demos/tabular_kitsune/tabular_example_kitsune.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "0f0a1d8b",
   "metadata": {},
   "source": [
    "This exmaple will show you how to use DeepAID Interpreter upon a SOTA NIDS *Kitsune* (NDSS'18).\n",
    "\n",
    "We simplify the DL model to a single autoencoder for ease of illustration."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cf8cca62",
   "metadata": {},
   "source": [
    " # Prepare your DL model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "b71e28fd",
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "31d1b0d8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch:0/0 |Loss: 0.050325267016887665\n",
      "epoch:0/10 |Loss: 0.03627059981226921\n",
      "epoch:0/20 |Loss: 0.02359006181359291\n",
      "epoch:0/30 |Loss: 0.017687782645225525\n",
      "epoch:0/40 |Loss: 0.012251004576683044\n",
      "epoch:0/50 |Loss: 0.011910964734852314\n",
      "epoch:0/60 |Loss: 0.010575455613434315\n",
      "epoch:0/70 |Loss: 0.010612315498292446\n",
      "epoch:0/80 |Loss: 0.008414724841713905\n",
      "epoch:0/90 |Loss: 0.011196578852832317\n",
      "epoch:0/100 |Loss: 0.007897039875388145\n",
      "epoch:0/110 |Loss: 0.00968477688729763\n",
      "epoch:0/120 |Loss: 0.007968610152602196\n",
      "epoch:0/130 |Loss: 0.007041106000542641\n",
      "epoch:0/140 |Loss: 0.01321518886834383\n",
      "epoch:0/150 |Loss: 0.007542287930846214\n",
      "epoch:0/160 |Loss: 0.007705105468630791\n",
      "epoch:0/170 |Loss: 0.008091779425740242\n",
      "epoch:0/180 |Loss: 0.007869303226470947\n",
      "epoch:0/190 |Loss: 0.008114654570817947\n",
      "epoch:0/200 |Loss: 0.006699248217046261\n",
      "epoch:0/210 |Loss: 0.007377055939286947\n",
      "epoch:0/220 |Loss: 0.007573551964014769\n",
      "epoch:0/230 |Loss: 0.0070730852894485\n",
      "epoch:0/240 |Loss: 0.011596474796533585\n",
      "epoch:0/250 |Loss: 0.005361731164157391\n",
      "epoch:0/260 |Loss: 0.006951004732400179\n",
      "epoch:0/270 |Loss: 0.007856125943362713\n",
      "epoch:0/280 |Loss: 0.008307741954922676\n",
      "epoch:0/290 |Loss: 0.006409224588423967\n",
      "epoch:0/300 |Loss: 0.0051156640984117985\n",
      "epoch:0/310 |Loss: 0.006705230101943016\n",
      "epoch:0/320 |Loss: 0.004329719580709934\n",
      "epoch:0/330 |Loss: 0.004316339269280434\n",
      "epoch:0/340 |Loss: 0.00337043683975935\n",
      "epoch:0/350 |Loss: 0.0053898910991847515\n",
      "epoch:0/360 |Loss: 0.003100291360169649\n",
      "epoch:0/370 |Loss: 0.005040488671511412\n",
      "epoch:0/380 |Loss: 0.0031234563793987036\n",
      "epoch:0/390 |Loss: 0.0033457146491855383\n",
      "epoch:1/0 |Loss: 0.0038628345355391502\n",
      "epoch:1/10 |Loss: 0.004464179743081331\n",
      "epoch:1/20 |Loss: 0.0035455524921417236\n",
      "epoch:1/30 |Loss: 0.0034786032047122717\n",
      "epoch:1/40 |Loss: 0.003083174116909504\n",
      "epoch:1/50 |Loss: 0.0030969639774411917\n",
      "epoch:1/60 |Loss: 0.0032839993946254253\n",
      "epoch:1/70 |Loss: 0.002400841098278761\n",
      "epoch:1/80 |Loss: 0.00324723357334733\n",
      "epoch:1/90 |Loss: 0.004016987979412079\n",
      "epoch:1/100 |Loss: 0.004022143315523863\n",
      "epoch:1/110 |Loss: 0.002981956582516432\n",
      "epoch:1/120 |Loss: 0.0025861330796033144\n",
      "epoch:1/130 |Loss: 0.004187081474810839\n",
      "epoch:1/140 |Loss: 0.0022581478115171194\n",
      "epoch:1/150 |Loss: 0.004256002604961395\n",
      "epoch:1/160 |Loss: 0.0024767834693193436\n",
      "epoch:1/170 |Loss: 0.0036459651309996843\n",
      "epoch:1/180 |Loss: 0.0023169966880232096\n",
      "epoch:1/190 |Loss: 0.0030443768482655287\n",
      "epoch:1/200 |Loss: 0.002797934925183654\n",
      "epoch:1/210 |Loss: 0.002373020863160491\n",
