
| pool = Pool(self._n_jobs) | pool = Pool(self._n_jobs) | ||||
| get_sp_graphs_fun = self._wrapper_get_sp_graphs | get_sp_graphs_fun = self._wrapper_get_sp_graphs | ||||
| @@ -76,21 +80,22 @@ class ShortestPath(GraphKernel): | |||||
| self._graphs[i] = g | self._graphs[i] = g | ||||
| pool.close() | pool.close() | ||||
| pool.join() | pool.join() | ||||
| # compute Gram matrix. | # compute Gram matrix. | ||||
| gram_matrix = np.zeros((len(self._graphs), len(self._graphs))) | gram_matrix = np.zeros((len(self._graphs), len(self._graphs))) | ||||
| def init_worker(gs_toshare): | def init_worker(gs_toshare): | ||||
| global G_gs | global G_gs | ||||
| G_gs = gs_toshare | G_gs = gs_toshare | ||||
| do_fun = self._wrapper_sp_do | do_fun = self._wrapper_sp_do | ||||
| parallel_gm(do_fun, gram_matrix, self._graphs, init_worker=init_worker, | |||||
| parallel_gm(do_fun, gram_matrix, self._graphs, init_worker=init_worker, | |||||
| glbv=(self._graphs,), n_jobs=self._n_jobs, verbose=self._verbose) | glbv=(self._graphs,), n_jobs=self._n_jobs, verbose=self._verbose) | ||||
| return gram_matrix | return gram_matrix | ||||
| def _compute_kernel_list_series(self, g1, g_list): | def _compute_kernel_list_series(self, g1, g_list): | ||||
| self._all_graphs_have_edges([g1] + g_list) | |||||
| # get shortest path graphs of g1 and each graph in g_list. | # get shortest path graphs of g1 and each graph in g_list. | ||||
| g1 = getSPGraph(g1, edge_weight=self._edge_weight) | g1 = getSPGraph(g1, edge_weight=self._edge_weight) | ||||
| if self._verbose >= 2: | if self._verbose >= 2: | ||||
| @@ -98,7 +103,7 @@ class ShortestPath(GraphKernel): | |||||
| else: | else: | ||||
| iterator = g_list | iterator = g_list | ||||
| g_list = [getSPGraph(g, edge_weight=self._edge_weight) for g in iterator] | g_list = [getSPGraph(g, edge_weight=self._edge_weight) for g in iterator] | ||||
| # compute kernel list. | # compute kernel list. | ||||
| kernel_list = [None] * len(g_list) | kernel_list = [None] * len(g_list) | ||||
| if self._verbose >= 2: | if self._verbose >= 2: | ||||
| @@ -108,11 +113,12 @@ class ShortestPath(GraphKernel): | |||||
| for i in iterator: | for i in iterator: | ||||
| kernel = self._sp_do(g1, g_list[i]) | kernel = self._sp_do(g1, g_list[i]) | ||||
| kernel_list[i] = kernel | kernel_list[i] = kernel | ||||
| return kernel_list | return kernel_list | ||||
| def _compute_kernel_list_imap_unordered(self, g1, g_list): | def _compute_kernel_list_imap_unordered(self, g1, g_list): | ||||
| self._all_graphs_have_edges([g1] + g_list) | |||||
| # get shortest path graphs of g1 and each graph in g_list. | # get shortest path graphs of g1 and each graph in g_list. | ||||
| g1 = getSPGraph(g1, edge_weight=self._edge_weight) | g1 = getSPGraph(g1, edge_weight=self._edge_weight) | ||||
| pool = Pool(self._n_jobs) | pool = Pool(self._n_jobs) | ||||
| @@ -131,49 +137,58 @@ class ShortestPath(GraphKernel): | |||||
| g_list[i] = g | g_list[i] = g | ||||
| pool.close() | pool.close() | ||||
| pool.join() | pool.join() | ||||
| # compute Gram matrix. | # compute Gram matrix. | ||||
| kernel_list = [None] * len(g_list) | kernel_list = [None] * len(g_list) | ||||
| def init_worker(g1_toshare, gl_toshare): | def init_worker(g1_toshare, gl_toshare): | ||||
| global G_g1, G_gl | global G_g1, G_gl | ||||
| G_g1 = g1_toshare | |||||
| G_gl = gl_toshare | |||||
| G_g1 = g1_toshare | |||||
| G_gl = gl_toshare | |||||
| do_fun = self._wrapper_kernel_list_do | do_fun = self._wrapper_kernel_list_do | ||||
| def func_assign(result, var_to_assign): | |||||
| def func_assign(result, var_to_assign): | |||||
| var_to_assign[result[0]] = result[1] | var_to_assign[result[0]] = result[1] | ||||
| itr = range(len(g_list)) | itr = range(len(g_list)) | ||||
| len_itr = len(g_list) | len_itr = len(g_list) | ||||
| parallel_me(do_fun, func_assign, kernel_list, itr, len_itr=len_itr, | parallel_me(do_fun, func_assign, kernel_list, itr, len_itr=len_itr, | ||||
| init_worker=init_worker, glbv=(g1, g_list), method='imap_unordered', n_jobs=self._n_jobs, itr_desc='Computing kernels', verbose=self._verbose) | init_worker=init_worker, glbv=(g1, g_list), method='imap_unordered', n_jobs=self._n_jobs, itr_desc='Computing kernels', verbose=self._verbose) | ||||
| return kernel_list | return kernel_list | ||||
| def _wrapper_kernel_list_do(self, itr): | def _wrapper_kernel_list_do(self, itr): | ||||
| return itr, self._sp_do(G_g1, G_gl[itr]) | return itr, self._sp_do(G_g1, G_gl[itr]) | ||||
| def _compute_single_kernel_series(self, g1, g2): | def _compute_single_kernel_series(self, g1, g2): | ||||
| self._all_graphs_have_edges([g1] + [g2]) | |||||
| g1 = getSPGraph(g1, edge_weight=self._edge_weight) | g1 = getSPGraph(g1, edge_weight=self._edge_weight) | ||||
| g2 = getSPGraph(g2, edge_weight=self._edge_weight) | g2 = getSPGraph(g2, edge_weight=self._edge_weight) | ||||
| kernel = self._sp_do(g1, g2) | kernel = self._sp_do(g1, g2) | ||||
| return kernel | |||||
| return kernel | |||||
| def _wrapper_get_sp_graphs(self, itr_item): | def _wrapper_get_sp_graphs(self, itr_item): | ||||
| g = itr_item[0] | g = itr_item[0] | ||||
| i = itr_item[1] | i = itr_item[1] | ||||
| return i, getSPGraph(g, edge_weight=self._edge_weight) | return i, getSPGraph(g, edge_weight=self._edge_weight) | ||||
| def _sp_do(self, g1, g2): | def _sp_do(self, g1, g2): | ||||
| if self._fcsp: # @todo: it may be put outside the _sp_do(). | |||||
| return self._sp_do_fcsp(g1, g2) | |||||
| else: | |||||
| return self._sp_do_naive(g1, g2) | |||||
| def _sp_do_fcsp(self, g1, g2): | |||||
| kernel = 0 | kernel = 0 | ||||
| # compute shortest path matrices first, method borrowed from FCSP. | # compute shortest path matrices first, method borrowed from FCSP. | ||||
| vk_dict = {} # shortest path matrices dict | vk_dict = {} # shortest path matrices dict | ||||
| if len(self._node_labels) > 0: | |||||
| if len(self._node_labels) > 0: # @todo: it may be put outside the _sp_do(). | |||||
| # node symb and non-synb labeled | # node symb and non-synb labeled | ||||
| if len(self._node_attrs) > 0: | if len(self._node_attrs) > 0: | ||||
| kn = self._node_kernels['mix'] | kn = self._node_kernels['mix'] | ||||
| @@ -208,7 +223,7 @@ class ShortestPath(GraphKernel): | |||||
| if e1[2]['cost'] == e2[2]['cost']: | if e1[2]['cost'] == e2[2]['cost']: | ||||
| kernel += 1 | kernel += 1 | ||||
| return kernel | return kernel | ||||
| # compute graph kernels | # compute graph kernels | ||||
| if self._ds_infos['directed']: | if self._ds_infos['directed']: | ||||
| for e1, e2 in product(g1.edges(data=True), g2.edges(data=True)): | for e1, e2 in product(g1.edges(data=True), g2.edges(data=True)): | ||||
| @@ -225,7 +240,7 @@ class ShortestPath(GraphKernel): | |||||
| kn1 = nk11 * nk22 | kn1 = nk11 * nk22 | ||||
| kn2 = nk12 * nk21 | kn2 = nk12 * nk21 | ||||
| kernel += kn1 + kn2 | kernel += kn1 + kn2 | ||||
| # # ---- exact implementation of the Fast Computation of Shortest Path Kernel (FCSP), reference [2], sadly it is slower than the current implementation | # # ---- exact implementation of the Fast Computation of Shortest Path Kernel (FCSP), reference [2], sadly it is slower than the current implementation | ||||
| # # compute vertex kernels | # # compute vertex kernels | ||||
| # try: | # try: | ||||
| @@ -238,7 +253,7 @@ class ShortestPath(GraphKernel): | |||||
| # vk_mat[i1][i2] = kn( | # vk_mat[i1][i2] = kn( | ||||
| # n1[1][node_label], n2[1][node_label], | # n1[1][node_label], n2[1][node_label], | ||||
| # [n1[1]['attributes']], [n2[1]['attributes']]) | # [n1[1]['attributes']], [n2[1]['attributes']]) | ||||
| # range1 = range(0, len(edge_w_g[i])) | # range1 = range(0, len(edge_w_g[i])) | ||||
| # range2 = range(0, len(edge_w_g[j])) | # range2 = range(0, len(edge_w_g[j])) | ||||
| # for i1 in range1: | # for i1 in range1: | ||||
| @@ -254,11 +269,74 @@ class ShortestPath(GraphKernel): | |||||
| # kn1 = vk_mat[x1][x2] * vk_mat[y1][y2] | # kn1 = vk_mat[x1][x2] * vk_mat[y1][y2] | ||||
| # kn2 = vk_mat[x1][y2] * vk_mat[y1][x2] | # kn2 = vk_mat[x1][y2] * vk_mat[y1][x2] | ||||
| # kernel += kn1 + kn2 | # kernel += kn1 + kn2 | ||||
| return kernel | return kernel | ||||
| def _sp_do_naive(self, g1, g2): | |||||
| kernel = 0 | |||||
| # Define the function to compute kernels between vertices in each condition. | |||||
| if len(self._node_labels) > 0: | |||||
| # node symb and non-synb labeled | |||||
| if len(self._node_attrs) > 0: | |||||
| def compute_vk(n1, n2): | |||||
| kn = self._node_kernels['mix'] | |||||
| n1_labels = [g1.nodes[n1][nl] for nl in self._node_labels] | |||||
| n2_labels = [g2.nodes[n2][nl] for nl in self._node_labels] | |||||
| n1_attrs = [g1.nodes[n1][na] for na in self._node_attrs] | |||||
| n2_attrs = [g2.nodes[n2][na] for na in self._node_attrs] | |||||
| return kn(n1_labels, n2_labels, n1_attrs, n2_attrs) | |||||
| # node symb labeled | |||||
| else: | |||||
| def compute_vk(n1, n2): | |||||
| kn = self._node_kernels['symb'] | |||||
| n1_labels = [g1.nodes[n1][nl] for nl in self._node_labels] | |||||
| n2_labels = [g2.nodes[n2][nl] for nl in self._node_labels] | |||||
| return kn(n1_labels, n2_labels) | |||||
| else: | |||||
| # node non-synb labeled | |||||
| if len(self._node_attrs) > 0: | |||||
| def compute_vk(n1, n2): | |||||
| kn = self._node_kernels['nsymb'] | |||||
| n1_attrs = [g1.nodes[n1][na] for na in self._node_attrs] | |||||
| n2_attrs = [g2.nodes[n2][na] for na in self._node_attrs] | |||||
| return kn(n1_attrs, n2_attrs) | |||||
| # node unlabeled | |||||
| else: | |||||
| for e1, e2 in product(g1.edges(data=True), g2.edges(data=True)): | |||||
| if e1[2]['cost'] == e2[2]['cost']: | |||||
| kernel += 1 | |||||
| return kernel | |||||
| # compute graph kernels | |||||
| if self._ds_infos['directed']: | |||||
| for e1, e2 in product(g1.edges(data=True), g2.edges(data=True)): | |||||
| if e1[2]['cost'] == e2[2]['cost']: | |||||
| nk11, nk22 = compute_vk(e1[0], e2[0]), compute_vk(e1[1], e2[1]) | |||||
| kn1 = nk11 * nk22 | |||||
| kernel += kn1 | |||||
| else: | |||||
