LSTM Malicious URL Detector.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/hvyd/anaconda3/lib/python3.5/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n",
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import re, os\n",
    "from string import printable\n",
    "from sklearn import model_selection\n",
    "\n",
    "import tensorflow as tf\n",
    "from keras.models import Sequential, Model, model_from_json, load_model\n",
    "from keras import regularizers\n",
    "from keras.layers.core import Dense, Dropout, Activation, Lambda, Flatten\n",
    "from keras.layers import *\n",
    "from keras.preprocessing import sequence\n",
    "from keras.optimizers import SGD, Adam, RMSprop\n",
    "from keras.utils import np_utils\n",
    "from keras import backend as K\n",
    "\n",
    "from pathlib import Path\n",
    "import json\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Preparations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>url</th>\n",
       "      <th>isMalicious</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>songlyrics.com/news/riffd-the-shins-heartworms</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>imaging-resource.com/PRODS/olympus-e-m1-ii/oly...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>gosugamers.net/lol/streams</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>thingiverse.com/corkyzett/collections/intlwome...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>bausch.com/our-products/contact-lens-care/spec...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                 url  isMalicious\n",
       "0     songlyrics.com/news/riffd-the-shins-heartworms            0\n",
       "1  imaging-resource.com/PRODS/olympus-e-m1-ii/oly...            0\n",
       "2                         gosugamers.net/lol/streams            0\n",
       "3  thingiverse.com/corkyzett/collections/intlwome...            0\n",
       "4  bausch.com/our-products/contact-lens-care/spec...            0"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "DATA = 'data/URL.csv'\n",
    "df = pd.read_csv(DATA)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(194798, 2)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>isMalicious</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>194798.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.500001</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         isMalicious\n",
       "count  194798.000000\n",
       "mean        0.500000\n",
       "std         0.500001\n",
       "min         0.000000\n",
       "25%         0.000000\n",
       "50%         0.500000\n",
       "75%         1.000000\n",
       "max         1.000000"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 194798 entries, 0 to 194797\n",
      "Data columns (total 2 columns):\n",
      "url            194798 non-null object\n",
      "isMalicious    194798 non-null int64\n",
      "dtypes: int64(1), object(1)\n",
      "memory usage: 3.0+ MB\n"
     ]
    }
   ],
   "source": [
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Matrix dimensions of X:  (194798, 75) Vector dimension of target:  (194798,)\n"
     ]
    }
   ],
   "source": [
    "max_len=75\n",
    "url_int_tokens = [[printable.index(x) + 1 for x in url if x in printable] for url in df.url]\n",
    "X = sequence.pad_sequences(url_int_tokens, maxlen=max_len)\n",
    "target = np.array(df.isMalicious)\n",
    "print('Matrix dimensions of X: ', X.shape, 'Vector dimension of target: ', target.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, target_train, target_test = model_selection.train_test_split(X, target, test_size=0.2, random_state=42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model Preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def print_layers_dims(model):\n",
    "    l_layers = model.layers\n",
    "    for i in range(len(l_layers)):\n",
    "        print(l_layers[i])\n",
    "        print('Input Shape: ', l_layers[i].input_shape, 'Output Shape: ', l_layers[i].output_shape)\n",
    "\n",
    "\n",
    "def save_model(fileModelJSON,fileWeights):\n",
    "    if Path(fileModelJSON).is_file():\n",
    "        os.remove(fileModelJSON)\n",
    "    json_string = model.to_json()\n",
    "    with open(fileModelJSON,'w' ) as f:\n",
    "        json.dump(json_string, f)\n",
    "    if Path(fileWeights).is_file():\n",
    "        os.remove(fileWeights)\n",
    "    model.save_weights(fileWeights)\n",
    "    \n",
    "def load_model(fileModelJSON,fileWeights):\n",
    "    with open(fileModelJSON, 'r') as f:\n",
    "         model_json = json.load(f)\n",
    "         model = model_from_json(model_json)\n",
    "    \n",