      "epoch:1/220 |Loss: 0.002218068577349186\n",
      "epoch:1/230 |Loss: 0.002855229889973998\n",
      "epoch:1/240 |Loss: 0.0016848144587129354\n",
      "epoch:1/250 |Loss: 0.0024556771386414766\n",
      "epoch:1/260 |Loss: 0.0021225165110081434\n",
      "epoch:1/270 |Loss: 0.0034225028939545155\n",
      "epoch:1/280 |Loss: 0.0029378824401646852\n",
      "epoch:1/290 |Loss: 0.0026784955989569426\n",
      "epoch:1/300 |Loss: 0.0021183935459703207\n",
      "epoch:1/310 |Loss: 0.0016185662243515253\n",
      "epoch:1/320 |Loss: 0.0026244735345244408\n",
      "epoch:1/330 |Loss: 0.002685840940102935\n",
      "epoch:1/340 |Loss: 0.0022861575707793236\n",
      "epoch:1/350 |Loss: 0.0013575527118518949\n",
      "epoch:1/360 |Loss: 0.0020253441762179136\n",
      "epoch:1/370 |Loss: 0.0019165551057085395\n",
      "epoch:1/380 |Loss: 0.002657421166077256\n",
      "epoch:1/390 |Loss: 0.0015685366233810782\n",
      "epoch:2/0 |Loss: 0.00232278136536479\n",
      "epoch:2/10 |Loss: 0.0029039066284894943\n",
      "epoch:2/20 |Loss: 0.0021594783756881952\n",
      "epoch:2/30 |Loss: 0.0017693719128146768\n",
      "epoch:2/40 |Loss: 0.002571800956502557\n",
      "epoch:2/50 |Loss: 0.0032666001934558153\n",
      "epoch:2/60 |Loss: 0.001570073887705803\n",
      "epoch:2/70 |Loss: 0.0015154615975916386\n",
      "epoch:2/80 |Loss: 0.0018475247779861093\n",
      "epoch:2/90 |Loss: 0.001702390261925757\n",
      "epoch:2/100 |Loss: 0.002314829733222723\n",
      "epoch:2/110 |Loss: 0.002125282771885395\n",
      "epoch:2/120 |Loss: 0.0016278231050819159\n",
      "epoch:2/130 |Loss: 0.0017578088445588946\n",
      "epoch:2/140 |Loss: 0.0018152121920138597\n",
      "epoch:2/150 |Loss: 0.0016138614155352116\n",
      "epoch:2/160 |Loss: 0.0014542266726493835\n",
      "epoch:2/170 |Loss: 0.0017631753580644727\n",
      "epoch:2/180 |Loss: 0.0019125946564599872\n",
      "epoch:2/190 |Loss: 0.0012341036926954985\n",
      "epoch:2/200 |Loss: 0.0014733034186065197\n",
      "epoch:2/210 |Loss: 0.0019256646046414971\n",
      "epoch:2/220 |Loss: 0.0019772076047956944\n",
      "epoch:2/230 |Loss: 0.0019708171021193266\n",
      "epoch:2/240 |Loss: 0.0012469268403947353\n",
      "epoch:2/250 |Loss: 0.0014844831312075257\n",
      "epoch:2/260 |Loss: 0.001704549416899681\n",
      "epoch:2/270 |Loss: 0.0016958836931735277\n",
      "epoch:2/280 |Loss: 0.0025130026042461395\n",
      "epoch:2/290 |Loss: 0.0022826315835118294\n",
      "epoch:2/300 |Loss: 0.001563460100442171\n",
      "epoch:2/310 |Loss: 0.0014427624410018325\n",
      "epoch:2/320 |Loss: 0.0015313387848436832\n",
      "epoch:2/330 |Loss: 0.0016081504290923476\n",
      "epoch:2/340 |Loss: 0.0013760164147242904\n",
      "epoch:2/350 |Loss: 0.0021990404929965734\n",
      "epoch:2/360 |Loss: 0.0015894025564193726\n",
      "epoch:2/370 |Loss: 0.0015072234673425555\n",
      "epoch:2/380 |Loss: 0.0018394431099295616\n",
      "epoch:2/390 |Loss: 0.0018913329113274813\n",
      "epoch:3/0 |Loss: 0.001685205614194274\n",
      "epoch:3/10 |Loss: 0.0016344197792932391\n",
      "epoch:3/20 |Loss: 0.0023301108740270138\n",
      "epoch:3/30 |Loss: 0.0017432396998628974\n",
      "epoch:3/40 |Loss: 0.0015499192522838712\n",
      "epoch:3/50 |Loss: 0.001785294502042234\n",
      "epoch:3/60 |Loss: 0.001116467290557921\n",
      "epoch:3/70 |Loss: 0.0019956170581281185\n",
      "epoch:3/80 |Loss: 0.002037512604147196\n",
      "epoch:3/90 |Loss: 0.0016433956334367394\n",
      "epoch:3/100 |Loss: 0.0017306818626821041\n",
      "epoch:3/110 |Loss: 0.002142460085451603\n",
      "epoch:3/120 |Loss: 0.0020319754257798195\n",
      "epoch:3/130 |Loss: 0.0011512554483488202\n",
      "epoch:3/140 |Loss: 0.0011671106331050396\n",