| for e1, e2 in product(g1.edges(data=True), g2.edges(data=True)): | |||||
| if e1[2]['cost'] == e2[2]['cost']: | |||||
| # each edge walk is counted twice, starting from both its extreme nodes. | |||||
| nk11, nk12, nk21, nk22 = compute_vk(e1[0], e2[0]), compute_vk( | |||||
| e1[0], e2[1]), compute_vk(e1[1], e2[0]), compute_vk(e1[1], e2[1]) | |||||
| kn1 = nk11 * nk22 | |||||
| kn2 = nk12 * nk21 | |||||
| kernel += kn1 + kn2 | |||||
| return kernel | |||||
| def _wrapper_sp_do(self, itr): | def _wrapper_sp_do(self, itr): | ||||
| i = itr[0] | i = itr[0] | ||||
| j = itr[1] | j = itr[1] | ||||
| return i, j, self._sp_do(G_gs[i], G_gs[j]) | |||||
| return i, j, self._sp_do(G_gs[i], G_gs[j]) | |||||
| def _all_graphs_have_edges(self, graphs): | |||||
| for G in graphs: | |||||
| if nx.number_of_edges(G) == 0: | |||||
| raise ValueError('Not all graphs have edges!!!') | |||||
+ 182
- 54
gklearn/kernels/structural_sp.py
View File
| @@ -5,9 +5,9 @@ Created on Mon Mar 30 11:59:57 2020 | |||||
| @author: ljia | @author: ljia | ||||
| @references: | |||||
| @references: | |||||
| [1] Suard F, Rakotomamonjy A, Bensrhair A. Kernel on Bag of Paths For | |||||
| [1] Suard F, Rakotomamonjy A, Bensrhair A. Kernel on Bag of Paths For | |||||
| Measuring Similarity of Shapes. InESANN 2007 Apr 25 (pp. 355-360). | Measuring Similarity of Shapes. InESANN 2007 Apr 25 (pp. 355-360). | ||||
| """ | """ | ||||
| import sys | import sys | ||||
| @@ -23,7 +23,7 @@ from gklearn.kernels import GraphKernel | |||||
| class StructuralSP(GraphKernel): | class StructuralSP(GraphKernel): | ||||
| def __init__(self, **kwargs): | def __init__(self, **kwargs): | ||||
| GraphKernel.__init__(self) | GraphKernel.__init__(self) | ||||
| self._node_labels = kwargs.get('node_labels', []) | self._node_labels = kwargs.get('node_labels', []) | ||||
| @@ -34,6 +34,7 @@ class StructuralSP(GraphKernel): | |||||
| self._node_kernels = kwargs.get('node_kernels', None) | self._node_kernels = kwargs.get('node_kernels', None) | ||||
| self._edge_kernels = kwargs.get('edge_kernels', None) | self._edge_kernels = kwargs.get('edge_kernels', None) | ||||
| self._compute_method = kwargs.get('compute_method', 'naive') | self._compute_method = kwargs.get('compute_method', 'naive') | ||||
| self._fcsp = kwargs.get('fcsp', True) | |||||
| self._ds_infos = kwargs.get('ds_infos', {}) | self._ds_infos = kwargs.get('ds_infos', {}) | ||||
| @@ -50,10 +51,10 @@ class StructuralSP(GraphKernel): | |||||
| else: | else: | ||||
| for g in iterator: | for g in iterator: | ||||
| splist.append(get_shortest_paths(g, self._edge_weight, self._ds_infos['directed'])) | splist.append(get_shortest_paths(g, self._edge_weight, self._ds_infos['directed'])) | ||||
| # compute Gram matrix. | # compute Gram matrix. | ||||
| gram_matrix = np.zeros((len(self._graphs), len(self._graphs))) | gram_matrix = np.zeros((len(self._graphs), len(self._graphs))) | ||||
| from itertools import combinations_with_replacement | from itertools import combinations_with_replacement | ||||
| itr = combinations_with_replacement(range(0, len(self._graphs)), 2) | itr = combinations_with_replacement(range(0, len(self._graphs)), 2) | ||||
| if self._verbose >= 2: | if self._verbose >= 2: | ||||
| @@ -72,10 +73,10 @@ class StructuralSP(GraphKernel): | |||||
| # print("error here ") | # print("error here ") | ||||
| gram_matrix[i][j] = kernel | gram_matrix[i][j] = kernel | ||||
| gram_matrix[j][i] = kernel | gram_matrix[j][i] = kernel | ||||
| return gram_matrix | return gram_matrix | ||||
| def _compute_gm_imap_unordered(self): | def _compute_gm_imap_unordered(self): | ||||
| # get shortest paths of each graph in the graphs. | # get shortest paths of each graph in the graphs. | ||||
| splist = [None] * len(self._graphs) | splist = [None] * len(self._graphs) | ||||
| @@ -87,9 +88,9 @@ class StructuralSP(GraphKernel): | |||||
| chunksize = 100 | chunksize = 100 | ||||
| # get shortest path graphs of self._graphs | # get shortest path graphs of self._graphs | ||||
| if self._compute_method == 'trie': | if self._compute_method == 'trie': | ||||
| get_sps_fun = self._wrapper_get_sps_trie | |||||
| get_sps_fun = self._wrapper_get_sps_trie | |||||
| else: | else: | ||||
| get_sps_fun = self._wrapper_get_sps_naive | |||||
| get_sps_fun = self._wrapper_get_sps_naive | |||||
| if self.verbose >= 2: | if self.verbose >= 2: | ||||
| iterator = tqdm(pool.imap_unordered(get_sps_fun, itr, chunksize), | iterator = tqdm(pool.imap_unordered(get_sps_fun, itr, chunksize), | ||||
| desc='getting shortest paths', file=sys.stdout) | desc='getting shortest paths', file=sys.stdout) | ||||
| @@ -99,24 +100,24 @@ class StructuralSP(GraphKernel): | |||||
| splist[i] = sp | splist[i] = sp | ||||
| pool.close() | pool.close() | ||||
| pool.join() | pool.join() | ||||
| # compute Gram matrix. | # compute Gram matrix. | ||||
| gram_matrix = np.zeros((len(self._graphs), len(self._graphs))) | gram_matrix = np.zeros((len(self._graphs), len(self._graphs))) | ||||
| def init_worker(spl_toshare, gs_toshare): | def init_worker(spl_toshare, gs_toshare): | ||||
| global G_spl, G_gs | global G_spl, G_gs | ||||
| G_spl = spl_toshare | G_spl = spl_toshare | ||||
| G_gs = gs_toshare | |||||
| if self._compute_method == 'trie': | |||||
| G_gs = gs_toshare | |||||
| if self._compute_method == 'trie': | |||||
| do_fun = self._wrapper_ssp_do_trie | do_fun = self._wrapper_ssp_do_trie | ||||
| else: | |||||
| do_fun = self._wrapper_ssp_do_naive | |||||
| parallel_gm(do_fun, gram_matrix, self._graphs, init_worker=init_worker, | |||||
| else: | |||||
| do_fun = self._wrapper_ssp_do_naive | |||||
| parallel_gm(do_fun, gram_matrix, self._graphs, init_worker=init_worker, | |||||
| glbv=(splist, self._graphs), n_jobs=self._n_jobs, verbose=self._verbose) | glbv=(splist, self._graphs), n_jobs=self._n_jobs, verbose=self._verbose) | ||||
| return gram_matrix | return gram_matrix | ||||
| def _compute_kernel_list_series(self, g1, g_list): | def _compute_kernel_list_series(self, g1, g_list): | ||||
| # get shortest paths of g1 and each graph in g_list. | # get shortest paths of g1 and each graph in g_list. | ||||
| sp1 = get_shortest_paths(g1, self._edge_weight, self._ds_infos['directed']) | sp1 = get_shortest_paths(g1, self._edge_weight, self._ds_infos['directed']) | ||||
| @@ -131,7 +132,7 @@ class StructuralSP(GraphKernel): | |||||
| else: | else: | ||||
| for g in iterator: | for g in iterator: | ||||
| splist.append(get_shortest_paths(g, self._edge_weight, self._ds_infos['directed'])) | splist.append(get_shortest_paths(g, self._edge_weight, self._ds_infos['directed'])) | ||||
| # compute kernel list. | # compute kernel list. | ||||
| kernel_list = [None] * len(g_list) | kernel_list = [None] * len(g_list) | ||||
| if self._verbose >= 2: | if self._verbose >= 2: | ||||
| @@ -146,10 +147,10 @@ class StructuralSP(GraphKernel): | |||||
| for i in iterator: | for i in iterator: | ||||
| kernel = self._ssp_do_naive(g1, g_list[i], sp1, splist[i]) | kernel = self._ssp_do_naive(g1, g_list[i], sp1, splist[i]) | ||||
| kernel_list[i] = kernel | kernel_list[i] = kernel | ||||
| return kernel_list | return kernel_list | ||||
| def _compute_kernel_list_imap_unordered(self, g1, g_list): | def _compute_kernel_list_imap_unordered(self, g1, g_list): | ||||
| # get shortest paths of g1 and each graph in g_list. | # get shortest paths of g1 and each graph in g_list. | ||||
| sp1 = get_shortest_paths(g1, self._edge_weight, self._ds_infos['directed']) | sp1 = get_shortest_paths(g1, self._edge_weight, self._ds_infos['directed']) | ||||
| @@ -162,9 +163,9 @@ class StructuralSP(GraphKernel): | |||||
| chunksize = 100 | chunksize = 100 | ||||
| # get shortest path graphs of g_list | # get shortest path graphs of g_list | ||||
| if self._compute_method == 'trie': | if self._compute_method == 'trie': | ||||
| get_sps_fun = self._wrapper_get_sps_trie | |||||
| get_sps_fun = self._wrapper_get_sps_trie | |||||
| else: | else: | ||||
| get_sps_fun = self._wrapper_get_sps_naive | |||||
| get_sps_fun = self._wrapper_get_sps_naive | |||||
| if self.verbose >= 2: | if self.verbose >= 2: | ||||
| iterator = tqdm(pool.imap_unordered(get_sps_fun, itr, chunksize), | iterator = tqdm(pool.imap_unordered(get_sps_fun, itr, chunksize), | ||||
| desc='getting shortest paths', file=sys.stdout) | desc='getting shortest paths', file=sys.stdout) | ||||
| @@ -174,7 +175,7 @@ class StructuralSP(GraphKernel): | |||||
| splist[i] = sp | splist[i] = sp | ||||
| pool.close() | pool.close() | ||||
| pool.join() | pool.join() | ||||
| # compute Gram matrix. | # compute Gram matrix. | ||||
| kernel_list = [None] * len(g_list) | kernel_list = [None] * len(g_list) | ||||
| @@ -182,27 +183,27 @@ class StructuralSP(GraphKernel): | |||||
| global G_sp1, G_spl, G_g1, G_gl | global G_sp1, G_spl, G_g1, G_gl | ||||
| G_sp1 = sp1_toshare | G_sp1 = sp1_toshare | ||||
| G_spl = spl_toshare | G_spl = spl_toshare | ||||
| G_g1 = g1_toshare | |||||
| G_gl = gl_toshare | |||||
| if self._compute_method == 'trie': | |||||
| G_g1 = g1_toshare | |||||
| G_gl = gl_toshare | |||||
| if self._compute_method == 'trie': | |||||
| do_fun = self._wrapper_ssp_do_trie | do_fun = self._wrapper_ssp_do_trie | ||||
| else: | |||||
| else: | |||||
| do_fun = self._wrapper_kernel_list_do | do_fun = self._wrapper_kernel_list_do | ||||
| def func_assign(result, var_to_assign): | |||||
| def func_assign(result, var_to_assign): | |||||
| var_to_assign[result[0]] = result[1] | var_to_assign[result[0]] = result[1] | ||||
| itr = range(len(g_list)) | itr = range(len(g_list)) | ||||
| len_itr = len(g_list) | len_itr = len(g_list) | ||||
| parallel_me(do_fun, func_assign, kernel_list, itr, len_itr=len_itr, | parallel_me(do_fun, func_assign, kernel_list, itr, len_itr=len_itr, | ||||
| init_worker=init_worker, glbv=(sp1, splist, g1, g_list), method='imap_unordered', n_jobs=self._n_jobs, itr_desc='Computing kernels', verbose=self._verbose) | init_worker=init_worker, glbv=(sp1, splist, g1, g_list), method='imap_unordered', n_jobs=self._n_jobs, itr_desc='Computing kernels', verbose=self._verbose) | ||||
