Our model IBDM Test.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 184,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import sys\n",
    "import pandas as pd\n",
    "from tensorflow.keras import  layers\n",
    "from tensorflow.keras import regularizers\n",
    "from keras.models import Sequential, load_model\n",
    "from keras.layers import Dense, Dropout, Activation, GaussianNoise, GaussianDropout\n",
    "from keras.layers import Embedding, Flatten\n",
    "from keras.layers import Conv1D, GlobalMaxPooling1D, MaxPooling1D\n",
    "from keras.callbacks import ModelCheckpoint, ReduceLROnPlateau\n",
    "from utl import utils\n",
    "from keras.preprocessing.sequence import pad_sequences\n",
    "\n",
    "from keras.layers import LSTM, Bidirectional\n",
    "from matplotlib import pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd  \n",
    "import numpy as np\n",
    "import time\n",
    "from ast import literal_eval\n",
    "from operator import add\n",
    "import matplotlib.pyplot as plt\n",
    "from mpl_toolkits.axes_grid1.inset_locator import zoomed_inset_axes\n",
    "from sklearn.model_selection import StratifiedKFold\n",
    "from collections import Counter\n",
    "from sklearn.utils import resample\n",
    "plt.style.use('fivethirtyeight')\n",
    "import re\n",
    "from bs4 import BeautifulSoup\n",
    "from nltk.tokenize import WordPunctTokenizer\n",
    "from keras.models import model_from_json\n",
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras.preprocessing.sequence import pad_sequences\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from keras.utils import np_utils\n",
    "from keras.wrappers.scikit_learn import KerasClassifier\n",
    "from keras.models import Sequential\n",
    "from keras import layers\n",
    "from keras.layers import Dense\n",
    "from keras.layers import Dropout, Dense, Embedding, LSTM, SpatialDropout1D, Input, MaxPooling1D, Flatten, GRU, Conv2D, Reshape, MaxPool2D, Concatenate, Bidirectional, GlobalAveragePooling1D, GlobalMaxPooling1D\n",
    "from tensorflow.compat.v1.keras.layers import CuDNNLSTM\n",
    "from keras.layers.merge import concatenate\n",
    "from keras.models import Model\n",
    "from keras.utils.vis_utils import plot_model\n",
    "from keras.callbacks import ModelCheckpoint\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from keras.layers import Conv1D, GlobalMaxPooling1D\n",
    "from keras.layers import Input, Dense, concatenate, Activation\n",
    "from keras.models import load_model\n",
    "from sklearn.metrics import roc_curve, auc\n",
    "from keras.callbacks import ModelCheckpoint\n",
    "import os\n",
    "import gc\n",
    "import csv\n",
    "from keras.layers import LeakyReLU\n",
    "from nltk import TweetTokenizer\n",
    "from nltk.corpus import stopwords\n",
    "from keras.preprocessing.sequence import pad_sequences\n",
    "from keras.layers import Dense, Input, LSTM, Embedding, Dropout, Activation, Conv1D, GRU,  BatchNormalization\n",
    "from tensorflow.compat.v1.keras.layers import CuDNNGRU, CuDNNLSTM\n",
    "from keras.layers import Bidirectional, GlobalMaxPool1D, MaxPooling1D, Add, Flatten, TimeDistributed\n",
    "from keras.layers import GlobalAveragePooling1D, GlobalMaxPooling1D, concatenate, SpatialDropout1D\n",
    "from keras.models import Model, load_model\n",
    "from keras import initializers, regularizers, constraints, optimizers, layers, callbacks\n",
    "from keras import backend as K\n",
    "from tensorflow.keras.layers import InputSpec, Layer\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "from keras.callbacks import ModelCheckpoint, TensorBoard, Callback, EarlyStopping\n",
    "use_gpu=True\n",
    "from keras import backend as K\n",
    "import keras.initializers\n",
    "import keras.regularizers\n",
    "import keras.constraints\n",
    "from sklearn.metrics import f1_score, classification_report, accuracy_score, confusion_matrix\n",
    "from numpy import dstack\n",
    "import tensorflow as tf\n",
    "from keras.layers.core import Lambda"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "metadata": {},
   "outputs": [],
   "source": [
    "path = 'D:/ACR-SA/Data/IMDB/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {},
   "outputs": [],
   "source": [
    "FREQ_DIST_FILE = path+'processed-freqdist.pkl'\n",
    "BI_FREQ_DIST_FILE = path+'processed-freqdist-bi.pkl'\n",
    "TRAIN_PROCESSED_FILE =  path+'train.csv'\n",
    "TEST_PROCESSED_FILE = path+'test.csv'\n",
    "GLOVE_FILE =  'E:/Pycharm/twitter-sentiment-afg/dataset/glove-seeds.txt'\n",
