OpenI
/
graphkit-learn
Not watched
1
0
Fork 0
Code
Releases 2 Wiki Activity
Issues 0 Pull Requests 0 Datasets Model Cloudbrain
You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
80 Commits
4 Branches
345 MB
156 Download
Tree: 6b4738df30
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '6b4738df30'
${ noResults }
graphkit-learn/notebooks/run_structuralspkernel.ipynb

run_structuralspkernel.ipynb 15 kB
Raw Normal View History

* use global variable to tackle big read-only variables when they are used in subprocesses. * rewrite the implementation of the marginalized kernel. * implement four computing methods of the generalized random walk kernel. * in the path kernel up to length h, use trie to save all paths, saving tremendous memories; use the Deep-first search to get paths from graphs. * in model_selection_for_precomputed_kernel method, complete the part to do cross validation when Gram matrices are read from file. * in get_dataset_attributes methods, correct three sub-methods about getting node degrees, add sub-methods to get fill factors of graphs. * change default chunksize of pool.imap_unordered parallel method to 100. * remove try... except blocks in case they hidden bugs.
6 years ago
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308 
  1. {

  2.  "cells": [

  3.   {

  4.    "cell_type": "code",

  5.    "execution_count": 1,

  6.    "metadata": {

  7.     "scrolled": false

  8.    },

  9.    "outputs": [

  10.     {

  11.      "name": "stdout",

  12.      "output_type": "stream",

  13.      "text": [

  14.       "\n",

  15.       "MAO\n",

  16.       "\n",

  17.       "--- This is a classification problem ---\n",

  18.       "\n",

  19.       "\n",

  20.       "1. Loading dataset from file...\n",

  21.       "\n",

  22.       "2. Calculating gram matrices. This could take a while...\n",

  23.       "\n",

  24.       " None edge weight specified. Set all weight to 1.\n",

  25.       "\n",

  26.       "getting shortest paths: 68it [00:00, 629.46it/s]\n",

  27.       "calculating kernels: 2346it [00:22, 102.31it/s]\n",

  28.       "\n",

  29.       " --- shortest path kernel matrix of size 68 built in 23.390946626663208 seconds ---\n",

  30.       "\n",

  31.       "the gram matrix with parameters {'edge_kernels': {'symb': <function deltakernel at 0x7f90ea71dae8>, 'nsymb': <function gaussiankernel at 0x7f90ea71d620>, 'mix': functools.partial(<function kernelproduct at 0x7f90ea71d6a8>, <function deltakernel at 0x7f90ea71dae8>, <function gaussiankernel at 0x7f90ea71d620>)}, 'node_kernels': {'symb': <function deltakernel at 0x7f90ea71dae8>, 'nsymb': <function gaussiankernel at 0x7f90ea71d620>, 'mix': functools.partial(<function kernelproduct at 0x7f90ea71d6a8>, <function deltakernel at 0x7f90ea71dae8>, <function gaussiankernel at 0x7f90ea71d620>)}, 'n_jobs': 8} is: \n",

  32.       "\n",

  33.       "1 gram matrices are calculated, 0 of which are ignored.\n",

  34.       "\n",

  35.       "3. Fitting and predicting using nested cross validation. This could really take a while...\n",