| return kernel_list | return kernel_list | ||||
| def _wrapper_kernel_list_do(self, itr): | def _wrapper_kernel_list_do(self, itr): | ||||
| return itr, self._ssp_do_naive(G_g1, G_gl[itr], G_sp1, G_spl[itr]) | return itr, self._ssp_do_naive(G_g1, G_gl[itr], G_sp1, G_spl[itr]) | ||||
| def _compute_single_kernel_series(self, g1, g2): | def _compute_single_kernel_series(self, g1, g2): | ||||
| sp1 = get_shortest_paths(g1, self._edge_weight, self._ds_infos['directed']) | sp1 = get_shortest_paths(g1, self._edge_weight, self._ds_infos['directed']) | ||||
| sp2 = get_shortest_paths(g2, self._edge_weight, self._ds_infos['directed']) | sp2 = get_shortest_paths(g2, self._edge_weight, self._ds_infos['directed']) | ||||
| @@ -210,26 +211,33 @@ class StructuralSP(GraphKernel): | |||||
| kernel = self._ssp_do_trie(g1, g2, sp1, sp2) | kernel = self._ssp_do_trie(g1, g2, sp1, sp2) | ||||
| else: | else: | ||||
| kernel = self._ssp_do_naive(g1, g2, sp1, sp2) | kernel = self._ssp_do_naive(g1, g2, sp1, sp2) | ||||
| return kernel | |||||
| return kernel | |||||
| def _wrapper_get_sps_naive(self, itr_item): | def _wrapper_get_sps_naive(self, itr_item): | ||||
| g = itr_item[0] | g = itr_item[0] | ||||
| i = itr_item[1] | i = itr_item[1] | ||||
| return i, get_shortest_paths(g, self._edge_weight, self._ds_infos['directed']) | return i, get_shortest_paths(g, self._edge_weight, self._ds_infos['directed']) | ||||
| def _ssp_do_naive(self, g1, g2, spl1, spl2): | def _ssp_do_naive(self, g1, g2, spl1, spl2): | ||||
| if self._fcsp: # @todo: it may be put outside the _sp_do(). | |||||
| return self._sp_do_naive_fcsp(g1, g2, spl1, spl2) | |||||
| else: | |||||
| return self._sp_do_naive_naive(g1, g2, spl1, spl2) | |||||
| def _sp_do_naive_fcsp(self, g1, g2, spl1, spl2): | |||||
| kernel = 0 | kernel = 0 | ||||
| # First, compute shortest path matrices, method borrowed from FCSP. | # First, compute shortest path matrices, method borrowed from FCSP. | ||||
| vk_dict = self._get_all_node_kernels(g1, g2) | vk_dict = self._get_all_node_kernels(g1, g2) | ||||
| # Then, compute kernels between all pairs of edges, which is an idea of | # Then, compute kernels between all pairs of edges, which is an idea of | ||||
| # extension of FCSP. It suits sparse graphs, which is the most case we | # extension of FCSP. It suits sparse graphs, which is the most case we | ||||
| # went though. For dense graphs, this would be slow. | # went though. For dense graphs, this would be slow. | ||||
| ek_dict = self._get_all_edge_kernels(g1, g2) | ek_dict = self._get_all_edge_kernels(g1, g2) | ||||
| # compute graph kernels | # compute graph kernels | ||||
| if vk_dict: | if vk_dict: | ||||
| if ek_dict: | if ek_dict: | ||||
| @@ -244,6 +252,7 @@ class StructuralSP(GraphKernel): | |||||
| if not kpath: | if not kpath: | ||||
| break | break | ||||
| kernel += kpath # add up kernels of all paths | kernel += kpath # add up kernels of all paths | ||||
| # print(kernel, ',', p1, ',', p2) | |||||
| else: | else: | ||||
| for p1, p2 in product(spl1, spl2): | for p1, p2 in product(spl1, spl2): | ||||
| if len(p1) == len(p2): | if len(p1) == len(p2): | ||||
| @@ -279,7 +288,7 @@ class StructuralSP(GraphKernel): | |||||
| print(g1.nodes(data=True)) | print(g1.nodes(data=True)) | ||||
| print(g1.edges(data=True)) | print(g1.edges(data=True)) | ||||
| raise Exception | raise Exception | ||||
| # # ---- exact implementation of the Fast Computation of Shortest Path Kernel (FCSP), reference [2], sadly it is slower than the current implementation | # # ---- exact implementation of the Fast Computation of Shortest Path Kernel (FCSP), reference [2], sadly it is slower than the current implementation | ||||
| # # compute vertex kernel matrix | # # compute vertex kernel matrix | ||||
| # try: | # try: | ||||
| @@ -292,7 +301,7 @@ class StructuralSP(GraphKernel): | |||||
| # vk_mat[i1][i2] = kn( | # vk_mat[i1][i2] = kn( | ||||
| # n1[1][node_label], n2[1][node_label], | # n1[1][node_label], n2[1][node_label], | ||||
| # [n1[1]['attributes']], [n2[1]['attributes']]) | # [n1[1]['attributes']], [n2[1]['attributes']]) | ||||
| # range1 = range(0, len(edge_w_g[i])) | # range1 = range(0, len(edge_w_g[i])) | ||||
| # range2 = range(0, len(edge_w_g[j])) | # range2 = range(0, len(edge_w_g[j])) | ||||
| # for i1 in range1: | # for i1 in range1: | ||||
| @@ -309,18 +318,137 @@ class StructuralSP(GraphKernel): | |||||
| # kn2 = vk_mat[x1][y2] * vk_mat[y1][x2] | # kn2 = vk_mat[x1][y2] * vk_mat[y1][x2] | ||||
| # Kmatrix += kn1 + kn2 | # Kmatrix += kn1 + kn2 | ||||
| return kernel | return kernel | ||||
| def _sp_do_naive_naive(self, g1, g2, spl1, spl2): | |||||
| kernel = 0 | |||||
| # Define the function to compute kernels between vertices in each condition. | |||||
| if len(self._node_labels) > 0: | |||||
| # node symb and non-synb labeled | |||||
| if len(self._node_attrs) > 0: | |||||
| def compute_vk(n1, n2): | |||||
| kn = self._node_kernels['mix'] | |||||
| n1_labels = [g1.nodes[n1][nl] for nl in self._node_labels] | |||||
| n2_labels = [g2.nodes[n2][nl] for nl in self._node_labels] | |||||
| n1_attrs = [g1.nodes[n1][na] for na in self._node_attrs] | |||||
| n2_attrs = [g2.nodes[n2][na] for na in self._node_attrs] | |||||
| return kn(n1_labels, n2_labels, n1_attrs, n2_attrs) | |||||
| # node symb labeled | |||||
| else: | |||||
| def compute_vk(n1, n2): | |||||
| kn = self._node_kernels['symb'] | |||||
| n1_labels = [g1.nodes[n1][nl] for nl in self._node_labels] | |||||
| n2_labels = [g2.nodes[n2][nl] for nl in self._node_labels] | |||||
| return kn(n1_labels, n2_labels) | |||||
| else: | |||||
| # node non-synb labeled | |||||
| if len(self._node_attrs) > 0: | |||||
| def compute_vk(n1, n2): | |||||
| kn = self._node_kernels['nsymb'] | |||||
| n1_attrs = [g1.nodes[n1][na] for na in self._node_attrs] | |||||
| n2_attrs = [g2.nodes[n2][na] for na in self._node_attrs] | |||||
| return kn(n1_attrs, n2_attrs) | |||||
| # # node unlabeled | |||||
| # else: | |||||
| # for e1, e2 in product(g1.edges(data=True), g2.edges(data=True)): | |||||
| # if e1[2]['cost'] == e2[2]['cost']: | |||||
| # kernel += 1 | |||||
| # return kernel | |||||
| # Define the function to compute kernels between edges in each condition. | |||||
| if len(self._edge_labels) > 0: | |||||
| # edge symb and non-synb labeled | |||||
| if len(self._edge_attrs) > 0: | |||||
| def compute_ek(e1, e2): | |||||
| ke = self._edge_kernels['mix'] | |||||
| e1_labels = [g1.edges[e1][el] for el in self._edge_labels] | |||||
| e2_labels = [g2.edges[e2][el] for el in self._edge_labels] | |||||
| e1_attrs = [g1.edges[e1][ea] for ea in self._edge_attrs] | |||||
| e2_attrs = [g2.edges[e2][ea] for ea in self._edge_attrs] | |||||
| return ke(e1_labels, e2_labels, e1_attrs, e2_attrs) | |||||
| # edge symb labeled | |||||
| else: | |||||
| def compute_ek(e1, e2): | |||||
| ke = self._edge_kernels['symb'] | |||||
| e1_labels = [g1.edges[e1][el] for el in self._edge_labels] | |||||
| e2_labels = [g2.edges[e2][el] for el in self._edge_labels] | |||||
| return ke(e1_labels, e2_labels) | |||||
| else: | |||||
| # edge non-synb labeled | |||||
| if len(self._edge_attrs) > 0: | |||||
| def compute_ek(e1, e2): | |||||
| ke = self._edge_kernels['nsymb'] | |||||
| e1_attrs = [g1.edges[e1][ea] for ea in self._edge_attrs] | |||||
| e2_attrs = [g2.edges[e2][ea] for ea in self._edge_attrs] | |||||
| return ke(e1_attrs, e2_attrs) | |||||
| # compute graph kernels | |||||
| if len(self._node_labels) > 0 or len(self._node_attrs) > 0: | |||||
| if len(self._edge_labels) > 0 or len(self._edge_attrs) > 0: | |||||
| for p1, p2 in product(spl1, spl2): | |||||
| if len(p1) == len(p2): | |||||
| kpath = compute_vk(p1[0], p2[0]) | |||||
| if kpath: | |||||
| for idx in range(1, len(p1)): | |||||
| kpath *= compute_vk(p1[idx], p2[idx]) * \ | |||||
| compute_ek((p1[idx-1], p1[idx]), | |||||
| (p2[idx-1], p2[idx])) | |||||
| if not kpath: | |||||
| break | |||||
| kernel += kpath # add up kernels of all paths | |||||
| # print(kernel, ',', p1, ',', p2) | |||||
| else: | |||||
| for p1, p2 in product(spl1, spl2): | |||||
| if len(p1) == len(p2): | |||||
| kpath = compute_vk(p1[0], p2[0]) | |||||
| if kpath: | |||||
| for idx in range(1, len(p1)): | |||||
| kpath *= compute_vk(p1[idx], p2[idx]) | |||||
| if not kpath: | |||||
| break | |||||
| kernel += kpath # add up kernels of all paths | |||||
| else: | |||||
| if len(self._edge_labels) > 0 or len(self._edge_attrs) > 0: | |||||
| for p1, p2 in product(spl1, spl2): | |||||
| if len(p1) == len(p2): | |||||
| if len(p1) == 0: | |||||
| kernel += 1 | |||||
| else: | |||||
| kpath = 1 | |||||
| for idx in range(0, len(p1) - 1): | |||||
| kpath *= compute_ek((p1[idx], p1[idx+1]), | |||||
| (p2[idx], p2[idx+1])) | |||||
| if not kpath: | |||||
| break | |||||
| kernel += kpath # add up kernels of all paths | |||||
| else: | |||||
| for p1, p2 in product(spl1, spl2): | |||||
| if len(p1) == len(p2): | |||||
| kernel += 1 | |||||
| try: | |||||
| kernel = kernel / (len(spl1) * len(spl2)) # Compute mean average | |||||
| except ZeroDivisionError: | |||||
| print(spl1, spl2) | |||||
| print(g1.nodes(data=True)) | |||||
| print(g1.edges(data=True)) | |||||
| raise Exception | |||||
| return kernel | |||||
| def _wrapper_ssp_do_naive(self, itr): | def _wrapper_ssp_do_naive(self, itr): | ||||
| i = itr[0] | i = itr[0] | ||||
| j = itr[1] | j = itr[1] | ||||
| return i, j, self._ssp_do_naive(G_gs[i], G_gs[j], G_spl[i], G_spl[j]) | return i, j, self._ssp_do_naive(G_gs[i], G_gs[j], G_spl[i], G_spl[j]) | ||||
| def _get_all_node_kernels(self, g1, g2): | def _get_all_node_kernels(self, g1, g2): | ||||
| return compute_vertex_kernels(g1, g2, self._node_kernels, node_labels=self._node_labels, node_attrs=self._node_attrs) | return compute_vertex_kernels(g1, g2, self._node_kernels, node_labels=self._node_labels, node_attrs=self._node_attrs) | ||||
| def _get_all_edge_kernels(self, g1, g2): | def _get_all_edge_kernels(self, g1, g2): | ||||