    "dim = 300"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "def get_glove_vectors(vocab):\n",
    "    \"\"\"\n",
    "    Extracts glove vectors from seed file only for words present in vocab.\n",
    "    \"\"\"\n",
    "    print ('Looking for GLOVE seeds')\n",
    "    glove_vectors = {}\n",
    "    found = 0\n",
    "    with open(GLOVE_FILE, 'r',encoding='utf-8') as glove_file:\n",
    "        for i, line in enumerate(glove_file):\n",
    "            utils.write_status(i + 1, 0)\n",
    "            tokens = line.strip().split()\n",
    "            word = tokens[0]\n",
    "            if vocab.get(word):\n",
    "                vector = [float(e) for e in tokens[1:]]\n",
    "                glove_vectors[word] = np.array(vector)\n",
    "                found += 1\n",
    "    print ('\\n')\n",
    "    return glove_vectors\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "def get_feature_vector(review):\n",
    "    \"\"\"\n",
    "    Generates a feature vector for each reviews where each word is\n",
    "    represented by integer index based on rank in vocabulary.\n",
    "    \"\"\"\n",
    "    words = review.split()\n",
    "    feature_vector = []\n",
    "    for i in range(len(words) - 1):\n",
    "        word = words[i]\n",
    "        if vocab.get(word) is not None:\n",
    "            feature_vector.append(vocab.get(word))\n",
    "    if len(words) >= 1:\n",
    "        if vocab.get(words[-1]) is not None:\n",
    "            feature_vector.append(vocab.get(words[-1]))\n",
    "    return feature_vector\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "def process_reviews(csv_file, test_file=True):\n",
    "    \"\"\"\n",
    "    Generates training X, y pairs.\n",
    "    \"\"\"\n",
    "    reviews = []\n",
    "    labels = []\n",
    "    print ('Generating feature vectors')\n",
    "    with open(csv_file, 'r') as csv:\n",
    "        lines = csv.readlines()\n",
    "        total = len(lines)\n",
    "        for i, line in enumerate(lines):\n",
    "            if test_file:\n",
    "                review_id, sentiment, review = line.split(',')\n",
    "            else:\n",
    "                review_id, sentiment, review = line.split(',')\n",
    "            feature_vector = get_feature_vector(review)\n",
    "            if test_file:\n",
    "                review.append(feature_vector)\n",
    "                labels.append(int(sentiment))\n",
    "            else:\n",
    "                review.append(feature_vector)\n",
    "                labels.append(int(sentiment))\n",
    "            utils.write_status(i + 1, total)\n",
    "    print ('\\n')\n",
    "    return reviews, np.array(labels)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {},
   "outputs": [],
   "source": [
    "#train = len(sys.argv) == 1\n",
    "np.random.seed(1337)\n",
    "vocab_size = 90000\n",
    "batch_size = 1000\n",
    "max_length = 200\n",
    "filters = 600\n",
    "kernel_size = 3\n",
    "vocab = utils.top_n_words(FREQ_DIST_FILE, vocab_size, shift=1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "glove_vectors = get_glove_vectors(vocab)\n",
    "reviews, labels = process_reviews(TRAIN_PROCESSED_FILE, test_file=False)\n",
    "embedding_matrix = np.random.randn(vocab_size + 1, dim) * 0.01\n",
    "# Seed it with GloVe vectors\n",
    "for word, i in vocab.items():\n",
    "    glove_vector = glove_vectors.get(word)\n",
    "    if glove_vector is not None:\n",
    "        embedding_matrix[i] = glove_vector\n",
    "reviews = pad_sequences(reviews, maxlen=max_length, padding='post')\n",
    "shuffled_indices = np.random.permutation(reviews.shape[0])\n",
    "reviews = reviews[shuffled_indices]\n",
    "labels = labels[shuffled_indices]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "  def dot_product(x, kernel):\n",
    "      if K.backend() == 'tensorflow':\n",
    "          return K.squeeze(K.dot(x, K.expand_dims(kernel)), axis=-1)\n",
    "      else:\n",
    "          return K.dot(x, kernel)\n",
    "\n",
    "  from keras.layers import Layer\n",
    "\n",
    "  class AttentionWithContext(keras.layers.Layer):\n",
    "\n",
    "    def __init__(self,\n",
    "                W_regularizer=None, u_regularizer=None, b_regularizer=None,\n",
    "                W_constraint=None, u_constraint=None, b_constraint=None,\n",
    "                bias=True, **kwargs):\n",
    "\n",
    "        self.supports_masking = True\n",
    "        self.init = keras.initializers.get('glorot_uniform')\n",
    "\n",