    "    model.load_weights(fileWeights)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### LSTM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def simple_lstm(max_len=75, emb_dim=32, max_vocab_len=100, lstm_output_size=32, W_reg=regularizers.l2(1e-4)):\n",
    "    main_input = Input(shape=(max_len,), dtype='int32', name='main_input')\n",
    "    emb = Embedding(input_dim=max_vocab_len, output_dim=emb_dim, input_length=max_len,\n",
    "                dropout=0.2, W_regularizer=W_reg)(main_input) \n",
    "\n",
    "    lstm = LSTM(lstm_output_size)(emb)\n",
    "    lstm = Dropout(0.5)(lstm)\n",
    "    \n",
    "    output = Dense(1, activation='sigmoid', name='output')(lstm)\n",
    "\n",
    "    model = Model(input=[main_input], output=[output])\n",
    "    adam = Adam(lr=1e-4, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0)\n",
    "    model.compile(optimizer=adam, loss='binary_crossentropy', metrics=['accuracy'])\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/hvyd/anaconda3/lib/python3.5/site-packages/keras/backend/tensorflow_backend.py:1029: calling reduce_sum (from tensorflow.python.ops.math_ops) with keep_dims is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "keep_dims is deprecated, use keepdims instead\n",
      "WARNING:tensorflow:From /home/hvyd/anaconda3/lib/python3.5/site-packages/keras/backend/tensorflow_backend.py:1108: calling reduce_mean (from tensorflow.python.ops.math_ops) with keep_dims is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "keep_dims is deprecated, use keepdims instead\n",
      "Epoch 1/5\n",
      "155838/155838 [==============================] - 4083s - loss: 0.4606 - acc: 0.7880  \n",
      "Epoch 2/5\n",
      "155838/155838 [==============================] - 4129s - loss: 0.3963 - acc: 0.8257  \n",
      "Epoch 3/5\n",
      "155838/155838 [==============================] - 4161s - loss: 0.3705 - acc: 0.8389  \n",
      "Epoch 4/5\n",
      "155838/155838 [==============================] - 4182s - loss: 0.3522 - acc: 0.8481  \n",
      "Epoch 5/5\n",
      "155838/155838 [==============================] - 4218s - loss: 0.3359 - acc: 0.8571  \n",
      "38912/38960 [============================>.] - ETA: 0s\n",
      "Final Cross-Validation Accuracy 0.8829568788501027 \n",
      "\n",
      "<keras.engine.topology.InputLayer object at 0x7f3273718da0>\n",
      "Input Shape:  (None, 75) Output Shape:  (None, 75)\n",
      "<keras.layers.embeddings.Embedding object at 0x7f32736bb0f0>\n",
      "Input Shape:  (None, 75) Output Shape:  (None, 75, 32)\n",
      "<keras.layers.recurrent.LSTM object at 0x7f3272685518>\n",
      "Input Shape:  (None, 75, 32) Output Shape:  (None, 32)\n",
      "<keras.layers.core.Dropout object at 0x7f32705c0dd8>\n",
      "Input Shape:  (None, 32) Output Shape:  (None, 32)\n",
      "<keras.layers.core.Dense object at 0x7f327055b3c8>\n",
      "Input Shape:  (None, 32) Output Shape:  (None, 1)\n"
     ]
    }
   ],
   "source": [
    "epochs = 2\n",
    "batch_size = 5\n",
    "\n",
    "model = simple_lstm()\n",
    "model.fit(X_train, target_train, epochs, batch_size)\n",
    "loss, accuracy = model.evaluate(X_test, target_test, verbose=1)\n",
    "\n",
    "print('\\nFinal Cross-Validation Accuracy', accuracy, '\\n')\n",
    "print_layers_dims(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "____________________________________________________________________________________________________\n",
      "Layer (type)                     Output Shape          Param #     Connected to                     \n",
      "====================================================================================================\n",
      "main_input (InputLayer)          (None, 75)            0                                            \n",
      "____________________________________________________________________________________________________\n",
      "embedding_1 (Embedding)          (None, 75, 32)        3200        main_input[0][0]                 \n",
      "____________________________________________________________________________________________________\n",
      "lstm_1 (LSTM)                    (None, 32)            8320        embedding_1[0][0]                \n",
      "____________________________________________________________________________________________________\n",
      "dropout_1 (Dropout)              (None, 32)            0           lstm_1[0][0]                     \n",
      "____________________________________________________________________________________________________\n",
      "output (Dense)                   (None, 1)             33          dropout_1[0][0]                  \n",
      "====================================================================================================\n",
      "Total params: 11,553\n",
      "Trainable params: 11,553\n",
      "Non-trainable params: 0\n",