      "epoch:3/150 |Loss: 0.0014392656739801168\n",
      "epoch:3/160 |Loss: 0.0019960009958595037\n",
      "epoch:3/170 |Loss: 0.001458079437725246\n",
      "epoch:3/180 |Loss: 0.0014460848178714514\n",
      "epoch:3/190 |Loss: 0.00135422358289361\n",
      "epoch:3/200 |Loss: 0.0019192337058484554\n",
      "epoch:3/210 |Loss: 0.0015596328303217888\n",
      "epoch:3/220 |Loss: 0.0013185185380280018\n",
      "epoch:3/230 |Loss: 0.0015578367747366428\n",
      "epoch:3/240 |Loss: 0.0022624062839895487\n",
      "epoch:3/250 |Loss: 0.0023743566125631332\n",
      "epoch:3/260 |Loss: 0.0014028421137481928\n",
      "epoch:3/270 |Loss: 0.0010667757596820593\n",
      "epoch:3/280 |Loss: 0.00198633735999465\n",
      "epoch:3/290 |Loss: 0.0016214322531595826\n",
      "epoch:3/300 |Loss: 0.0031898734159767628\n",
      "epoch:3/310 |Loss: 0.0013733028899878263\n",
      "epoch:3/320 |Loss: 0.001628084690310061\n",
      "epoch:3/330 |Loss: 0.0011552704963833094\n",
      "epoch:3/340 |Loss: 0.0012287137797102332\n",
      "epoch:3/350 |Loss: 0.0011162623995915055\n",
      "epoch:3/360 |Loss: 0.0017652156529948115\n",
      "epoch:3/370 |Loss: 0.0012978205922991037\n",
      "epoch:3/380 |Loss: 0.0011822329834103584\n",
      "epoch:3/390 |Loss: 0.0013772716047242284\n",
      "epoch:4/0 |Loss: 0.0010377406142652035\n",
      "epoch:4/10 |Loss: 0.001564549165777862\n",
      "epoch:4/20 |Loss: 0.0015781630063429475\n",
      "epoch:4/30 |Loss: 0.0014954006765037775\n",
      "epoch:4/40 |Loss: 0.0020295821595937014\n",
      "epoch:4/50 |Loss: 0.001755604869686067\n",
      "epoch:4/60 |Loss: 0.001435768324881792\n",
      "epoch:4/70 |Loss: 0.0011050428729504347\n",
      "epoch:4/80 |Loss: 0.0014876436907798052\n",
      "epoch:4/90 |Loss: 0.0016420247266069055\n",
      "epoch:4/100 |Loss: 0.0012799014803022146\n",
      "epoch:4/110 |Loss: 0.0016708022449165583\n",
      "epoch:4/120 |Loss: 0.0018290167208760977\n",
      "epoch:4/130 |Loss: 0.0009261663071811199\n",
      "epoch:4/140 |Loss: 0.0013843950582668185\n",
      "epoch:4/150 |Loss: 0.0016529449494555593\n",
      "epoch:4/160 |Loss: 0.0015769515885040164\n",
      "epoch:4/170 |Loss: 0.0018727830611169338\n",
      "epoch:4/180 |Loss: 0.0011414922773838043\n",
      "epoch:4/190 |Loss: 0.0012360161636024714\n",
      "epoch:4/200 |Loss: 0.0012412231881171465\n",
      "epoch:4/210 |Loss: 0.001102812122553587\n",
      "epoch:4/220 |Loss: 0.002097640885040164\n",
      "epoch:4/230 |Loss: 0.001972807105630636\n",
      "epoch:4/240 |Loss: 0.001285040401853621\n",
      "epoch:4/250 |Loss: 0.002720164367929101\n",
      "epoch:4/260 |Loss: 0.0019576388876885176\n",
      "epoch:4/270 |Loss: 0.001643347553908825\n",
      "epoch:4/280 |Loss: 0.0014596080873161554\n",
      "epoch:4/290 |Loss: 0.0014529097825288773\n",
      "epoch:4/300 |Loss: 0.0023539713583886623\n",
      "epoch:4/310 |Loss: 0.0016572007443755865\n",
      "epoch:4/320 |Loss: 0.001838783617131412\n",
      "epoch:4/330 |Loss: 0.002076963661238551\n",
      "epoch:4/340 |Loss: 0.0020481825340539217\n",
      "epoch:4/350 |Loss: 0.0014223232865333557\n",
      "epoch:4/360 |Loss: 0.0012290411395952106\n",
      "epoch:4/370 |Loss: 0.001254181144759059\n",
      "epoch:4/380 |Loss: 0.0013801043387502432\n",
      "epoch:4/390 |Loss: 0.0011211500968784094\n",
      "max AD score 0.38217393\n",
      "thres: 0.3455014\n"
     ]
    }
   ],
   "source": [
    "# Train an autoencoder-based DL model\n",
    "import numpy as np\n",
    "import torch\n",
    "from autoencoder import train, test, test_plot\n",
    "import sys\n",
    "sys.path.append('../../deepaid/')\n",
    "from utils import validate_by_rmse, Normalizer\n",
    "train_feat = np.load('data/train_benign_feat.npy')\n",
