Add fcsp experiments.

gklearn/experiments/thesis/graph_kernels/fcsp/compare_fcsp.py

@@ -0,0 +1,82 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Created on Wed Dec  2 17:41:54 2020
+
+@author: ljia
+
+This script compares the results with and without FCSP.
+"""
+from gklearn.dataset import Dataset
+from gklearn.utils import get_graph_kernel_by_name
+from gklearn.utils.kernels import deltakernel, gaussiankernel, kernelproduct
+import functools
+import os
+import pickle
+import sys
+
+
+def run_all(fcsp):
+	save_dir = 'outputs/' + ('fscp' if fcsp == True else 'naive') + '/'
+	os.makedirs(save_dir, exist_ok=True)
+
+	from sklearn.model_selection import ParameterGrid
+
+	Dataset_List = ['Alkane_unlabeled', 'Alkane', 'Acyclic', 'MAO_lite', 'MAO',
+				    'PAH', 'MUTAG', 'Monoterpenoids', 'Letter-high',
+					'Letter-med', 'Letter-low',
+					'ENZYMES', 'AIDS_lite', 'AIDS', 'NCI1', 'NCI109', 'DD']
+
+	Kernel_List = ['ShortestPath', 'StructuralSP']
+
+	work_grid = ParameterGrid({'kernel': Kernel_List[0:], 'dataset': Dataset_List[2:3]})
+
+	for work in list(work_grid):
+
+		save_file_suffix = '.' + work['kernel'] + '.' + work['dataset']
+		file_name = os.path.join(save_dir, 'run_time' + save_file_suffix + '.pkl')
+		if not os.path.isfile(file_name):
+			print()
+			print((work['kernel'], work['dataset']))
+
+			gram_matrix, run_time = run_work(work['kernel'], work['dataset'], fcsp)
+
+			save_file_suffix = '.' + work['kernel'] + work['dataset']
+
+			with open(file_name, 'wb') as f:
+				pickle.dump(run_time, f)
+
+
+def run_work(kernel_name, ds_name, fcsp):
+	dataset = Dataset(ds_name, verbose=True)
+
+	mixkernel = functools.partial(kernelproduct, deltakernel, gaussiankernel)
+	node_kernels = {'symb': deltakernel, 'nsymb': gaussiankernel, 'mix': mixkernel}
+	edge_kernels = {'symb': deltakernel, 'nsymb': gaussiankernel, 'mix': mixkernel}
+
+	graph_kernel = get_graph_kernel_by_name(name=kernel_name,
+							  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=fcsp,
+							  compute_method='naive',
+							  node_kernels=node_kernels,
+							  edge_kernels=edge_kernels,
+							  )
+	gram_matrix, run_time = graph_kernel.compute(dataset.graphs,
+											  parallel=None,
+											  normalize=True,
+											  verbose=2
+											  )
+	return gram_matrix, run_time
+
+
+if __name__ == '__main__':
+	if len(sys.argv) > 1:
+		fcsp = True if sys.argv[1] == 'True' else False
+	else:
+		fcsp = True
+	run_all(fcsp)
+

gklearn/experiments/thesis/graph_kernels/fcsp/run_jobs_compare_fcsp.py

@@ -0,0 +1,52 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Created on Mon Dec 14 11:49:43 2020
+
+@author: ljia
+"""
+
+import os
+import re
+
+
+def get_job_script(param):
+	script = r"""
+#!/bin/bash
+
+#SBATCH --exclusive
+#SBATCH --job-name="fcsp.""" + param + r""""
+#SBATCH --partition=long
+#SBATCH --mail-type=ALL
+#SBATCH --mail-user=jajupmochi@gmail.com
+#SBATCH --output="outputs/output_fcsp.""" + param + r""".txt"
+#SBATCH --error="errors/error_fcsp.""" + param + r""".txt"
+#
+#SBATCH --ntasks=1
+#SBATCH --nodes=1
+#SBATCH --cpus-per-task=1
+#SBATCH --time=100:00:00
+#SBATCH --mem-per-cpu=4000
+
+srun hostname
+srun cd /home/2019015/ljia02/graphkit-learn/gklearn/experiments/thesis/graph_kernels/fcsp
+srun python3 compare_fcsp.py """ + param
+	script = script.strip()
+	script = re.sub('\n\t+', '\n', script)
+	script = re.sub('\n +', '\n', script)
+
+	return script
+
+
+if __name__ == '__main__':
+	os.makedirs('outputs/', exist_ok=True)
+	os.makedirs('errors/', exist_ok=True)
+
+	param_list = ['True', 'False']
+	for param in param_list[:]:
+		job_script = get_job_script(param)
+		command = 'sbatch <<EOF\n' + job_script + '\nEOF'
+# 			print(command)
+		os.system(command)
+	# 		os.popen(command)
+	# 		output = stream.readlines()