    "        self.W_regularizer = keras.regularizers.get(W_regularizer)\n",
    "        self.u_regularizer = keras.regularizers.get(u_regularizer)\n",
    "        self.b_regularizer = keras.regularizers.get(b_regularizer)\n",
    "\n",
    "        self.W_constraint = keras.constraints.get(W_constraint)\n",
    "        self.u_constraint = keras.constraints.get(u_constraint)\n",
    "        self.b_constraint = keras.constraints.get(b_constraint)\n",
    "\n",
    "        self.bias = bias\n",
    "        super(AttentionWithContext, self).__init__(**kwargs)\n",
    "\n",
    "    def get_config(self):\n",
    "        config = super().get_config().copy()\n",
    "        config.update({\n",
    "                'W_regularizer': self.W_regularizer,\n",
    "                'u_regularizer': self.u_regularizer,\n",
    "                'b_regularizer': self.b_regularizer,\n",
    "                'W_constraint': self.W_constraint,\n",
    "                'u_constraint': self.u_constraint,\n",
    "                'b_constraint': self.b_constraint,\n",
    "                'bias': self.bias,\n",
    "        })\n",
    "        return config\n",
    "\n",
    "    def build(self, input_shape):\n",
    "        assert len(input_shape) == 3\n",
    "\n",
    "        self.W = self.add_weight(shape=(input_shape[-1], input_shape[-1],),\n",
    "                                initializer=self.init,\n",
    "                                name='{}_W'.format(self.name),\n",
    "                                regularizer=self.W_regularizer,\n",
    "                                constraint=self.W_constraint)\n",
    "        if self.bias:\n",
    "            self.b = self.add_weight(shape=(input_shape[-1],),\n",
    "                                    initializer='zero',\n",
    "                                    name='{}_b'.format(self.name),\n",
    "                                    regularizer=self.b_regularizer,\n",
    "                                    constraint=self.b_constraint)\n",
    "\n",
    "        self.u = self.add_weight(shape=(input_shape[-1],),\n",
    "                                initializer=self.init,\n",
    "                                name='{}_u'.format(self.name),\n",
    "                                regularizer=self.u_regularizer,\n",
    "                                constraint=self.u_constraint)\n",
    "\n",
    "        super(AttentionWithContext, self).build(input_shape)\n",
    "\n",
    "    def compute_mask(self, input, input_mask=None):\n",
    "        # do not pass the mask to the next layers\n",
    "        return None\n",
    "\n",
    "    def call(self, x, mask=None):\n",
    "        uit = dot_product(x, self.W)\n",
    "\n",
    "        if self.bias:\n",
    "            uit += self.b\n",
    "\n",
    "        uit = K.tanh(uit)\n",
    "        ait = dot_product(uit, self.u)\n",
    "\n",
    "        a = K.exp(ait)\n",
    "\n",
    "        # apply mask after the exp. will be re-normalized next\n",
    "        if mask is not None:\n",
    "            # Cast the mask to floatX to avoid float64 upcasting in theano\n",
    "            a *= K.cast(mask, K.floatx())\n",
    "\n",
    "        # in some cases especially in the early stages of training the sum may be almost zero\n",
    "        # and this results in NaN's. A workaround is to add a very small positive number ε to the sum.\n",
    "        # a /= K.cast(K.sum(a, axis=1, keepdims=True), K.floatx())\n",
    "        a /= K.cast(K.sum(a, axis=1, keepdims=True) + K.epsilon(), K.floatx())\n",
    "\n",
    "        a = K.expand_dims(a)\n",
    "        weighted_input = x * a\n",
    "        return K.sum(weighted_input, axis=1)\n",
    "\n",
    "    def compute_output_shape(self, input_shape):\n",
    "        return input_shape[0], input_shape[-1]\n",
    "      \n",
    "  def createHierarchicalAttentionModel(maxSeq, embWeights=None, embeddingSize = None, vocabSize = None,wordRnnSize=100, sentenceRnnSize=100,dropWordEmb = 0.2, dropWordRnnOut = 0.2, dropSentenceRnnOut = 0.5):\t\n",
    "    '''\n",
    "    maxSeq: maximum size for sentences\n",
    "    '''\n",
    "    # Sentence level logic\n",
    "    wordInp = Input(shape=(maxSeq,),dtype='int32')\t\n",
    "    if embWeights is None:\n",
    "      x = Embedding(vocabSize, embeddingSize, input_length=maxSeq,trainable=True)(wordInp)\n",
    "    else:\n",
    "      x = Embedding(embWeights.shape[0], embWeights.shape[1], weights=[embWeights], trainable=False)(wordInp)    \n",
    "    \n",
    "    # if dropWordEmb!=0.0:\n",