  36.       "cross validation:   0%|          | 0/30 [00:00<?, ?it/s]0 0\n",

  37.       "params_in:  {'C': 1e-10}\n",

  38.       "0 1\n",

  39.       "params_in:  {'C': 3.1622776601683795e-10}\n",

  40.       "0 2\n",

  41.       "params_in:  {'C': 1e-09}\n",

  42.       "0 3\n",

  43.       "params_in:  {'C': 3.1622776601683795e-09}\n",

  44.       "0 4\n",

  45.       "params_in:  {'C': 1e-08}\n",

  46.       "0 5\n",

  47.       "params_in:  {'C': 3.162277660168379e-08}\n",

  48.       "0 6\n",

  49.       "params_in:  {'C': 1e-07}\n",

  50.       "0 7\n",

  51.       "params_in:  {'C': 3.162277660168379e-07}\n",

  52.       "0 8\n",

  53.       "params_in:  {'C': 1e-06}\n",

  54.       "0 9\n",

  55.       "params_in:  {'C': 3.162277660168379e-06}\n",

  56.       "0 10\n",

  57.       "params_in:  {'C': 1e-05}\n",

  58.       "0 11\n",

  59.       "params_in:  {'C': 3.1622776601683795e-05}\n",

  60.       "0 12\n",

  61.       "params_in:  {'C': 0.0001}\n",

  62.       "0 13\n",

  63.       "params_in:  {'C': 0.00031622776601683794}\n",

  64.       "0 14\n",

  65.       "params_in:  {'C': 0.001}\n",

  66.       "0 15\n",

  67.       "params_in:  {'C': 0.0031622776601683794}\n",

  68.       "0 16\n",

  69.       "params_in:  {'C': 0.01}\n",

  70.       "0 17\n",

  71.       "params_in:  {'C': 0.03162277660168379}\n",

  72.       "0 18\n",

  73.       "params_in:  {'C': 0.1}\n",

  74.       "0 19\n",

  75.       "params_in:  {'C': 0.31622776601683794}\n",

  76.       "0 20\n",

  77.       "params_in:  {'C': 1.0}\n",

  78.       "0 21\n",

  79.       "params_in:  {'C': 3.1622776601683795}\n",

  80.       "0 22\n",

  81.       "params_in:  {'C': 10.0}\n",

  82.       "0 23\n",

  83.       "params_in:  {'C': 31.622776601683793}\n",

  84.       "0 24\n",

  85.       "params_in:  {'C': 100.0}\n",

  86.       "0 25\n",

  87.       "params_in:  {'C': 316.22776601683796}\n",

  88.       "0 26\n",

  89.       "params_in:  {'C': 1000.0}\n",

  90.       "0 27\n",

  91.       "params_in:  {'C': 3162.2776601683795}\n",

  92.       "0 28\n",

  93.       "params_in:  {'C': 10000.0}\n",

  94.       "0 29\n",

  95.       "params_in:  {'C': 31622.776601683792}\n",

  96.       "0 30\n",

  97.       "params_in:  {'C': 100000.0}\n",

  98.       "0 31\n",

  99.       "params_in:  {'C': 316227.7660168379}\n",

  100.       "0 32\n",

  101.       "params_in:  {'C': 1000000.0}\n",

  102.       "0 33\n",

  103.       "params_in:  {'C': 3162277.6601683795}\n",

  104.       "0 34\n",

  105.       "params_in:  {'C': 10000000.0}\n",

  106.       "0 35\n",

  107.       "params_in:  {'C': 31622776.60168379}\n",

  108.       "0 36\n",

  109.       "params_in:  {'C': 100000000.0}\n",

  110.       "0 37\n",

  111.       "params_in:  {'C': 316227766.01683795}\n",

  112.       "0 38\n",

  113.       "params_in:  {'C': 1000000000.0}\n",

  114.       "0 39\n",

  115.       "params_in:  {'C': 3162277660.1683793}\n",

  116.       "0 40\n",

  117.       "params_in:  {'C': 10000000000.0}\n",

  118.       "val_pref:  [[0.59285714 0.59285714 0.59285714 0.59285714 0.59285714 0.59285714\n",

  119.       "  0.59285714 0.59285714 0.59285714 0.59285714 0.59285714 0.59285714\n",

  120.       "  0.59285714 0.59285714 0.59285714 0.59285714 0.59285714 0.59285714\n",

  121.       "  0.59285714 0.59285714 0.55952381 0.71666667 0.81666667 0.81666667\n",