| # compute kernels between all pairs of edges, which is an idea of | # compute kernels between all pairs of edges, which is an idea of | ||||
| # extension of FCSP. It suits sparse graphs, which is the most case we | # extension of FCSP. It suits sparse graphs, which is the most case we | ||||
| @@ -368,5 +496,5 @@ class StructuralSP(GraphKernel): | |||||
| # edge unlabeled | # edge unlabeled | ||||
| else: | else: | ||||
| pass | pass | ||||
| return ek_dict | |||||
| return ek_dict | |||||
+ 97
- 97
gklearn/kernels/structuralspKernel.py
View File
| @@ -5,9 +5,9 @@ Created on Thu Sep 27 10:56:23 2018 | |||||
| @author: linlin | @author: linlin | ||||
| @references: | |||||
| @references: | |||||
| [1] Suard F, Rakotomamonjy A, Bensrhair A. Kernel on Bag of Paths For | |||||
| [1] Suard F, Rakotomamonjy A, Bensrhair A. Kernel on Bag of Paths For | |||||
| Measuring Similarity of Shapes. InESANN 2007 Apr 25 (pp. 355-360). | Measuring Similarity of Shapes. InESANN 2007 Apr 25 (pp. 355-360). | ||||
| """ | """ | ||||
| @@ -43,7 +43,7 @@ def structuralspkernel(*args, | |||||
| ---------- | ---------- | ||||
| Gn : List of NetworkX graph | Gn : List of NetworkX graph | ||||
| List of graphs between which the kernels are computed. | List of graphs between which the kernels are computed. | ||||
| G1, G2 : NetworkX graphs | G1, G2 : NetworkX graphs | ||||
| Two graphs between which the kernel is computed. | Two graphs between which the kernel is computed. | ||||
| @@ -51,25 +51,25 @@ def structuralspkernel(*args, | |||||
| Node attribute used as label. The default node label is atom. | Node attribute used as label. The default node label is atom. | ||||
| edge_weight : string | edge_weight : string | ||||
| Edge attribute name corresponding to the edge weight. Applied for the | |||||
| Edge attribute name corresponding to the edge weight. Applied for the | |||||
| computation of the shortest paths. | computation of the shortest paths. | ||||
| edge_label : string | edge_label : string | ||||
| Edge attribute used as label. The default edge label is bond_type. | Edge attribute used as label. The default edge label is bond_type. | ||||
| node_kernels : dict | node_kernels : dict | ||||
| A dictionary of kernel functions for nodes, including 3 items: 'symb' | |||||
| for symbolic node labels, 'nsymb' for non-symbolic node labels, 'mix' | |||||
| for both labels. The first 2 functions take two node labels as | |||||
| A dictionary of kernel functions for nodes, including 3 items: 'symb' | |||||
| for symbolic node labels, 'nsymb' for non-symbolic node labels, 'mix' | |||||
| for both labels. The first 2 functions take two node labels as | |||||
| parameters, and the 'mix' function takes 4 parameters, a symbolic and a | parameters, and the 'mix' function takes 4 parameters, a symbolic and a | ||||
| non-symbolic label for each the two nodes. Each label is in form of 2-D | non-symbolic label for each the two nodes. Each label is in form of 2-D | ||||
| dimension array (n_samples, n_features). Each function returns a number | dimension array (n_samples, n_features). Each function returns a number | ||||
| as the kernel value. Ignored when nodes are unlabeled. | as the kernel value. Ignored when nodes are unlabeled. | ||||
| edge_kernels : dict | edge_kernels : dict | ||||
| A dictionary of kernel functions for edges, including 3 items: 'symb' | |||||
| for symbolic edge labels, 'nsymb' for non-symbolic edge labels, 'mix' | |||||
| for both labels. The first 2 functions take two edge labels as | |||||
| A dictionary of kernel functions for edges, including 3 items: 'symb' | |||||
| for symbolic edge labels, 'nsymb' for non-symbolic edge labels, 'mix' | |||||
| for both labels. The first 2 functions take two edge labels as | |||||
| parameters, and the 'mix' function takes 4 parameters, a symbolic and a | parameters, and the 'mix' function takes 4 parameters, a symbolic and a | ||||
| non-symbolic label for each the two edges. Each label is in form of 2-D | non-symbolic label for each the two edges. Each label is in form of 2-D | ||||
| dimension array (n_samples, n_features). Each function returns a number | dimension array (n_samples, n_features). Each function returns a number | ||||
| @@ -89,7 +89,7 @@ def structuralspkernel(*args, | |||||
| Return | Return | ||||
| ------ | ------ | ||||
| Kmatrix : Numpy matrix | Kmatrix : Numpy matrix | ||||
| Kernel matrix, each element of which is the mean average structural | |||||
| Kernel matrix, each element of which is the mean average structural | |||||
| shortest path kernel between 2 praphs. | shortest path kernel between 2 praphs. | ||||
| """ | """ | ||||
| # pre-process | # pre-process | ||||
| @@ -135,9 +135,9 @@ def structuralspkernel(*args, | |||||
| chunksize = 100 | chunksize = 100 | ||||
| # get shortest path graphs of Gn | # get shortest path graphs of Gn | ||||
| if compute_method == 'trie': | if compute_method == 'trie': | ||||
| getsp_partial = partial(wrapper_getSP_trie, weight, ds_attrs['is_directed']) | |||||
| getsp_partial = partial(wrapper_getSP_trie, weight, ds_attrs['is_directed']) | |||||
| else: | else: | ||||
| getsp_partial = partial(wrapper_getSP_naive, weight, ds_attrs['is_directed']) | |||||
| getsp_partial = partial(wrapper_getSP_naive, weight, ds_attrs['is_directed']) | |||||
| if verbose: | if verbose: | ||||
| iterator = tqdm(pool.imap_unordered(getsp_partial, itr, chunksize), | iterator = tqdm(pool.imap_unordered(getsp_partial, itr, chunksize), | ||||
| desc='getting shortest paths', file=sys.stdout) | desc='getting shortest paths', file=sys.stdout) | ||||
| @@ -161,17 +161,17 @@ def structuralspkernel(*args, | |||||
| else: | else: | ||||
| for g in iterator: | for g in iterator: | ||||
| splist.append(get_shortest_paths(g, weight, ds_attrs['is_directed'])) | splist.append(get_shortest_paths(g, weight, ds_attrs['is_directed'])) | ||||
| # ss = 0 | # ss = 0 | ||||
| # ss += sys.getsizeof(splist) | # ss += sys.getsizeof(splist) | ||||
| # for spss in splist: | # for spss in splist: | ||||
| # ss += sys.getsizeof(spss) | # ss += sys.getsizeof(spss) | ||||
| # for spp in spss: | # for spp in spss: | ||||
| # ss += sys.getsizeof(spp) | # ss += sys.getsizeof(spp) | ||||
| # time.sleep(20) | # time.sleep(20) | ||||
| # # ---- only for the Fast Computation of Shortest Path Kernel (FCSP) | # # ---- only for the Fast Computation of Shortest Path Kernel (FCSP) | ||||
| @@ -194,21 +194,21 @@ def structuralspkernel(*args, | |||||
| Kmatrix = np.zeros((len(Gn), len(Gn))) | Kmatrix = np.zeros((len(Gn), len(Gn))) | ||||
| # ---- use pool.imap_unordered to parallel and track progress. ---- | |||||
| # ---- use pool.imap_unordered to parallel and track progress. ---- | |||||
| if parallel == 'imap_unordered': | if parallel == 'imap_unordered': | ||||
| def init_worker(spl_toshare, gs_toshare): | def init_worker(spl_toshare, gs_toshare): | ||||
| global G_spl, G_gs | global G_spl, G_gs | ||||
| G_spl = spl_toshare | G_spl = spl_toshare | ||||
| G_gs = gs_toshare | |||||
| if compute_method == 'trie': | |||||
| do_partial = partial(wrapper_ssp_do_trie, ds_attrs, node_label, edge_label, | |||||
| node_kernels, edge_kernels) | |||||
| parallel_gm(do_partial, Kmatrix, Gn, init_worker=init_worker, | |||||
| glbv=(splist, Gn), n_jobs=n_jobs, chunksize=chunksize, verbose=verbose) | |||||
| else: | |||||
| do_partial = partial(wrapper_ssp_do, ds_attrs, node_label, edge_label, | |||||
| node_kernels, edge_kernels) | |||||
| parallel_gm(do_partial, Kmatrix, Gn, init_worker=init_worker, | |||||
| G_gs = gs_toshare | |||||
| if compute_method == 'trie': | |||||
| do_partial = partial(wrapper_ssp_do_trie, ds_attrs, node_label, edge_label, | |||||
| node_kernels, edge_kernels) | |||||
| parallel_gm(do_partial, Kmatrix, Gn, init_worker=init_worker, | |||||
| glbv=(splist, Gn), n_jobs=n_jobs, chunksize=chunksize, verbose=verbose) | |||||
| else: | |||||
| do_partial = partial(wrapper_ssp_do, ds_attrs, node_label, edge_label, | |||||
| node_kernels, edge_kernels) | |||||
| parallel_gm(do_partial, Kmatrix, Gn, init_worker=init_worker, | |||||
| glbv=(splist, Gn), n_jobs=n_jobs, chunksize=chunksize, verbose=verbose) | glbv=(splist, Gn), n_jobs=n_jobs, chunksize=chunksize, verbose=verbose) | ||||
| # ---- direct running, normally use single CPU core. ---- | # ---- direct running, normally use single CPU core. ---- | ||||
| elif parallel is None: | elif parallel is None: | ||||
| @@ -232,10 +232,10 @@ def structuralspkernel(*args, | |||||
| # print("error here ") | # print("error here ") | ||||
| Kmatrix[i][j] = kernel | Kmatrix[i][j] = kernel | ||||
| Kmatrix[j][i] = kernel | Kmatrix[j][i] = kernel | ||||
| # # ---- use pool.map to parallel. ---- | # # ---- use pool.map to parallel. ---- | ||||
| # pool = Pool(n_jobs) | # pool = Pool(n_jobs) | ||||
| # do_partial = partial(wrapper_ssp_do, ds_attrs, node_label, edge_label, | |||||
| # do_partial = partial(wrapper_ssp_do, ds_attrs, node_label, edge_label, | |||||
| # node_kernels, edge_kernels) | # node_kernels, edge_kernels) | ||||
| # itr = zip(combinations_with_replacement(Gn, 2), | # itr = zip(combinations_with_replacement(Gn, 2), | ||||
| # combinations_with_replacement(splist, 2), | # combinations_with_replacement(splist, 2), | ||||
| @@ -249,7 +249,7 @@ def structuralspkernel(*args, | |||||
| # pool.join() | # pool.join() | ||||
| # # ---- use pool.imap_unordered to parallel and track progress. ---- | # # ---- use pool.imap_unordered to parallel and track progress. ---- | ||||
| # do_partial = partial(wrapper_ssp_do, ds_attrs, node_label, edge_label, | |||||
| # do_partial = partial(wrapper_ssp_do, ds_attrs, node_label, edge_label, | |||||
| # node_kernels, edge_kernels) | # node_kernels, edge_kernels) | ||||
| # itr = zip(combinations_with_replacement(Gn, 2), | # itr = zip(combinations_with_replacement(Gn, 2), | ||||
| # combinations_with_replacement(splist, 2), | # combinations_with_replacement(splist, 2), | ||||
| @@ -282,7 +282,7 @@ def structuralspkernel(*args, | |||||