      "____________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "DATA_HOME = 'data'\n",
    "model_name = \"LSTM\"\n",
    "save_model(DATA_HOME + model_name + \".json\", DATA_HOME + model_name + \".h5\")\n",
    "model = load_model(DATA_HOME + model_name + \".json\", DATA_HOME + model_name + \".h5\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Convolutional LSTM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "____________________________________________________________________________________________________\n",
      "Layer (type)                     Output Shape          Param #     Connected to                     \n",
      "====================================================================================================\n",
      "main_input (InputLayer)          (None, 75)            0                                            \n",
      "____________________________________________________________________________________________________\n",
      "embedding_1 (Embedding)          (None, 75, 32)        3200        main_input[0][0]                 \n",
      "____________________________________________________________________________________________________\n",
      "lstm_1 (LSTM)                    (None, 32)            8320        embedding_1[0][0]                \n",
      "____________________________________________________________________________________________________\n",
      "dropout_1 (Dropout)              (None, 32)            0           lstm_1[0][0]                     \n",
      "____________________________________________________________________________________________________\n",
      "output (Dense)                   (None, 1)             33          dropout_1[0][0]                  \n",
      "====================================================================================================\n",
      "Total params: 11,553\n",
      "Trainable params: 11,553\n",
      "Non-trainable params: 0\n",
      "____________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "def lstm_conv(max_len=75, emb_dim=32, max_vocab_len=100, lstm_output_size=32, W_reg=regularizers.l2(1e-4)):\n",
    "    main_input = Input(shape=(max_len,), dtype='int32', name='main_input')\n",
    "    emb = Embedding(input_dim=max_vocab_len, output_dim=emb_dim, input_length=max_len,\n",
    "                W_regularizer=W_reg)(main_input) \n",
    "    emb = Dropout(0.25)(emb)\n",
    "\n",
    "    conv = Convolution1D(kernel_size=5, filters=256, \\\n",
    "                     border_mode='same')(emb)\n",
    "    conv = ELU()(conv)\n",
    "\n",
    "    conv = MaxPooling1D(pool_size=4)(conv)\n",
    "    conv = Dropout(0.5)(conv)\n",
    "\n",
    "    lstm = LSTM(lstm_output_size)(conv)\n",
    "    lstm = Dropout(0.5)(lstm)\n",
    "    \n",
    "    output = Dense(1, activation='sigmoid', name='output')(lstm)\n",
    "\n",
    "    model = Model(input=[main_input], output=[output])\n",
    "    adam = Adam(lr=1e-4, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0)\n",
    "    model.compile(optimizer=adam, loss='binary_crossentropy', metrics=['accuracy'])\n",
    "    return model\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "epochs = 2\n",
    "batch_size = 5\n",
    "\n",
    "model = lstm_conv()\n",
    "model.fit(X_train, target_train, epochs, batch_size)\n",
    "loss, accuracy = model.evaluate(X_test, target_test, verbose=1)\n",
    "\n",
    "print('\\nFinal Cross-Validation Accuracy', accuracy, '\\n')\n",
    "print_layers_dims(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "DATA_HOME = 'data'\n",
    "model_name = \"1DConvLSTM\"\n",
    "save_model(DATA_HOME + model_name + \".json\", DATA_HOME + model_name + \".h5\")\n",
    "model = load_model(DATA_HOME + model_name + \".json\", DATA_HOME + model_name + \".h5\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Convolutional Fully Connected"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "____________________________________________________________________________________________________\n",
      "Layer (type)                     Output Shape          Param #     Connected to                     \n",
      "====================================================================================================\n",
      "main_input (InputLayer)          (None, 75)            0                                            \n",
      "____________________________________________________________________________________________________\n",
      "embedding_1 (Embedding)          (None, 75, 32)        3200        main_input[0][0]                 \n",
      "____________________________________________________________________________________________________\n",