  122.       "  0.83571429 0.86666667 0.9        0.9        0.9        0.9\n",

  123.       "  0.9        0.9        0.9        0.9        0.9        0.9\n",

  124.       "  0.9        0.9        0.9        0.9        0.9       ]]\n",

  125.       "test_pref:  [[0.28571429 0.28571429 0.28571429 0.28571429 0.28571429 0.28571429\n",

  126.       "  0.28571429 0.28571429 0.28571429 0.28571429 0.28571429 0.28571429\n",

  127.       "  0.28571429 0.28571429 0.28571429 0.28571429 0.28571429 0.28571429\n",

  128.       "  0.28571429 0.28571429 0.61428571 0.84285714 0.84285714 0.85714286\n",

  129.       "  0.85714286 0.85714286 0.85714286 0.85714286 0.85714286 0.85714286\n",

  130.       "  0.85714286 0.85714286 0.85714286 0.85714286 0.85714286 0.85714286\n",

  131.       "  0.85714286 0.85714286 0.85714286 0.85714286 0.85714286]]\n",

  132.       "cross validation: 100%|██████████| 30/30 [00:11<00:00,  2.75it/s]\n",

  133.       "\n",

  134.       "\n",

  135.       "4. Getting final performance...\n",

  136.       "val_pref:  [0.59285714 0.59285714 0.59285714 0.59285714 0.59285714 0.59285714\n",

  137.       " 0.59285714 0.59285714 0.59285714 0.59285714 0.59285714 0.59285714\n",

  138.       " 0.59285714 0.59285714 0.59285714 0.59285714 0.59285714 0.59285714\n",

  139.       " 0.59285714 0.59285714 0.55952381 0.71666667 0.81666667 0.81666667\n",

  140.       " 0.83571429 0.86666667 0.9        0.9        0.9        0.9\n",

  141.       " 0.9        0.9        0.9        0.9        0.9        0.9\n",

  142.       " 0.9        0.9        0.9        0.9        0.9       ]\n",

  143.       "test_pref:  [0.28571429 0.28571429 0.28571429 0.28571429 0.28571429 0.28571429\n",

  144.       " 0.28571429 0.28571429 0.28571429 0.28571429 0.28571429 0.28571429\n",

  145.       " 0.28571429 0.28571429 0.28571429 0.28571429 0.28571429 0.28571429\n",

  146.       " 0.28571429 0.28571429 0.61428571 0.84285714 0.84285714 0.85714286\n",

  147.       " 0.85714286 0.85714286 0.85714286 0.85714286 0.85714286 0.85714286\n",

  148.       " 0.85714286 0.85714286 0.85714286 0.85714286 0.85714286 0.85714286\n",

  149.       " 0.85714286 0.85714286 0.85714286 0.85714286 0.85714286]\n",

  150.       "average_val_scores:  [[0.55301587 0.55301587 0.55301587 0.55301587 0.55301587 0.55301587\n",

  151.       "  0.55301587 0.55301587 0.55301587 0.55301587 0.55301587 0.55301587\n",

  152.       "  0.55301587 0.55301587 0.55301587 0.55301587 0.55301587 0.55301587\n",

  153.       "  0.55301587 0.55468254 0.61507937 0.71777778 0.78039683 0.80531746\n",

  154.       "  0.86198413 0.89531746 0.89420635 0.87190476 0.85761905 0.85761905\n",

  155.       "  0.85761905 0.85761905 0.85761905 0.85761905 0.85761905 0.85761905\n",