| def structuralspkernel_do(g1, g2, spl1, spl2, ds_attrs, node_label, edge_label, | def structuralspkernel_do(g1, g2, spl1, spl2, ds_attrs, node_label, edge_label, | ||||
| node_kernels, edge_kernels): | node_kernels, edge_kernels): | ||||
| kernel = 0 | kernel = 0 | ||||
| # First, compute shortest path matrices, method borrowed from FCSP. | # First, compute shortest path matrices, method borrowed from FCSP. | ||||
| @@ -373,25 +373,25 @@ def structuralspkernel_do(g1, g2, spl1, spl2, ds_attrs, node_label, edge_label, | |||||
| return kernel | return kernel | ||||
| def wrapper_ssp_do(ds_attrs, node_label, edge_label, node_kernels, | |||||
| def wrapper_ssp_do(ds_attrs, node_label, edge_label, node_kernels, | |||||
| edge_kernels, itr): | edge_kernels, itr): | ||||
| i = itr[0] | i = itr[0] | ||||
| j = itr[1] | j = itr[1] | ||||
| return i, j, structuralspkernel_do(G_gs[i], G_gs[j], G_spl[i], G_spl[j], | |||||
| ds_attrs, node_label, edge_label, | |||||
| return i, j, structuralspkernel_do(G_gs[i], G_gs[j], G_spl[i], G_spl[j], | |||||
| ds_attrs, node_label, edge_label, | |||||
| node_kernels, edge_kernels) | node_kernels, edge_kernels) | ||||
| def ssp_do_trie(g1, g2, trie1, trie2, ds_attrs, node_label, edge_label, | def ssp_do_trie(g1, g2, trie1, trie2, ds_attrs, node_label, edge_label, | ||||
| node_kernels, edge_kernels): | node_kernels, edge_kernels): | ||||
| # # traverse all paths in graph1. Deep-first search is applied. | # # traverse all paths in graph1. Deep-first search is applied. | ||||
| # def traverseBothTrie(root, trie2, kernel, pcurrent=[]): | # def traverseBothTrie(root, trie2, kernel, pcurrent=[]): | ||||
| # for key, node in root['children'].items(): | # for key, node in root['children'].items(): | ||||
| # pcurrent.append(key) | # pcurrent.append(key) | ||||
| # if node['isEndOfWord']: | # if node['isEndOfWord']: | ||||
| # # print(node['count']) | # # print(node['count']) | ||||
| # traverseTrie2(trie2.root, pcurrent, kernel, | |||||
| # traverseTrie2(trie2.root, pcurrent, kernel, | |||||
| # pcurrent=[]) | # pcurrent=[]) | ||||
| # if node['children'] != {}: | # if node['children'] != {}: | ||||
| # traverseBothTrie(node, trie2, kernel, pcurrent) | # traverseBothTrie(node, trie2, kernel, pcurrent) | ||||
| @@ -399,14 +399,14 @@ def ssp_do_trie(g1, g2, trie1, trie2, ds_attrs, node_label, edge_label, | |||||
| # del pcurrent[-1] | # del pcurrent[-1] | ||||
| # if pcurrent != []: | # if pcurrent != []: | ||||
| # del pcurrent[-1] | # del pcurrent[-1] | ||||
| # | |||||
| # | |||||
| # # traverse all paths in graph2 and find out those that are not in | |||||
| # # graph1. Deep-first search is applied. | |||||
| # | |||||
| # | |||||
| # # traverse all paths in graph2 and find out those that are not in | |||||
| # # graph1. Deep-first search is applied. | |||||
| # def traverseTrie2(root, p1, kernel, pcurrent=[]): | # def traverseTrie2(root, p1, kernel, pcurrent=[]): | ||||
| # for key, node in root['children'].items(): | # for key, node in root['children'].items(): | ||||
| # pcurrent.append(key) | # pcurrent.append(key) | ||||
| # if node['isEndOfWord']: | |||||
| # if node['isEndOfWord']: | |||||
| # # print(node['count']) | # # print(node['count']) | ||||
| # kernel[0] += computePathKernel(p1, pcurrent, vk_dict, ek_dict) | # kernel[0] += computePathKernel(p1, pcurrent, vk_dict, ek_dict) | ||||
| # if node['children'] != {}: | # if node['children'] != {}: | ||||
| @@ -415,8 +415,8 @@ def ssp_do_trie(g1, g2, trie1, trie2, ds_attrs, node_label, edge_label, | |||||
| # del pcurrent[-1] | # del pcurrent[-1] | ||||
| # if pcurrent != []: | # if pcurrent != []: | ||||
| # del pcurrent[-1] | # del pcurrent[-1] | ||||
| # | |||||
| # | |||||
| # | |||||
| # | |||||
| # kernel = [0] | # kernel = [0] | ||||
| # | # | ||||
| # # First, compute shortest path matrices, method borrowed from FCSP. | # # First, compute shortest path matrices, method borrowed from FCSP. | ||||
| @@ -437,7 +437,7 @@ def ssp_do_trie(g1, g2, trie1, trie2, ds_attrs, node_label, edge_label, | |||||
| # pcurrent.append(key) | # pcurrent.append(key) | ||||
| # if node['isEndOfWord']: | # if node['isEndOfWord']: | ||||
| # # print(node['count']) | # # print(node['count']) | ||||
| # traverseTrie2(trie2.root, pcurrent, kernel, vk_dict, ek_dict, | |||||
| # traverseTrie2(trie2.root, pcurrent, kernel, vk_dict, ek_dict, | |||||
| # pcurrent=[]) | # pcurrent=[]) | ||||
| # if node['children'] != {}: | # if node['children'] != {}: | ||||
| # traverseBothTrie(node, trie2, kernel, vk_dict, ek_dict, pcurrent) | # traverseBothTrie(node, trie2, kernel, vk_dict, ek_dict, pcurrent) | ||||
| @@ -445,14 +445,14 @@ def ssp_do_trie(g1, g2, trie1, trie2, ds_attrs, node_label, edge_label, | |||||
| # del pcurrent[-1] | # del pcurrent[-1] | ||||
| # if pcurrent != []: | # if pcurrent != []: | ||||
| # del pcurrent[-1] | # del pcurrent[-1] | ||||
| # | |||||
| # | |||||
| # # traverse all paths in graph2 and find out those that are not in | |||||
| # # graph1. Deep-first search is applied. | |||||
| # | |||||
| # | |||||
| # # traverse all paths in graph2 and find out those that are not in | |||||
| # # graph1. Deep-first search is applied. | |||||
| # def traverseTrie2(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | # def traverseTrie2(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | ||||
| # for key, node in root['children'].items(): | # for key, node in root['children'].items(): | ||||
| # pcurrent.append(key) | # pcurrent.append(key) | ||||
| # if node['isEndOfWord']: | |||||
| # if node['isEndOfWord']: | |||||
| # # print(node['count']) | # # print(node['count']) | ||||
| # kernel[0] += computePathKernel(p1, pcurrent, vk_dict, ek_dict) | # kernel[0] += computePathKernel(p1, pcurrent, vk_dict, ek_dict) | ||||
| # if node['children'] != {}: | # if node['children'] != {}: | ||||
| @@ -461,8 +461,8 @@ def ssp_do_trie(g1, g2, trie1, trie2, ds_attrs, node_label, edge_label, | |||||
| # del pcurrent[-1] | # del pcurrent[-1] | ||||
| # if pcurrent != []: | # if pcurrent != []: | ||||
| # del pcurrent[-1] | # del pcurrent[-1] | ||||
| kernel = [0] | kernel = [0] | ||||
| # First, compute shortest path matrices, method borrowed from FCSP. | # First, compute shortest path matrices, method borrowed from FCSP. | ||||
| @@ -483,20 +483,20 @@ def ssp_do_trie(g1, g2, trie1, trie2, ds_attrs, node_label, edge_label, | |||||
| if ek_dict: | if ek_dict: | ||||
| traverseBothTriee(trie1[0].root, trie2[0], kernel, vk_dict, ek_dict) | traverseBothTriee(trie1[0].root, trie2[0], kernel, vk_dict, ek_dict) | ||||
| else: | else: | ||||
| traverseBothTrieu(trie1[0].root, trie2[0], kernel, vk_dict, ek_dict) | |||||
| traverseBothTrieu(trie1[0].root, trie2[0], kernel, vk_dict, ek_dict) | |||||
| kernel = kernel[0] / (trie1[1] * trie2[1]) # Compute mean average | kernel = kernel[0] / (trie1[1] * trie2[1]) # Compute mean average | ||||
| return kernel | return kernel | ||||
| def wrapper_ssp_do_trie(ds_attrs, node_label, edge_label, node_kernels, | |||||
| def wrapper_ssp_do_trie(ds_attrs, node_label, edge_label, node_kernels, | |||||
| edge_kernels, itr): | edge_kernels, itr): | ||||
| i = itr[0] | i = itr[0] | ||||
| j = itr[1] | j = itr[1] | ||||
| return i, j, ssp_do_trie(G_gs[i], G_gs[j], G_spl[i], G_spl[j], ds_attrs, | |||||
| return i, j, ssp_do_trie(G_gs[i], G_gs[j], G_spl[i], G_spl[j], ds_attrs, | |||||
| node_label, edge_label, node_kernels, edge_kernels) | node_label, edge_label, node_kernels, edge_kernels) | ||||
| def getAllNodeKernels(g1, g2, node_kernels, node_label, ds_attrs): | def getAllNodeKernels(g1, g2, node_kernels, node_label, ds_attrs): | ||||
| # compute shortest path matrices, method borrowed from FCSP. | # compute shortest path matrices, method borrowed from FCSP. | ||||
| @@ -528,7 +528,7 @@ def getAllNodeKernels(g1, g2, node_kernels, node_label, ds_attrs): | |||||
| # node unlabeled | # node unlabeled | ||||
| else: | else: | ||||
| pass | pass | ||||
| return vk_dict | return vk_dict | ||||
| @@ -573,17 +573,17 @@ def getAllEdgeKernels(g1, g2, edge_kernels, edge_label, ds_attrs): | |||||
| # edge unlabeled | # edge unlabeled | ||||
| else: | else: | ||||
| pass | pass | ||||
| return ek_dict | |||||
| return ek_dict | |||||
| # traverse all paths in graph1. Deep-first search is applied. | # traverse all paths in graph1. Deep-first search is applied. | ||||
| def traverseBothTriem(root, trie2, kernel, vk_dict, ek_dict, pcurrent=[]): | def traverseBothTriem(root, trie2, kernel, vk_dict, ek_dict, pcurrent=[]): | ||||
| for key, node in root['children'].items(): | for key, node in root['children'].items(): | ||||
| pcurrent.append(key) | pcurrent.append(key) | ||||
| if node['isEndOfWord']: | if node['isEndOfWord']: | ||||
| # print(node['count']) | # print(node['count']) | ||||
| traverseTrie2m(trie2.root, pcurrent, kernel, vk_dict, ek_dict, | |||||
| traverseTrie2m(trie2.root, pcurrent, kernel, vk_dict, ek_dict, | |||||
| pcurrent=[]) | pcurrent=[]) | ||||
| if node['children'] != {}: | if node['children'] != {}: | ||||
| traverseBothTriem(node, trie2, kernel, vk_dict, ek_dict, pcurrent) | traverseBothTriem(node, trie2, kernel, vk_dict, ek_dict, pcurrent) | ||||
| @@ -591,14 +591,14 @@ def traverseBothTriem(root, trie2, kernel, vk_dict, ek_dict, pcurrent=[]): | |||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| if pcurrent != []: | if pcurrent != []: | ||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| # traverse all paths in graph2 and find out those that are not in | |||||
| # graph1. Deep-first search is applied. | |||||
| # traverse all paths in graph2 and find out those that are not in | |||||
| # graph1. Deep-first search is applied. | |||||
| def traverseTrie2m(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | def traverseTrie2m(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | ||||
| for key, node in root['children'].items(): | for key, node in root['children'].items(): | ||||
| pcurrent.append(key) | pcurrent.append(key) | ||||
| if node['isEndOfWord']: | |||||
| if node['isEndOfWord']: | |||||
| # print(node['count']) | # print(node['count']) | ||||