    "    # \tx = Dropout(dropWordEmb)(x)\n",
    "    wordRNN = Bidirectional(CuDNNGRU(wordRnnSize,return_sequences=True))(x)\n",
    "    # if dropWordRnnOut>0.0:\n",
    "    # \twordRNN = Dropout(dropWordRnnOut)(wordRNN)\n",
    "    word_dense = TimeDistributed(Dense(200))(wordRNN)\n",
    "    word_attention = AttentionWithContext()(word_dense)\n",
    "    modelSentEncoder = Model(wordInp,word_attention)\n",
    "\n",
    "\n",
    "    # Document level logic\n",
    "    docInp = Input(shape=(1,maxSeq),dtype='int32')\t\n",
    "    sentEncoder = TimeDistributed(modelSentEncoder)(docInp)\t\n",
    "    sentRNN = Bidirectional(CuDNNGRU(sentenceRnnSize,return_sequences=True))(sentEncoder)\n",
    "    # if dropSentenceRnnOut!=0:\n",
    "    # \tsentRNN = Dropout(dropSentenceRnnOut)(sentRNN)\n",
    "    sent_dense = TimeDistributed(Dense(200))(sentRNN)\n",
    "    sent_attention = AttentionWithContext()(sent_dense)\n",
    "    docOut = Dense(1,activation='sigmoid')(sent_attention)\n",
    "    model = Model(input=docInp,output=docOut)\n",
    "    model.compile(loss='binary_crossentropy',optimizer='adam',metrics=['accuracy'])\n",
    "    plot_model(model, show_shapes=True, to_file=path+'model/IDBM-han.png')\n",
    "    return model\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generating feature vectors\n",
      "Processing 5000/5000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "x_test_new,y_test_new = process_reviews(TEST_PROCESSED_FILE, test_file=True)\n",
    "x_test_new = pad_sequences(x_test_new, maxlen=max_length, padding='post')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Our model loaded\n",
      "Our model loaded\n",
      "SS-BED model loaded\n",
      "ARC model loaded\n",
      "ABCDM model loaded\n",
      "CRNN model loaded\n",
      "IWV model loaded\n",
      "AC-BiLSTM model loaded\n",
      "LSTM model loaded\n",
      "CNN model loaded\n",
      "GloveBigruCNN model loaded\n",
      "GloveBilSTMCNN model loaded\n"
     ]
    }
   ],
   "source": [
    "\n",
    "dataset_name = 'IDBM'\n",
    "\n",
    "loaded_our_model1 = load_model(path + 'model/our_model_IMDB.hdf5', custom_objects={'AttentionWithContext': AttentionWithContext , \"tf\": tf})\n",
    "print('Our model loaded')\n",
    "print('Our model loaded')\n",
    "loaded_SSBED_model = load_model(path +  'model/SS-BED.hdf5')\n",
    "print('SS-BED model loaded')\n",
    "loaded_ARC_model = load_model(path + 'model/ARC_BiGRU.hdf5', custom_objects={'AttentionWithContext': AttentionWithContext , \"tf\": tf})\n",
    "print('ARC model loaded')\n",
    "loaded_ABCDM_model = load_model(path + 'model/ABCDM_model.hdf5', custom_objects={'AttentionWithContext': AttentionWithContext})\n",
    "print('ABCDM model loaded')\n",
    "loaded_CRNN_model = load_model(path  + 'model/CRNN.hdf5')\n",
    "print('CRNN model loaded')\n",
    "loaded_IWV_model = load_model(path  + 'model/IWV.hdf5')\n",
    "print('IWV model loaded')\n",
    "loaded_ACBiLSTM_model = load_model(path  + 'model/AC-BiLSTM.hdf5', custom_objects={'AttentionWithContext': AttentionWithContext})\n",
    "print('AC-BiLSTM model loaded')\n",
    "\n",
    "loaded_LSTM_model = load_model(path  + 'model/LSTM_APP.hdf5')\n",
    "print('LSTM model loaded')\n",
    "loaded_CNN_model = load_model(path  + 'model/CNN-Appdata.hdf5')\n",
    "print('CNN model loaded')\n",
    "\n",
    "loaded_glovegru_model = load_model(path  + 'model/GloveRGUCNN.hdf5')\n",
    "print('GloveBigruCNN model loaded')\n",
    "loaded_glovelstm_model = load_model(path  + 'model/GloveBilSTMCNN.hdf5')\n",
    "print('GloveBilSTMCNN model loaded')\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "metadata": {},
   "outputs": [],
   "source": [
    "embedding_dim = 200\n",
    "\n",
    "def build_model_ACR(lr=0.0, lr_d=0.0, units=0, spatial_dr=0.0, kernel_size1=3, kernel_size2=2, dense_units=128, dr=0.1, conv_size=64):\n",
    "    file_path = path + \"model/our_model_IMDB.hdf5\"\n",
    "    check_point = ModelCheckpoint(file_path, monitor = \"val_accuracy\", verbose = 1,\n",
    "                                  save_best_only = True, mode = \"max\")\n",
    "    early_stop = EarlyStopping(monitor = \"val_accuracy\", mode = \"max\", patience = 5)\n",
    "   \n",
    "    \n",
    "    inp = Input(shape = (review.shape[1],), name = 'our-inp')\n",
    "    inp = Input(shape = (review.shape[1],), name = 'our-inp')\n",
    "    embd1 = Embedding(embedding_matrix.shape[0], dim, weights = [embedding_matrix], trainable =True, name = 'our-embd_1')(inp)\n",