  156.       "  0.85761905 0.85761905 0.85761905 0.85761905 0.85761905]]\n",

  157.       "best_val_perf:  0.8953174603174604\n",

  158.       "\n",

  159.       "best_params_out:  [{'edge_kernels': {'symb': <function deltakernel at 0x7f90ea71dae8>, 'nsymb': <function gaussiankernel at 0x7f90ea71d620>, 'mix': functools.partial(<function kernelproduct at 0x7f90ea71d6a8>, <function deltakernel at 0x7f90ea71dae8>, <function gaussiankernel at 0x7f90ea71d620>)}, 'node_kernels': {'symb': <function deltakernel at 0x7f90ea71dae8>, 'nsymb': <function gaussiankernel at 0x7f90ea71d620>, 'mix': functools.partial(<function kernelproduct at 0x7f90ea71d6a8>, <function deltakernel at 0x7f90ea71dae8>, <function gaussiankernel at 0x7f90ea71d620>)}, 'n_jobs': 8}]\n",

  160.       "best_params_in:  [{'C': 316.22776601683796}]\n",

  161.       "\n",

  162.       "best_val_perf:  0.8953174603174604\n",

  163.       "best_val_std:  0.029090007386146643\n",

  164.       "(array([0]), array([25]))\n",

  165.       "[0]\n",

  166.       "[[0.5047619  0.5047619  0.5047619  0.5047619  0.5047619  0.5047619\n",

  167.       "  0.5047619  0.5047619  0.5047619  0.5047619  0.5047619  0.5047619\n",

  168.       "  0.5047619  0.5047619  0.5047619  0.5047619  0.5047619  0.5047619\n",

  169.       "  0.5047619  0.49761905 0.66       0.75857143 0.78857143 0.82857143\n",

  170.       "  0.85285714 0.86380952 0.84428571 0.82190476 0.81571429 0.81571429\n",

  171.       "  0.81571429 0.81571429 0.81571429 0.81571429 0.81571429 0.81571429\n",

  172.       "  0.81571429 0.81571429 0.81571429 0.81571429 0.81571429]]\n",

  173.       "final_performance:  [0.8638095238095236]\n",

  174.       "final_confidence:  [0.10509426306201483]\n",

  175.       "train_performance: [0.9857934904601572]\n",

  176.       "train_std:  [0.00730576290039335]\n",

  177.       "\n",

  178.       "time to calculate gram matrix with different hyper-params: 23.39±nans\n",

  179.       "time to calculate best gram matrix: 23.39±nans\n",

  180.       "total training time with all hyper-param choices: 34.88s\n",

  181.       "\n",

  182.       "\n"

  183.      ]

  184.     },

  185.     {

  186.      "name": "stderr",

  187.      "output_type": "stream",

  188.      "text": [

  189.       "/usr/local/lib/python3.6/dist-packages/numpy/core/_methods.py:140: RuntimeWarning: Degrees of freedom <= 0 for slice\n",

  190.       "  keepdims=keepdims)\n",

  191.       "/usr/local/lib/python3.6/dist-packages/numpy/core/_methods.py:132: RuntimeWarning: invalid value encountered in double_scalars\n",

  192.       "  ret = ret.dtype.type(ret / rcount)\n"

  193.      ]

  194.     }

  195.    ],

  196.    "source": [

  197.     "#!/usr/bin/env python3\n",

  198.     "# -*- coding: utf-8 -*-\n",

  199.     "\"\"\"\n",

  200.     "Created on Fri Sep 28 16:37:29 2018\n",

  201.     "\n",

  202.     "@author: ljia\n",

  203.     "\"\"\"\n",

  204.     "\n",

  205.     "import functools\n",

  206.     "from libs import *\n",

  207.     "import multiprocessing\n",

  208.     "\n",

  209.     "from pygraph.kernels.structuralspKernel import structuralspkernel\n",

  210.     "from pygraph.utils.kernels import deltakernel, gaussiankernel, kernelproduct\n",

  211.     "\n",

  212.     "dslist = [\n",

  213.     "#    {'name': 'Acyclic', 'dataset': '../datasets/acyclic/dataset_bps.ds',\n",

  214.     "#        'task': 'regression'},  # node symb\n",

  215.     "#    {'name': 'Alkane', 'dataset': '../datasets/Alkane/dataset.ds', 'task': 'regression',\n",