| if len(p1) == len(pcurrent): | if len(p1) == len(pcurrent): | ||||
| kpath = vk_dict[(p1[0], pcurrent[0])] | kpath = vk_dict[(p1[0], pcurrent[0])] | ||||
| @@ -616,7 +616,7 @@ def traverseTrie2m(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | |||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| if pcurrent != []: | if pcurrent != []: | ||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| # traverse all paths in graph1. Deep-first search is applied. | # traverse all paths in graph1. Deep-first search is applied. | ||||
| def traverseBothTriev(root, trie2, kernel, vk_dict, ek_dict, pcurrent=[]): | def traverseBothTriev(root, trie2, kernel, vk_dict, ek_dict, pcurrent=[]): | ||||
| @@ -624,7 +624,7 @@ def traverseBothTriev(root, trie2, kernel, vk_dict, ek_dict, pcurrent=[]): | |||||
| pcurrent.append(key) | pcurrent.append(key) | ||||
| if node['isEndOfWord']: | if node['isEndOfWord']: | ||||
| # print(node['count']) | # print(node['count']) | ||||
| traverseTrie2v(trie2.root, pcurrent, kernel, vk_dict, ek_dict, | |||||
| traverseTrie2v(trie2.root, pcurrent, kernel, vk_dict, ek_dict, | |||||
| pcurrent=[]) | pcurrent=[]) | ||||
| if node['children'] != {}: | if node['children'] != {}: | ||||
| traverseBothTriev(node, trie2, kernel, vk_dict, ek_dict, pcurrent) | traverseBothTriev(node, trie2, kernel, vk_dict, ek_dict, pcurrent) | ||||
| @@ -632,14 +632,14 @@ def traverseBothTriev(root, trie2, kernel, vk_dict, ek_dict, pcurrent=[]): | |||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| if pcurrent != []: | if pcurrent != []: | ||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| # traverse all paths in graph2 and find out those that are not in | |||||
| # graph1. Deep-first search is applied. | |||||
| # traverse all paths in graph2 and find out those that are not in | |||||
| # graph1. Deep-first search is applied. | |||||
| def traverseTrie2v(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | def traverseTrie2v(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | ||||
| for key, node in root['children'].items(): | for key, node in root['children'].items(): | ||||
| pcurrent.append(key) | pcurrent.append(key) | ||||
| if node['isEndOfWord']: | |||||
| if node['isEndOfWord']: | |||||
| # print(node['count']) | # print(node['count']) | ||||
| if len(p1) == len(pcurrent): | if len(p1) == len(pcurrent): | ||||
| kpath = vk_dict[(p1[0], pcurrent[0])] | kpath = vk_dict[(p1[0], pcurrent[0])] | ||||
| @@ -655,15 +655,15 @@ def traverseTrie2v(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | |||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| if pcurrent != []: | if pcurrent != []: | ||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| # traverse all paths in graph1. Deep-first search is applied. | # traverse all paths in graph1. Deep-first search is applied. | ||||
| def traverseBothTriee(root, trie2, kernel, vk_dict, ek_dict, pcurrent=[]): | def traverseBothTriee(root, trie2, kernel, vk_dict, ek_dict, pcurrent=[]): | ||||
| for key, node in root['children'].items(): | for key, node in root['children'].items(): | ||||
| pcurrent.append(key) | pcurrent.append(key) | ||||
| if node['isEndOfWord']: | if node['isEndOfWord']: | ||||
| # print(node['count']) | # print(node['count']) | ||||
| traverseTrie2e(trie2.root, pcurrent, kernel, vk_dict, ek_dict, | |||||
| traverseTrie2e(trie2.root, pcurrent, kernel, vk_dict, ek_dict, | |||||
| pcurrent=[]) | pcurrent=[]) | ||||
| if node['children'] != {}: | if node['children'] != {}: | ||||
| traverseBothTriee(node, trie2, kernel, vk_dict, ek_dict, pcurrent) | traverseBothTriee(node, trie2, kernel, vk_dict, ek_dict, pcurrent) | ||||
| @@ -671,14 +671,14 @@ def traverseBothTriee(root, trie2, kernel, vk_dict, ek_dict, pcurrent=[]): | |||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| if pcurrent != []: | if pcurrent != []: | ||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| # traverse all paths in graph2 and find out those that are not in | |||||
| # graph1. Deep-first search is applied. | |||||
| # traverse all paths in graph2 and find out those that are not in | |||||
| # graph1. Deep-first search is applied. | |||||
| def traverseTrie2e(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | def traverseTrie2e(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | ||||
| for key, node in root['children'].items(): | for key, node in root['children'].items(): | ||||
| pcurrent.append(key) | pcurrent.append(key) | ||||
| if node['isEndOfWord']: | |||||
| if node['isEndOfWord']: | |||||
| # print(node['count']) | # print(node['count']) | ||||
| if len(p1) == len(pcurrent): | if len(p1) == len(pcurrent): | ||||
| if len(p1) == 0: | if len(p1) == 0: | ||||
| @@ -697,15 +697,15 @@ def traverseTrie2e(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | |||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| if pcurrent != []: | if pcurrent != []: | ||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| # traverse all paths in graph1. Deep-first search is applied. | # traverse all paths in graph1. Deep-first search is applied. | ||||
| def traverseBothTrieu(root, trie2, kernel, vk_dict, ek_dict, pcurrent=[]): | def traverseBothTrieu(root, trie2, kernel, vk_dict, ek_dict, pcurrent=[]): | ||||
| for key, node in root['children'].items(): | for key, node in root['children'].items(): | ||||
| pcurrent.append(key) | pcurrent.append(key) | ||||
| if node['isEndOfWord']: | if node['isEndOfWord']: | ||||
| # print(node['count']) | # print(node['count']) | ||||
| traverseTrie2u(trie2.root, pcurrent, kernel, vk_dict, ek_dict, | |||||
| traverseTrie2u(trie2.root, pcurrent, kernel, vk_dict, ek_dict, | |||||
| pcurrent=[]) | pcurrent=[]) | ||||
| if node['children'] != {}: | if node['children'] != {}: | ||||
| traverseBothTrieu(node, trie2, kernel, vk_dict, ek_dict, pcurrent) | traverseBothTrieu(node, trie2, kernel, vk_dict, ek_dict, pcurrent) | ||||
| @@ -713,14 +713,14 @@ def traverseBothTrieu(root, trie2, kernel, vk_dict, ek_dict, pcurrent=[]): | |||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| if pcurrent != []: | if pcurrent != []: | ||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| # traverse all paths in graph2 and find out those that are not in | |||||
| # graph1. Deep-first search is applied. | |||||
| # traverse all paths in graph2 and find out those that are not in | |||||
| # graph1. Deep-first search is applied. | |||||
| def traverseTrie2u(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | def traverseTrie2u(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | ||||
| for key, node in root['children'].items(): | for key, node in root['children'].items(): | ||||
| pcurrent.append(key) | pcurrent.append(key) | ||||
| if node['isEndOfWord']: | |||||
| if node['isEndOfWord']: | |||||
| # print(node['count']) | # print(node['count']) | ||||
| if len(p1) == len(pcurrent): | if len(p1) == len(pcurrent): | ||||
| kernel[0] += 1 | kernel[0] += 1 | ||||
| @@ -730,8 +730,8 @@ def traverseTrie2u(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | |||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| if pcurrent != []: | if pcurrent != []: | ||||
| del pcurrent[-1] | del pcurrent[-1] | ||||
| #def computePathKernel(p1, p2, vk_dict, ek_dict): | #def computePathKernel(p1, p2, vk_dict, ek_dict): | ||||
| # kernel = 0 | # kernel = 0 | ||||
| # if vk_dict: | # if vk_dict: | ||||
| @@ -771,7 +771,7 @@ def traverseTrie2u(root, p1, kernel, vk_dict, ek_dict, pcurrent=[]): | |||||
| # else: | # else: | ||||
| # if len(p1) == len(p2): | # if len(p1) == len(p2): | ||||
| # kernel += 1 | # kernel += 1 | ||||
| # | |||||
| # | |||||
| # return kernel | # return kernel | ||||
| @@ -804,7 +804,7 @@ def get_shortest_paths(G, weight, directed): | |||||
| # each edge walk is counted twice, starting from both its extreme nodes. | # each edge walk is counted twice, starting from both its extreme nodes. | ||||
| if not directed: | if not directed: | ||||
| sp += [sptemp[::-1] for sptemp in spltemp] | sp += [sptemp[::-1] for sptemp in spltemp] | ||||
| # add single nodes as length 0 paths. | # add single nodes as length 0 paths. | ||||
| sp += [[n] for n in G.nodes()] | sp += [[n] for n in G.nodes()] | ||||
| return sp | return sp | ||||
| @@ -849,7 +849,7 @@ def get_sps_as_trie(G, weight, directed): | |||||
| # each edge walk is counted twice, starting from both its extreme nodes. | # each edge walk is counted twice, starting from both its extreme nodes. | ||||
| if not directed: | if not directed: | ||||
| sptrie.insertWord(sp[::-1]) | sptrie.insertWord(sp[::-1]) | ||||
| # add single nodes as length 0 paths. | # add single nodes as length 0 paths. | ||||
| for n in G.nodes(): | for n in G.nodes(): | ||||
| sptrie.insertWord([n]) | sptrie.insertWord([n]) | ||||
+ 90
- 52
gklearn/tests/test_graph_kernels.py
View File
| @@ -3,13 +3,14 @@ | |||||
| import pytest | import pytest | ||||
| import multiprocessing | import multiprocessing | ||||
| import numpy as np | |||||
| def chooseDataset(ds_name): | def chooseDataset(ds_name): | ||||
| """Choose dataset according to name. | """Choose dataset according to name. | ||||
| """ | """ | ||||
| from gklearn.utils import Dataset | from gklearn.utils import Dataset | ||||
| dataset = Dataset() | dataset = Dataset() | ||||
| # no node labels (and no edge labels). | # no node labels (and no edge labels). | ||||
| @@ -18,6 +19,7 @@ def chooseDataset(ds_name): | |||||
| dataset.trim_dataset(edge_required=False) | dataset.trim_dataset(edge_required=False) | ||||
| irrelevant_labels = {'node_attrs': ['x', 'y', 'z'], 'edge_labels': ['bond_stereo']} | irrelevant_labels = {'node_attrs': ['x', 'y', 'z'], 'edge_labels': ['bond_stereo']} | ||||
| dataset.remove_labels(**irrelevant_labels) | dataset.remove_labels(**irrelevant_labels) | ||||
| dataset.cut_graphs(range(1, 10)) | |||||
| # node symbolic labels. | # node symbolic labels. | ||||
| elif ds_name == 'Acyclic': | elif ds_name == 'Acyclic': | ||||
| dataset.load_predefined_dataset(ds_name) | dataset.load_predefined_dataset(ds_name) | ||||