      "lstm_1 (LSTM)                    (None, 32)            8320        embedding_1[0][0]                \n",
      "____________________________________________________________________________________________________\n",
      "dropout_1 (Dropout)              (None, 32)            0           lstm_1[0][0]                     \n",
      "____________________________________________________________________________________________________\n",
      "output (Dense)                   (None, 1)             33          dropout_1[0][0]                  \n",
      "====================================================================================================\n",
      "Total params: 11,553\n",
      "Trainable params: 11,553\n",
      "Non-trainable params: 0\n",
      "____________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "def conv_fully(max_len=75, emb_dim=32, max_vocab_len=100, W_reg=regularizers.l2(1e-4)):\n",
    "    main_input = Input(shape=(max_len,), dtype='int32', name='main_input')\n",
    "    emb = Embedding(input_dim=max_vocab_len, output_dim=emb_dim, input_length=max_len,\n",
    "                W_regularizer=W_reg)(main_input) \n",
    "    emb = Dropout(0.25)(emb)\n",
    "\n",
    "    \n",
    "    def sum_1d(X):\n",
    "        return K.sum(X, axis=1)\n",
    "    \n",
    "    def get_conv_layer(emb, kernel_size=5, filters=256):\n",
    "        conv = Convolution1D(kernel_size=kernel_size, filters=filters, \\\n",
    "                     border_mode='same')(emb)\n",
    "        conv = ELU()(conv)\n",
    "\n",
    "        conv = Lambda(sum_1d, output_shape=(filters,))(conv)\n",
    "        conv = Dropout(0.5)(conv)\n",
    "        return conv\n",
    "        \n",
    " \n",
    "    conv1 = get_conv_layer(emb, kernel_size=2, filters=256)\n",
    "    conv2 = get_conv_layer(emb, kernel_size=3, filters=256)\n",
    "    conv3 = get_conv_layer(emb, kernel_size=4, filters=256)\n",
    "    conv4 = get_conv_layer(emb, kernel_size=5, filters=256)\n",
    "\n",
    "    merged = concatenate([conv1,conv2,conv3,conv4], axis=1)\n",
    "\n",
    "    hidden1 = Dense(1024)(merged)\n",
    "    hidden1 = ELU()(hidden1)\n",
    "    hidden1 = BatchNormalization(mode=0)(hidden1)\n",
    "    hidden1 = Dropout(0.5)(hidden1)\n",
    "\n",
    "    hidden2 = Dense(1024)(hidden1)\n",
    "    hidden2 = ELU()(hidden2)\n",
    "    hidden2 = BatchNormalization(mode=0)(hidden2)\n",
    "    hidden2 = Dropout(0.5)(hidden2)\n",
    "    \n",
    "    output = Dense(1, activation='sigmoid', name='output')(hidden2)\n",
    "\n",
    "    model = Model(input=[main_input], output=[output])\n",
    "    adam = Adam(lr=1e-4, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0)\n",
    "    model.compile(optimizer=adam, loss='binary_crossentropy', metrics=['accuracy'])\n",
    "    return model\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "epochs = 2\n",
    "batch_size = 5\n",
    "\n",
    "model = conv_fully()\n",
    "model.fit(X_train, target_train, epochs, batch_size)\n",
    "loss, accuracy = model.evaluate(X_test, target_test, verbose=1)\n",
    "\n",
    "print('\\nFinal Cross-Validation Accuracy', accuracy, '\\n')\n",
    "print_layers_dims(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "target_proba = model.predict(X_test, batch_size=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "target_proba[0:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "DATA_HOME = 'data'\n",
    "model_name = \"1DConv\"\n",
    "save_model(DATA_HOME + model_name + \".json\", DATA_HOME + model_name + \".h5\")\n",
    "model = load_model(DATA_HOME + model_name + \".json\", DATA_HOME + model_name + \".h5\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(100, 32)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "l_layers = model.layers\n",
    "weights = l_layers[1].get_weights()\n",
    "weights[0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_url_mal = \"mydrivers.com/1/524/524241.htm\"\n",
    "test_url_benign = \"ubuntulinux.org/server/hyperscale\"\n",
    "url = test_url_benign"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "max_len=75\n",
    "url_int_tokens = [[printable.index(x) + 1 for x in url if x in printable]]\n",
    "X = sequence.pad_sequences(url_int_tokens, maxlen=max_len)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test URL: ubuntulinux.org/server/hyperscale is benign\n"
     ]
    }
   ],
   "source": [
    "target_proba = model.predict(X, batch_size=1)\n",
    "def print_result(proba):\n",
    "    if proba > 0.5:\n",
    "        return \"malicious\"\n",
    "    else:\n",
    "        return \"benign\"\n",
    "print(\"Test URL:\", url, \"is\", print_result(target_proba[0]))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.5.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}