    "    embd2 = Embedding(embedding_matrix.shape[0], dim, weights = [embedding_matrix], trainable = True, name = 'our-embd_2')(inp)\n",
    "    first=GaussianNoise(0.3)(embd1)\n",
    "    second=GaussianNoise(0.3)(embd2)\n",
    "    x1 = SpatialDropout1D(spatial_dr, name = 'our-spdrop_1')(first)\n",
    "    \n",
    "    x4 = SpatialDropout1D(spatial_dr, name = 'our-spdrop_2')(second)\n",
    "    \n",
    "    x1_cnn = Conv1D(conv_size, kernel_size=2, padding='valid', kernel_initializer='he_uniform')(x1)\n",
    "    #avg_pool1_cnn = GlobalAveragePooling1D()(x1_cnn)\n",
    "    max_pool1_cnn = MaxPooling1D(pool_size=2)(x1_cnn)\n",
    "\n",
    "    x2_cnn = Conv1D(conv_size, kernel_size=3, padding='valid', kernel_initializer='he_uniform')(x1)\n",
    "    #avg_pool2_cnn = GlobalAveragePooling1D()(x2_cnn)\n",
    "    max_pool2_cnn = MaxPooling1D(pool_size=2)(x2_cnn)\n",
    "\n",
    "    x1_cnn_c = concatenate([max_pool1_cnn, max_pool2_cnn])#, x1])\n",
    "    x_lstm1 = Bidirectional(LSTM(units, return_sequences = True))(x1_cnn_c)\n",
    "    x_lstm11 = Dropout(dr)(x_lstm1)\n",
    "    x_lstm111 = AttentionWithContext()(x_lstm11)\n",
    "\n",
    "\n",
    "    x3_cnn = Conv1D(conv_size, kernel_size=2, padding='valid', kernel_initializer='he_uniform')(x4)\n",
    "  \n",
    "    max_pool3_cnn = MaxPooling1D(pool_size=2)(x3_cnn)\n",
    "\n",
    "    x4_cnn = Conv1D(conv_size, kernel_size=3, padding='valid', kernel_initializer='he_uniform')(x4)\n",
    " \n",
    "    max_pool4_cnn = MaxPooling1D(pool_size=2)(x4_cnn)\n",
    "\n",
    "    x2_cnn_c = concatenate([max_pool3_cnn, max_pool4_cnn])#, x1])\n",
    "    x_gru2 = Bidirectional(GRU(units, return_sequences = True))(x2_cnn_c)\n",
    "    x_gru22 = Dropout(dr)(x_gru2)\n",
    "    x_gru222 = AttentionWithContext()(x_gru22)\n",
    "\n",
    "\n",
    "    x = concatenate([x_gru222, x_lstm111])#, x1])\n",
    "    x = BatchNormalization()(x)\n",
    "   \n",
    "    x = Dropout(dr)(Dense(dense_units, activation='relu') (x))\n",
    "    x = BatchNormalization()(x)\n",
    "    x = Dropout(dr)(x)\n",
    "    x = Dense(int(dense_units / 2), activation='relu') (x)\n",
    "    x = Dense(1, activation = \"sigmoid\")(x)\n",
    "    model = Model(inputs = inp, outputs = x)\n",
    "    model.compile(loss = \"binary_crossentropy\", optimizer = Adam(lr = lr, decay = lr_d), metrics = [\"accuracy\"])\n",
    "    \n",
    "    \n",
    "    print(model.summary())\n",
    "    plot_model(model, show_shapes=True, to_file= path + 'model/our_model_IDBM.png')\n",
    "    history = model.fit(reviews, labels, batch_size = 128, epochs = 50,  validation_split=0.10, \n",
    "                        verbose = 1, callbacks = [check_point])\n",
    "    hist_df = pd.DataFrame(history.history)\n",
    "    hist_csv_file = path+'model/result_IMDB.csv' \n",
    "    with open(hist_csv_file, mode='w') as f:\n",
    "        hist_df.to_csv(f)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_3\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "our-inp (InputLayer)            [(None, 200)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "our-embd_2 (Embedding)          (None, 200, 200)     20000200    our-inp[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "our-embd_1 (Embedding)          (None, 200, 200)     20000200    our-inp[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_noise_7 (GaussianNoise (None, 200, 200)     0           our-embd_2[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_noise_6 (GaussianNoise (None, 200, 200)     0           our-embd_1[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "our-spdrop_2 (SpatialDropout1D) (None, 200, 200)     0           gaussian_noise_7[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "our-spdrop_1 (SpatialDropout1D) (None, 200, 200)     0           gaussian_noise_6[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_18 (Conv1D)              (None, 199, 64)      25664       our-spdrop_2[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_19 (Conv1D)              (None, 198, 64)      38464       our-spdrop_2[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_16 (Conv1D)              (None, 199, 64)      25664       our-spdrop_1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_17 (Conv1D)              (None, 198, 64)      38464       our-spdrop_1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_18 (MaxPooling1D) (None, 99, 64)       0           conv1d_18[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_19 (MaxPooling1D) (None, 99, 64)       0           conv1d_19[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_16 (MaxPooling1D) (None, 99, 64)       0           conv1d_16[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_17 (MaxPooling1D) (None, 99, 64)       0           conv1d_17[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_10 (Concatenate)    (None, 99, 128)      0           max_pooling1d_18[0][0]           \n",