  216.     "#             'dataset_y': '../datasets/Alkane/dataset_boiling_point_names.txt', },  \n",

  217.     "#    # contains single node graph, node symb\n",

  218.     "    {'name': 'MAO', 'dataset': '../datasets/MAO/dataset.ds', },  # node/edge symb\n",

  219.     "#    {'name': 'PAH', 'dataset': '../datasets/PAH/dataset.ds', },  # unlabeled\n",

  220.     "#    {'name': 'MUTAG', 'dataset': '../datasets/MUTAG/MUTAG.mat',\n",

  221.     "#             'extra_params': {'am_sp_al_nl_el': [0, 0, 3, 1, 2]}},  # node/edge symb\n",

  222.     "#    {'name': 'Letter-med', 'dataset': '../datasets/Letter-med/Letter-med_A.txt'},\n",

  223.     "    # node nsymb\n",

  224.     "#    {'name': 'ENZYMES', 'dataset': '../datasets/ENZYMES_txt/ENZYMES_A_sparse.txt'},\n",

  225.     "#    # node symb/nsymb\n",

  226.     "#    {'name': 'Mutagenicity', 'dataset': '../datasets/Mutagenicity/Mutagenicity_A.txt'},\n",

  227.     "#    # node/edge symb\n",

  228.     "#    {'name': 'D&D', 'dataset': '../datasets/D&D/DD.mat',\n",

  229.     "#     'extra_params': {'am_sp_al_nl_el': [0, 1, 2, 1, -1]}},  # node symb\n",

  230.     "\n",

  231.     "    #     {'name': 'COIL-DEL', 'dataset': '../datasets/COIL-DEL/COIL-DEL_A.txt'}, # edge symb, node nsymb\n",

  232.     "    # # #     {'name': 'BZR', 'dataset': '../datasets/BZR_txt/BZR_A_sparse.txt'}, # node symb/nsymb\n",

  233.     "    # # #     {'name': 'COX2', 'dataset': '../datasets/COX2_txt/COX2_A_sparse.txt'}, # node symb/nsymb\n",

  234.     "    #     {'name': 'Fingerprint', 'dataset': '../datasets/Fingerprint/Fingerprint_A.txt'},\n",

  235.     "    #\n",

  236.     "    # #     {'name': 'DHFR', 'dataset': '../datasets/DHFR_txt/DHFR_A_sparse.txt'}, # node symb/nsymb\n",

  237.     "    # #     {'name': 'SYNTHETIC', 'dataset': '../datasets/SYNTHETIC_txt/SYNTHETIC_A_sparse.txt'}, # node symb/nsymb\n",

  238.     "#    {'name': 'MSRC9', 'dataset': '../datasets/MSRC_9_txt/MSRC_9_A.txt'}, # node symb, missing values\n",

  239.     "#    {'name': 'MSRC21', 'dataset': '../datasets/MSRC_21_txt/MSRC_21_A.txt'}, # node symb, missing values\n",

  240.     "    # #     {'name': 'FIRSTMM_DB', 'dataset': '../datasets/FIRSTMM_DB/FIRSTMM_DB_A.txt'}, # node symb/nsymb ,edge nsymb\n",

  241.     "\n",

  242.     "    # #     {'name': 'PROTEINS', 'dataset': '../datasets/PROTEINS_txt/PROTEINS_A_sparse.txt'}, # node symb/nsymb\n",

  243.     "    # #     {'name': 'PROTEINS_full', 'dataset': '../datasets/PROTEINS_full_txt/PROTEINS_full_A_sparse.txt'}, # node symb/nsymb\n",

  244.     "    # #     {'name': 'AIDS', 'dataset': '../datasets/AIDS/AIDS_A.txt'}, # node symb/nsymb, edge symb\n",