| @@ -46,9 +48,9 @@ def chooseDataset(ds_name): | |||||
| elif ds_name == 'Cuneiform': | elif ds_name == 'Cuneiform': | ||||
| dataset.load_predefined_dataset(ds_name) | dataset.load_predefined_dataset(ds_name) | ||||
| dataset.trim_dataset(edge_required=True) | dataset.trim_dataset(edge_required=True) | ||||
| dataset.cut_graphs(range(0, 3)) | dataset.cut_graphs(range(0, 3)) | ||||
| return dataset | return dataset | ||||
| @@ -57,7 +59,7 @@ def test_list_graph_kernels(): | |||||
| """ | """ | ||||
| from gklearn.kernels import GRAPH_KERNELS, list_of_graph_kernels | from gklearn.kernels import GRAPH_KERNELS, list_of_graph_kernels | ||||
| assert list_of_graph_kernels() == [i for i in GRAPH_KERNELS] | assert list_of_graph_kernels() == [i for i in GRAPH_KERNELS] | ||||
| @pytest.mark.parametrize('ds_name', ['Alkane', 'AIDS']) | @pytest.mark.parametrize('ds_name', ['Alkane', 'AIDS']) | ||||
| @@ -68,10 +70,10 @@ def test_CommonWalk(ds_name, parallel, weight, compute_method): | |||||
| """ | """ | ||||
| from gklearn.kernels import CommonWalk | from gklearn.kernels import CommonWalk | ||||
| import networkx as nx | import networkx as nx | ||||
| dataset = chooseDataset(ds_name) | dataset = chooseDataset(ds_name) | ||||
| dataset.load_graphs([g for g in dataset.graphs if nx.number_of_nodes(g) > 1]) | dataset.load_graphs([g for g in dataset.graphs if nx.number_of_nodes(g) > 1]) | ||||
| try: | try: | ||||
| graph_kernel = CommonWalk(node_labels=dataset.node_labels, | graph_kernel = CommonWalk(node_labels=dataset.node_labels, | ||||
| edge_labels=dataset.edge_labels, | edge_labels=dataset.edge_labels, | ||||
| @@ -87,8 +89,8 @@ def test_CommonWalk(ds_name, parallel, weight, compute_method): | |||||
| except Exception as exception: | except Exception as exception: | ||||
| assert False, exception | assert False, exception | ||||
| @pytest.mark.parametrize('ds_name', ['Alkane', 'AIDS']) | @pytest.mark.parametrize('ds_name', ['Alkane', 'AIDS']) | ||||
| @pytest.mark.parametrize('remove_totters', [False]) #[True, False]) | @pytest.mark.parametrize('remove_totters', [False]) #[True, False]) | ||||
| @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | ||||
| @@ -96,9 +98,9 @@ def test_Marginalized(ds_name, parallel, remove_totters): | |||||
| """Test marginalized kernel. | """Test marginalized kernel. | ||||
| """ | """ | ||||
| from gklearn.kernels import Marginalized | from gklearn.kernels import Marginalized | ||||
| dataset = chooseDataset(ds_name) | dataset = chooseDataset(ds_name) | ||||
| try: | try: | ||||
| graph_kernel = Marginalized(node_labels=dataset.node_labels, | graph_kernel = Marginalized(node_labels=dataset.node_labels, | ||||
| edge_labels=dataset.edge_labels, | edge_labels=dataset.edge_labels, | ||||
| @@ -115,15 +117,15 @@ def test_Marginalized(ds_name, parallel, remove_totters): | |||||
| except Exception as exception: | except Exception as exception: | ||||
| assert False, exception | assert False, exception | ||||
| @pytest.mark.parametrize('ds_name', ['Acyclic']) | @pytest.mark.parametrize('ds_name', ['Acyclic']) | ||||
| @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | ||||
| def test_SylvesterEquation(ds_name, parallel): | def test_SylvesterEquation(ds_name, parallel): | ||||
| """Test sylvester equation kernel. | """Test sylvester equation kernel. | ||||
| """ | """ | ||||
| from gklearn.kernels import SylvesterEquation | from gklearn.kernels import SylvesterEquation | ||||
| dataset = chooseDataset(ds_name) | dataset = chooseDataset(ds_name) | ||||
| try: | try: | ||||
| @@ -139,11 +141,11 @@ def test_SylvesterEquation(ds_name, parallel): | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | ||||
| kernel, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1], | kernel, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1], | ||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | ||||
| except Exception as exception: | except Exception as exception: | ||||
| assert False, exception | assert False, exception | ||||
| @pytest.mark.parametrize('ds_name', ['Acyclic', 'AIDS']) | @pytest.mark.parametrize('ds_name', ['Acyclic', 'AIDS']) | ||||
| @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | ||||
| def test_ConjugateGradient(ds_name, parallel): | def test_ConjugateGradient(ds_name, parallel): | ||||
| @@ -152,9 +154,9 @@ def test_ConjugateGradient(ds_name, parallel): | |||||
| from gklearn.kernels import ConjugateGradient | from gklearn.kernels import ConjugateGradient | ||||
| from gklearn.utils.kernels import deltakernel, gaussiankernel, kernelproduct | from gklearn.utils.kernels import deltakernel, gaussiankernel, kernelproduct | ||||
| import functools | import functools | ||||
| dataset = chooseDataset(ds_name) | dataset = chooseDataset(ds_name) | ||||
| mixkernel = functools.partial(kernelproduct, deltakernel, gaussiankernel) | mixkernel = functools.partial(kernelproduct, deltakernel, gaussiankernel) | ||||
| sub_kernels = {'symb': deltakernel, 'nsymb': gaussiankernel, 'mix': mixkernel} | sub_kernels = {'symb': deltakernel, 'nsymb': gaussiankernel, 'mix': mixkernel} | ||||
| @@ -177,11 +179,11 @@ def test_ConjugateGradient(ds_name, parallel): | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | ||||
| kernel, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1], | kernel, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1], | ||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | ||||
| except Exception as exception: | except Exception as exception: | ||||
| assert False, exception | assert False, exception | ||||
| @pytest.mark.parametrize('ds_name', ['Acyclic', 'AIDS']) | @pytest.mark.parametrize('ds_name', ['Acyclic', 'AIDS']) | ||||
| @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | ||||
| def test_FixedPoint(ds_name, parallel): | def test_FixedPoint(ds_name, parallel): | ||||
| @@ -190,9 +192,9 @@ def test_FixedPoint(ds_name, parallel): | |||||
| from gklearn.kernels import FixedPoint | from gklearn.kernels import FixedPoint | ||||
| from gklearn.utils.kernels import deltakernel, gaussiankernel, kernelproduct | from gklearn.utils.kernels import deltakernel, gaussiankernel, kernelproduct | ||||
| import functools | import functools | ||||
| dataset = chooseDataset(ds_name) | dataset = chooseDataset(ds_name) | ||||
| mixkernel = functools.partial(kernelproduct, deltakernel, gaussiankernel) | mixkernel = functools.partial(kernelproduct, deltakernel, gaussiankernel) | ||||
| sub_kernels = {'symb': deltakernel, 'nsymb': gaussiankernel, 'mix': mixkernel} | sub_kernels = {'symb': deltakernel, 'nsymb': gaussiankernel, 'mix': mixkernel} | ||||
| @@ -215,11 +217,11 @@ def test_FixedPoint(ds_name, parallel): | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | ||||
| kernel, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1], | kernel, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1], | ||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | ||||
| except Exception as exception: | except Exception as exception: | ||||
| assert False, exception | assert False, exception | ||||
| @pytest.mark.parametrize('ds_name', ['Acyclic']) | @pytest.mark.parametrize('ds_name', ['Acyclic']) | ||||
| @pytest.mark.parametrize('sub_kernel', ['exp', 'geo']) | @pytest.mark.parametrize('sub_kernel', ['exp', 'geo']) | ||||
| @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | ||||
| @@ -227,7 +229,7 @@ def test_SpectralDecomposition(ds_name, sub_kernel, parallel): | |||||
| """Test spectral decomposition kernel. | """Test spectral decomposition kernel. | ||||
| """ | """ | ||||
| from gklearn.kernels import SpectralDecomposition | from gklearn.kernels import SpectralDecomposition | ||||
| dataset = chooseDataset(ds_name) | dataset = chooseDataset(ds_name) | ||||
| try: | try: | ||||
| @@ -244,11 +246,11 @@ def test_SpectralDecomposition(ds_name, sub_kernel, parallel): | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | ||||
| kernel, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1], | kernel, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1], | ||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | ||||
| except Exception as exception: | except Exception as exception: | ||||
| assert False, exception | assert False, exception | ||||
| # @pytest.mark.parametrize( | # @pytest.mark.parametrize( | ||||
| # 'compute_method,ds_name,sub_kernel', | # 'compute_method,ds_name,sub_kernel', | ||||
| # [ | # [ | ||||
| @@ -268,7 +270,7 @@ def test_SpectralDecomposition(ds_name, sub_kernel, parallel): | |||||
| # from gklearn.kernels import RandomWalk | # from gklearn.kernels import RandomWalk | ||||
| # from gklearn.utils.kernels import deltakernel, gaussiankernel, kernelproduct | # from gklearn.utils.kernels import deltakernel, gaussiankernel, kernelproduct | ||||
| # import functools | # import functools | ||||
| # | |||||
| # | |||||
| # dataset = chooseDataset(ds_name) | # dataset = chooseDataset(ds_name) | ||||
| # mixkernel = functools.partial(kernelproduct, deltakernel, gaussiankernel) | # mixkernel = functools.partial(kernelproduct, deltakernel, gaussiankernel) | ||||
| @@ -297,7 +299,7 @@ def test_SpectralDecomposition(ds_name, sub_kernel, parallel): | |||||
| # except Exception as exception: | # except Exception as exception: | ||||
| # assert False, exception | # assert False, exception | ||||
| @pytest.mark.parametrize('ds_name', ['Alkane', 'Acyclic', 'Letter-med', 'AIDS', 'Fingerprint']) | @pytest.mark.parametrize('ds_name', ['Alkane', 'Acyclic', 'Letter-med', 'AIDS', 'Fingerprint']) | ||||
| @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | ||||
| def test_ShortestPath(ds_name, parallel): | def test_ShortestPath(ds_name, parallel): | ||||
| @@ -306,17 +308,30 @@ def test_ShortestPath(ds_name, parallel): | |||||
| from gklearn.kernels import ShortestPath | from gklearn.kernels import ShortestPath | ||||
| from gklearn.utils.kernels import deltakernel, gaussiankernel, kernelproduct | from gklearn.utils.kernels import deltakernel, gaussiankernel, kernelproduct | ||||
| import functools | import functools | ||||