      "                                                                 max_pooling1d_19[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_9 (Concatenate)     (None, 99, 128)      0           max_pooling1d_16[0][0]           \n",
      "                                                                 max_pooling1d_17[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "bidirectional_11 (Bidirectional (None, 99, 256)      198144      concatenate_10[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bidirectional_10 (Bidirectional (None, 99, 256)      263168      concatenate_9[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "dropout_17 (Dropout)            (None, 99, 256)      0           bidirectional_11[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dropout_16 (Dropout)            (None, 99, 256)      0           bidirectional_10[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "attention_with_context_7 (Atten (None, 256)          66048       dropout_17[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "attention_with_context_6 (Atten (None, 256)          66048       dropout_16[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_11 (Concatenate)    (None, 512)          0           attention_with_context_7[0][0]   \n",
      "                                                                 attention_with_context_6[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_6 (BatchNor (None, 512)          2048        concatenate_11[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dense_17 (Dense)                (None, 32)           16416       batch_normalization_6[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "dropout_18 (Dropout)            (None, 32)           0           dense_17[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_7 (BatchNor (None, 32)           128         dropout_18[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dropout_19 (Dropout)            (None, 32)           0           batch_normalization_7[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "dense_18 (Dense)                (None, 16)           528         dropout_19[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dense_19 (Dense)                (None, 1)            17          dense_18[0][0]                   \n",
      "==================================================================================================\n",
      "Total params: 40,741,201\n",
      "Trainable params: 40,740,113\n",
      "Non-trainable params: 1,088\n",
      "__________________________________________________________________________________________________\n",
      "None\n",
      "('You must install pydot (`pip install pydot`) and install graphviz (see instructions at https://graphviz.gitlab.io/download/) ', 'for plot_model/model_to_dot to work.')\n",
      "Epoch 1/50\n",
      "299/299 [==============================] - 1005s 3s/step - loss: 0.5119 - accuracy: 0.7338 - val_loss: 0.3292 - val_accuracy: 0.8590\n",
      "\n",
      "Epoch 00001: val_accuracy improved from -inf to 0.85896, saving model to D:/ACR-SA/Data/IMDB/model\\our_model_IMDBpeerj300.hdf5\n",
      "Epoch 2/50\n",
      "299/299 [==============================] - 1003s 3s/step - loss: 0.3342 - accuracy: 0.8553 - val_loss: 0.2701 - val_accuracy: 0.8867\n",
      "\n",
      "Epoch 00002: val_accuracy improved from 0.85896 to 0.88667, saving model to D:/ACR-SA/Data/IMDB/model\\our_model_IMDBpeerj300.hdf5\n",
      "Epoch 3/50\n",
      "299/299 [==============================] - 1007s 3s/step - loss: 0.2780 - accuracy: 0.8831 - val_loss: 0.3159 - val_accuracy: 0.8766\n",
      "\n",
      "Epoch 00003: val_accuracy did not improve from 0.88667\n",
      "Epoch 4/50\n",
      "299/299 [==============================] - 1010s 3s/step - loss: 0.2390 - accuracy: 0.9027 - val_loss: 0.2621 - val_accuracy: 0.9010\n",
      "\n",
      "Epoch 00004: val_accuracy improved from 0.88667 to 0.90104, saving model to D:/ACR-SA/Data/IMDB/model\\our_model_IMDBpeerj300.hdf5\n",
      "Epoch 5/50\n",
      "299/299 [==============================] - 983s 3s/step - loss: 0.2047 - accuracy: 0.9201 - val_loss: 0.2682 - val_accuracy: 0.9010\n",
      "\n",
      "Epoch 00005: val_accuracy did not improve from 0.90104\n",