  245.     "    #     {'name': 'NCI1', 'dataset': '../datasets/NCI1/NCI1.mat',\n",

  246.     "    #         'extra_params': {'am_sp_al_nl_el': [1, 1, 2, 0, -1]}}, # node symb\n",

  247.     "    #     {'name': 'NCI109', 'dataset': '../datasets/NCI109/NCI109.mat',\n",

  248.     "    #         'extra_params': {'am_sp_al_nl_el': [1, 1, 2, 0, -1]}}, # node symb\n",

  249.     "    #     {'name': 'NCI-HIV', 'dataset': '../datasets/NCI-HIV/AIDO99SD.sdf',\n",

  250.     "    #         'dataset_y': '../datasets/NCI-HIV/aids_conc_may04.txt',}, # node/edge symb\n",

  251.     "\n",

  252.     "#     # not working below\n",

  253.     "#     {'name': 'PTC_FM', 'dataset': '../datasets/PTC/Train/FM.ds',},\n",

  254.     "    #     {'name': 'PTC_FR', 'dataset': '../datasets/PTC/Train/FR.ds',},\n",

  255.     "    #     {'name': 'PTC_MM', 'dataset': '../datasets/PTC/Train/MM.ds',},\n",

  256.     "    #     {'name': 'PTC_MR', 'dataset': '../datasets/PTC/Train/MR.ds',},\n",

  257.     "]\n",

  258.     "estimator = structuralspkernel\n",

  259.     "mixkernel = functools.partial(kernelproduct, deltakernel, gaussiankernel)\n",

  260.     "param_grid_precomputed = {'node_kernels': \n",

  261.     "    [{'symb': deltakernel, 'nsymb': gaussiankernel, 'mix': mixkernel}],\n",

  262.     "    'edge_kernels': \n",

  263.     "    [{'symb': deltakernel, 'nsymb': gaussiankernel, 'mix': mixkernel}]}\n",

  264.     "param_grid = [{'C': np.logspace(-10, 10, num=41, base=10)},\n",

  265.     "              {'alpha': np.logspace(-10, 10, num=41, base=10)}]\n",

  266.     "\n",

  267.     "for ds in dslist:\n",

  268.     "    print()\n",

  269.     "    print(ds['name'])\n",

  270.     "    model_selection_for_precomputed_kernel(\n",

  271.     "        ds['dataset'],\n",

  272.     "        estimator,\n",

  273.     "        param_grid_precomputed,\n",

  274.     "        (param_grid[1] if ('task' in ds and ds['task']\n",

  275.     "                           == 'regression') else param_grid[0]),\n",

  276.     "        (ds['task'] if 'task' in ds else 'classification'),\n",

  277.     "        NUM_TRIALS=30,\n",

  278.     "        datafile_y=(ds['dataset_y'] if 'dataset_y' in ds else None),\n",

  279.     "        extra_params=(ds['extra_params'] if 'extra_params' in ds else None),\n",

  280.     "        ds_name=ds['name'],\n",

  281.     "        n_jobs=multiprocessing.cpu_count(),\n",

  282.     "        read_gm_from_file=False)\n",

  283.     "    print()"

  284.    ]

  285.   }

  286.  ],

  287.  "metadata": {

  288.   "kernelspec": {

  289.    "display_name": "Python 3",

  290.    "language": "python",

  291.    "name": "python3"

  292.   },

  293.   "language_info": {

  294.    "codemirror_mode": {

  295.     "name": "ipython",

  296.     "version": 3

  297.    },

  298.    "file_extension": ".py",

  299.    "mimetype": "text/x-python",

  300.    "name": "python",

  301.    "nbconvert_exporter": "python",

  302.    "pygments_lexer": "ipython3",

  303.    "version": "3.6.6"

  304.   }

  305.  },

  306.  "nbformat": 4,

  307.  "nbformat_minor": 2

  308. }

A Python package for graph kernels, graph edit distances and graph pre-image problem.