| dataset = chooseDataset(ds_name) | dataset = chooseDataset(ds_name) | ||||
| mixkernel = functools.partial(kernelproduct, deltakernel, gaussiankernel) | mixkernel = functools.partial(kernelproduct, deltakernel, gaussiankernel) | ||||
| sub_kernels = {'symb': deltakernel, 'nsymb': gaussiankernel, 'mix': mixkernel} | sub_kernels = {'symb': deltakernel, 'nsymb': gaussiankernel, 'mix': mixkernel} | ||||
| try: | try: | ||||
| graph_kernel = ShortestPath(node_labels=dataset.node_labels, | graph_kernel = ShortestPath(node_labels=dataset.node_labels, | ||||
| node_attrs=dataset.node_attrs, | node_attrs=dataset.node_attrs, | ||||
| ds_infos=dataset.get_dataset_infos(keys=['directed']), | ds_infos=dataset.get_dataset_infos(keys=['directed']), | ||||
| fcsp=True, | |||||
| node_kernels=sub_kernels) | node_kernels=sub_kernels) | ||||
| gram_matrix, run_time = graph_kernel.compute(dataset.graphs, | |||||
| gram_matrix1, run_time = graph_kernel.compute(dataset.graphs, | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | |||||
| kernel_list, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1:], | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | |||||
| kernel, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1], | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | |||||
| graph_kernel = ShortestPath(node_labels=dataset.node_labels, | |||||
| node_attrs=dataset.node_attrs, | |||||
| ds_infos=dataset.get_dataset_infos(keys=['directed']), | |||||
| fcsp=False, | |||||
| node_kernels=sub_kernels) | |||||
| gram_matrix2, run_time = graph_kernel.compute(dataset.graphs, | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | ||||
| kernel_list, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1:], | kernel_list, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1:], | ||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | ||||
| @@ -326,6 +341,8 @@ def test_ShortestPath(ds_name, parallel): | |||||
| except Exception as exception: | except Exception as exception: | ||||
| assert False, exception | assert False, exception | ||||
| assert np.array_equal(gram_matrix1, gram_matrix2) | |||||
| #@pytest.mark.parametrize('ds_name', ['Alkane', 'Acyclic', 'Letter-med', 'AIDS', 'Fingerprint']) | #@pytest.mark.parametrize('ds_name', ['Alkane', 'Acyclic', 'Letter-med', 'AIDS', 'Fingerprint']) | ||||
| @pytest.mark.parametrize('ds_name', ['Alkane', 'Acyclic', 'Letter-med', 'AIDS', 'Fingerprint', 'Fingerprint_edge', 'Cuneiform']) | @pytest.mark.parametrize('ds_name', ['Alkane', 'Acyclic', 'Letter-med', 'AIDS', 'Fingerprint', 'Fingerprint_edge', 'Cuneiform']) | ||||
| @@ -336,29 +353,47 @@ def test_StructuralSP(ds_name, parallel): | |||||
| from gklearn.kernels import StructuralSP | from gklearn.kernels import StructuralSP | ||||
| from gklearn.utils.kernels import deltakernel, gaussiankernel, kernelproduct | from gklearn.utils.kernels import deltakernel, gaussiankernel, kernelproduct | ||||
| import functools | import functools | ||||
| dataset = chooseDataset(ds_name) | dataset = chooseDataset(ds_name) | ||||
| mixkernel = functools.partial(kernelproduct, deltakernel, gaussiankernel) | mixkernel = functools.partial(kernelproduct, deltakernel, gaussiankernel) | ||||
| sub_kernels = {'symb': deltakernel, 'nsymb': gaussiankernel, 'mix': mixkernel} | sub_kernels = {'symb': deltakernel, 'nsymb': gaussiankernel, 'mix': mixkernel} | ||||
| try: | try: | ||||
| graph_kernel = StructuralSP(node_labels=dataset.node_labels, | graph_kernel = StructuralSP(node_labels=dataset.node_labels, | ||||
| edge_labels=dataset.edge_labels, | |||||
| edge_labels=dataset.edge_labels, | |||||
| node_attrs=dataset.node_attrs, | node_attrs=dataset.node_attrs, | ||||
| edge_attrs=dataset.edge_attrs, | edge_attrs=dataset.edge_attrs, | ||||
| ds_infos=dataset.get_dataset_infos(keys=['directed']), | ds_infos=dataset.get_dataset_infos(keys=['directed']), | ||||
| fcsp=True, | |||||
| node_kernels=sub_kernels, | node_kernels=sub_kernels, | ||||
| edge_kernels=sub_kernels) | edge_kernels=sub_kernels) | ||||
| gram_matrix, run_time = graph_kernel.compute(dataset.graphs, | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | |||||
| gram_matrix1, run_time = graph_kernel.compute(dataset.graphs, | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True, normalize=False) | |||||
| kernel_list, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1:], | kernel_list, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1:], | ||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | |||||
| kernel, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1], | kernel, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1], | ||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | |||||
| graph_kernel = StructuralSP(node_labels=dataset.node_labels, | |||||
| edge_labels=dataset.edge_labels, | |||||
| node_attrs=dataset.node_attrs, | |||||
| edge_attrs=dataset.edge_attrs, | |||||
| ds_infos=dataset.get_dataset_infos(keys=['directed']), | |||||
| fcsp=False, | |||||
| node_kernels=sub_kernels, | |||||
| edge_kernels=sub_kernels) | |||||
| gram_matrix2, run_time = graph_kernel.compute(dataset.graphs, | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True, normalize=False) | |||||
| kernel_list, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1:], | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | |||||
| kernel, run_time = graph_kernel.compute(dataset.graphs[0], dataset.graphs[1], | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | |||||
| except Exception as exception: | except Exception as exception: | ||||
| assert False, exception | assert False, exception | ||||
| assert np.array_equal(gram_matrix1, gram_matrix2) | |||||
| @pytest.mark.parametrize('ds_name', ['Alkane', 'AIDS']) | @pytest.mark.parametrize('ds_name', ['Alkane', 'AIDS']) | ||||
| @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | ||||
| @@ -369,9 +404,9 @@ def test_PathUpToH(ds_name, parallel, k_func, compute_method): | |||||
| """Test path kernel up to length $h$. | """Test path kernel up to length $h$. | ||||
| """ | """ | ||||
| from gklearn.kernels import PathUpToH | from gklearn.kernels import PathUpToH | ||||
| dataset = chooseDataset(ds_name) | dataset = chooseDataset(ds_name) | ||||
| try: | try: | ||||
| graph_kernel = PathUpToH(node_labels=dataset.node_labels, | graph_kernel = PathUpToH(node_labels=dataset.node_labels, | ||||
| edge_labels=dataset.edge_labels, | edge_labels=dataset.edge_labels, | ||||
| @@ -385,8 +420,8 @@ def test_PathUpToH(ds_name, parallel, k_func, compute_method): | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | ||||
| except Exception as exception: | except Exception as exception: | ||||
| assert False, exception | assert False, exception | ||||
| @pytest.mark.parametrize('ds_name', ['Alkane', 'AIDS']) | @pytest.mark.parametrize('ds_name', ['Alkane', 'AIDS']) | ||||
| @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | @pytest.mark.parametrize('parallel', ['imap_unordered', None]) | ||||
| def test_Treelet(ds_name, parallel): | def test_Treelet(ds_name, parallel): | ||||
| @@ -395,10 +430,10 @@ def test_Treelet(ds_name, parallel): | |||||
| from gklearn.kernels import Treelet | from gklearn.kernels import Treelet | ||||
| from gklearn.utils.kernels import polynomialkernel | from gklearn.utils.kernels import polynomialkernel | ||||
| import functools | import functools | ||||
| dataset = chooseDataset(ds_name) | dataset = chooseDataset(ds_name) | ||||
| pkernel = functools.partial(polynomialkernel, d=2, c=1e5) | |||||
| pkernel = functools.partial(polynomialkernel, d=2, c=1e5) | |||||
| try: | try: | ||||
| graph_kernel = Treelet(node_labels=dataset.node_labels, | graph_kernel = Treelet(node_labels=dataset.node_labels, | ||||
| edge_labels=dataset.edge_labels, | edge_labels=dataset.edge_labels, | ||||
| @@ -412,8 +447,8 @@ def test_Treelet(ds_name, parallel): | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | ||||
| except Exception as exception: | except Exception as exception: | ||||
| assert False, exception | assert False, exception | ||||
| @pytest.mark.parametrize('ds_name', ['Acyclic']) | @pytest.mark.parametrize('ds_name', ['Acyclic']) | ||||
| #@pytest.mark.parametrize('base_kernel', ['subtree', 'sp', 'edge']) | #@pytest.mark.parametrize('base_kernel', ['subtree', 'sp', 'edge']) | ||||
| # @pytest.mark.parametrize('base_kernel', ['subtree']) | # @pytest.mark.parametrize('base_kernel', ['subtree']) | ||||
| @@ -422,7 +457,7 @@ def test_WLSubtree(ds_name, parallel): | |||||
| """Test Weisfeiler-Lehman subtree kernel. | """Test Weisfeiler-Lehman subtree kernel. | ||||
| """ | """ | ||||
| from gklearn.kernels import WLSubtree | from gklearn.kernels import WLSubtree | ||||
| dataset = chooseDataset(ds_name) | dataset = chooseDataset(ds_name) | ||||
| try: | try: | ||||
| @@ -438,12 +473,15 @@ def test_WLSubtree(ds_name, parallel): | |||||
| parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | parallel=parallel, n_jobs=multiprocessing.cpu_count(), verbose=True) | ||||
| except Exception as exception: | except Exception as exception: | ||||
| assert False, exception | assert False, exception | ||||
| if __name__ == "__main__": | if __name__ == "__main__": | ||||
| test_list_graph_kernels() | test_list_graph_kernels() | ||||
| # test_spkernel('Alkane', 'imap_unordered') | # test_spkernel('Alkane', 'imap_unordered') | ||||
| # test_ShortestPath('Alkane', 'imap_unordered') | |||||
| # test_StructuralSP('Fingerprint_edge', 'imap_unordered') | # test_StructuralSP('Fingerprint_edge', 'imap_unordered') | ||||
| # test_StructuralSP('Alkane', None) | |||||
| # test_StructuralSP('Cuneiform', None) | |||||
| # test_WLSubtree('Acyclic', 'imap_unordered') | # test_WLSubtree('Acyclic', 'imap_unordered') | ||||
| # test_RandomWalk('Acyclic', 'sylvester', None, 'imap_unordered') | # test_RandomWalk('Acyclic', 'sylvester', None, 'imap_unordered') | ||||
| # test_RandomWalk('Acyclic', 'conjugate', None, 'imap_unordered') | # test_RandomWalk('Acyclic', 'conjugate', None, 'imap_unordered') | ||||