    "normer = Normalizer(train_feat.shape[-1],online_minmax=True)\n",
    "train_feat = normer.fit_transform(train_feat)\n",
    "model, thres = train(train_feat, train_feat.shape[-1])\n",
    "torch.save({'net':model,'thres':thres},'./save/autoencoder.pth.tar')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "cec19f07",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Validate the performance of trained model\n",
    "test_feat = np.load('data/test_mirai_ddos.npy')\n",
    "test_feat = normer.transform(test_feat)\n",
    "rmse_vec = test(model,thres,test_feat)\n",
    "test_plot(test_feat, rmse_vec, thres) # ACC = 0.84"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2f5f07e7",
   "metadata": {},
   "source": [
    "# Interpret your interested anomaly in four steps "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "659b1aa0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Successfully Initialize <Tabular Interptreter> for Model <autoencoder>\n",
      "Finish Interpretation after 499 steps (Final loss: 0.16, Time elasped: 1.20s)\n",
      "\n",
      "Visualize Interpretation (Table View)\n",
      "+---------------------+------------------+-------+--------------------+\n",
      "| Feature Description | Value in Anomaly | comp. | Value in Reference |\n",
      "+---------------------+------------------+-------+--------------------+\n",
      "|  MI_dir_0.1_weight  |     3227.63      |   >   |      483.413       |\n",
      "|  MI_dir_0.01_weight |    23117.713     |   >   |      4259.929      |\n",
      "|  MI_dir_1.0_weight  |     386.527      |   >   |      141.729       |\n",
      "|  MI_dir_3.0_weight  |     170.356      |   >   |       70.842       |\n",
      "|  MI_dir_5.0_weight  |      124.98      |   >   |       61.988       |\n",
      "+---------------------+------------------+-------+--------------------+\n",
      "\n",
      "Visualize Interpretation (Plot View)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Visualize Interpretation (HeatMap View)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\"\"\"Step 1: Load your model\"\"\"\n",
    "from autoencoder import autoencoder\n",
    "from utils import Normalizer\n",
    "model_dict = torch.load('save/autoencoder.pth.tar')\n",
    "model = model_dict['net']\n",
    "thres = model_dict['thres']\n",
    "\n",
    "\"\"\"Step 2: Find an anomaly you are interested in (here we use the IP scan in Mirai botnet attack)\"\"\"\n",
    "anomaly = test_feat[np.argsort(rmse_vec)[-100]]\n",
    "\n",
    "\"\"\"Step 3: Create a DeepAID Tabular Interpreter\"\"\"\n",
    "import sys\n",
    "sys.path.append(\"../../deepaid/interpreters/\")\n",
    "from tabular import TabularAID\n",
    "feature_desc = np.load('data/AI_feature_name.npy') # feature_description\n",
    "my_interpreter = TabularAID(model,thres,input_size=100,feature_desc=feature_desc)\n",
    "\n",
    "\"\"\"Step 4: Interpret your anomaly and show the result\"\"\"\n",
    "interpretation = my_interpreter(anomaly)\n",
    "# DeepAID supports three kinds of visualization of results:\n",
    "my_interpreter.show_table(anomaly,interpretation, normer) \n",
    "my_interpreter.show_plot(anomaly, interpretation, normer)\n",
    "my_interpreter.show_heatmap(anomaly,interpretation, normer)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2aeb3392",
   "metadata": {},
   "source": [
    "**As we can see that, the top-5 influencial features from the results of Interpreter are all related to *MI_dir* (packets sent from the same MAC&IP).\n",
    "This confirms the anomaly is a IP Scanning.**"
   ]
  }
 ],
 "metadata": {
  "hide_input": false,
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.11"
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}