      "Epoch 6/50\n",
      "299/299 [==============================] - 1010s 3s/step - loss: 0.1707 - accuracy: 0.9346 - val_loss: 0.2813 - val_accuracy: 0.9009\n",
      "\n",
      "Epoch 00006: val_accuracy did not improve from 0.90104\n",
      "Epoch 7/50\n",
      "299/299 [==============================] - 1006s 3s/step - loss: 0.1487 - accuracy: 0.9429 - val_loss: 0.3128 - val_accuracy: 0.9001\n",
      "\n",
      "Epoch 00007: val_accuracy did not improve from 0.90104\n",
      "Epoch 8/50\n",
      "299/299 [==============================] - 1020s 3s/step - loss: 0.1221 - accuracy: 0.9550 - val_loss: 0.2985 - val_accuracy: 0.9007\n",
      "\n",
      "Epoch 00008: val_accuracy did not improve from 0.90104\n",
      "Epoch 9/50\n",
      "299/299 [==============================] - 995s 3s/step - loss: 0.1068 - accuracy: 0.9610 - val_loss: 0.3065 - val_accuracy: 0.8967\n",
      "\n",
      "Epoch 00009: val_accuracy did not improve from 0.90104\n",
      "Epoch 10/50\n",
      "299/299 [==============================] - 979s 3s/step - loss: 0.0937 - accuracy: 0.9653 - val_loss: 0.3429 - val_accuracy: 0.8981\n",
      "\n",
      "Epoch 00010: val_accuracy did not improve from 0.90104\n",
      "Epoch 11/50\n",
      "299/299 [==============================] - 980s 3s/step - loss: 0.0800 - accuracy: 0.9716 - val_loss: 0.3912 - val_accuracy: 0.8919\n",
      "\n",
      "Epoch 00011: val_accuracy did not improve from 0.90104\n",
      "Epoch 12/50\n",
      "299/299 [==============================] - 985s 3s/step - loss: 0.0655 - accuracy: 0.9768 - val_loss: 0.4104 - val_accuracy: 0.8895\n",
      "\n",
      "Epoch 00012: val_accuracy did not improve from 0.90104\n",
      "Epoch 13/50\n",
      "299/299 [==============================] - 1007s 3s/step - loss: 0.0605 - accuracy: 0.9780 - val_loss: 0.3887 - val_accuracy: 0.8911\n",
      "\n",
      "Epoch 00013: val_accuracy did not improve from 0.90104\n",
      "Epoch 14/50\n",
      "299/299 [==============================] - 1017s 3s/step - loss: 0.0526 - accuracy: 0.9819 - val_loss: 0.4215 - val_accuracy: 0.8902\n",
      "\n",
      "Epoch 00014: val_accuracy did not improve from 0.90104\n",
      "Epoch 15/50\n",
      "299/299 [==============================] - 1031s 3s/step - loss: 0.0497 - accuracy: 0.9821 - val_loss: 0.5049 - val_accuracy: 0.8908\n",
      "\n",
      "Epoch 00015: val_accuracy did not improve from 0.90104\n",
      "Epoch 16/50\n",
      "299/299 [==============================] - 1031s 3s/step - loss: 0.0440 - accuracy: 0.9846 - val_loss: 0.5148 - val_accuracy: 0.8812\n",
      "\n",
      "Epoch 00016: val_accuracy did not improve from 0.90104\n",
      "Epoch 17/50\n",
      "299/299 [==============================] - 999s 3s/step - loss: 0.0396 - accuracy: 0.9858 - val_loss: 0.6230 - val_accuracy: 0.8744\n",
      "\n",
      "Epoch 00017: val_accuracy did not improve from 0.90104\n",
      "Epoch 18/50\n",
      "299/299 [==============================] - 1000s 3s/step - loss: 0.0368 - accuracy: 0.9876 - val_loss: 0.5182 - val_accuracy: 0.8824\n",
      "\n",
      "Epoch 00018: val_accuracy did not improve from 0.90104\n",
      "Epoch 19/50\n",
      "299/299 [==============================] - 998s 3s/step - loss: 0.0310 - accuracy: 0.9893 - val_loss: 0.5504 - val_accuracy: 0.8865\n",
      "\n",
      "Epoch 00019: val_accuracy did not improve from 0.90104\n",
      "Epoch 20/50\n",
      "299/299 [==============================] - 997s 3s/step - loss: 0.0308 - accuracy: 0.9884 - val_loss: 0.5812 - val_accuracy: 0.8822\n",
      "\n",
      "Epoch 00020: val_accuracy did not improve from 0.90104\n",
      "Epoch 21/50\n",
      "299/299 [==============================] - 996s 3s/step - loss: 0.0277 - accuracy: 0.9902 - val_loss: 0.6793 - val_accuracy: 0.8753\n",
      "\n",
      "Epoch 00021: val_accuracy did not improve from 0.90104\n",
      "Epoch 22/50\n",
      "299/299 [==============================] - 992s 3s/step - loss: 0.0251 - accuracy: 0.9911 - val_loss: 0.6546 - val_accuracy: 0.8841\n",
      "\n",
      "Epoch 00022: val_accuracy did not improve from 0.90104\n",
      "Epoch 23/50\n",
      "299/299 [==============================] - 992s 3s/step - loss: 0.0241 - accuracy: 0.9915 - val_loss: 0.6258 - val_accuracy: 0.8825\n",
      "\n",
      "Epoch 00023: val_accuracy did not improve from 0.90104\n",
      "Epoch 24/50\n",
      "299/299 [==============================] - 994s 3s/step - loss: 0.0229 - accuracy: 0.9922 - val_loss: 0.6903 - val_accuracy: 0.8831\n",
      "\n",
      "Epoch 00024: val_accuracy did not improve from 0.90104\n",
      "Epoch 25/50\n",
      "299/299 [==============================] - 993s 3s/step - loss: 0.0201 - accuracy: 0.9922 - val_loss: 0.6717 - val_accuracy: 0.8830\n",
      "\n",
      "Epoch 00025: val_accuracy did not improve from 0.90104\n",
      "Epoch 26/50\n",
      "299/299 [==============================] - 994s 3s/step - loss: 0.0213 - accuracy: 0.9928 - val_loss: 0.6113 - val_accuracy: 0.8870\n",
      "\n",
      "Epoch 00026: val_accuracy did not improve from 0.90104\n",
      "Epoch 27/50\n",
      "299/299 [==============================] - 995s 3s/step - loss: 0.0187 - accuracy: 0.9942 - val_loss: 0.6551 - val_accuracy: 0.8815\n",
      "\n",
      "Epoch 00027: val_accuracy did not improve from 0.90104\n",
      "Epoch 28/50\n",
      "299/299 [==============================] - 994s 3s/step - loss: 0.0194 - accuracy: 0.9928 - val_loss: 0.6726 - val_accuracy: 0.8852\n",
      "\n",
      "Epoch 00028: val_accuracy did not improve from 0.90104\n",
      "Epoch 29/50\n",
      "299/299 [==============================] - 994s 3s/step - loss: 0.0165 - accuracy: 0.9945 - val_loss: 0.7197 - val_accuracy: 0.8769\n",
      "\n",
      "Epoch 00029: val_accuracy did not improve from 0.90104\n",
      "Epoch 30/50\n",
      "299/299 [==============================] - 994s 3s/step - loss: 0.0153 - accuracy: 0.9949 - val_loss: 0.6977 - val_accuracy: 0.8834\n",
      "\n",
      "Epoch 00030: val_accuracy did not improve from 0.90104\n",
      "Epoch 31/50\n",
      "299/299 [==============================] - 994s 3s/step - loss: 0.0149 - accuracy: 0.9946 - val_loss: 0.7778 - val_accuracy: 0.8821\n",
      "\n",
      "Epoch 00031: val_accuracy did not improve from 0.90104\n",
      "Epoch 32/50\n",
      "299/299 [==============================] - 995s 3s/step - loss: 0.0155 - accuracy: 0.9946 - val_loss: 0.7512 - val_accuracy: 0.8801\n",
      "\n",
      "Epoch 00032: val_accuracy did not improve from 0.90104\n",
      "Epoch 33/50\n",
      "299/299 [==============================] - 994s 3s/step - loss: 0.0128 - accuracy: 0.9954 - val_loss: 0.7659 - val_accuracy: 0.8790\n",
      "\n",
      "Epoch 00033: val_accuracy did not improve from 0.90104\n",
      "Epoch 34/50\n",
      "299/299 [==============================] - 994s 3s/step - loss: 0.0143 - accuracy: 0.9948 - val_loss: 0.8687 - val_accuracy: 0.8674\n",
      "\n",
      "Epoch 00034: val_accuracy did not improve from 0.90104\n",
      "Epoch 35/50\n",
      "299/299 [==============================] - 993s 3s/step - loss: 0.0129 - accuracy: 0.9956 - val_loss: 0.8238 - val_accuracy: 0.8781\n",
      "\n",
      "Epoch 00035: val_accuracy did not improve from 0.90104\n",
      "Epoch 36/50\n",
      "299/299 [==============================] - 989s 3s/step - loss: 0.0134 - accuracy: 0.9957 - val_loss: 0.7830 - val_accuracy: 0.8796\n",
      "\n",
      "Epoch 00036: val_accuracy did not improve from 0.90104\n",
      "Epoch 37/50\n",
      "299/299 [==============================] - 988s 3s/step - loss: 0.0126 - accuracy: 0.9955 - val_loss: 0.8798 - val_accuracy: 0.8698\n",
      "\n",
      "Epoch 00037: val_accuracy did not improve from 0.90104\n",
      "Epoch 38/50\n",
      "299/299 [==============================] - 989s 3s/step - loss: 0.0131 - accuracy: 0.9952 - val_loss: 0.8192 - val_accuracy: 0.8773\n",
      "\n",
      "Epoch 00038: val_accuracy did not improve from 0.90104\n",
      "Epoch 39/50\n",
      "299/299 [==============================] - 993s 3s/step - loss: 0.0115 - accuracy: 0.9959 - val_loss: 0.8756 - val_accuracy: 0.8723\n",
      "\n",
      "Epoch 00039: val_accuracy did not improve from 0.90104\n",
      "Epoch 40/50\n",
      "299/299 [==============================] - 989s 3s/step - loss: 0.0116 - accuracy: 0.9959 - val_loss: 0.8391 - val_accuracy: 0.8756\n",
      "\n",
      "Epoch 00040: val_accuracy did not improve from 0.90104\n",
      "Epoch 41/50\n",
      "299/299 [==============================] - 988s 3s/step - loss: 0.0118 - accuracy: 0.9961 - val_loss: 0.8343 - val_accuracy: 0.8730\n",
      "\n",
      "Epoch 00041: val_accuracy did not improve from 0.90104\n",
      "Epoch 42/50\n",
      "299/299 [==============================] - 990s 3s/step - loss: 0.0091 - accuracy: 0.9970 - val_loss: 0.8912 - val_accuracy: 0.8748\n",
      "\n",
      "Epoch 00042: val_accuracy did not improve from 0.90104\n",
      "Epoch 43/50\n",
      "299/299 [==============================] - 989s 3s/step - loss: 0.0107 - accuracy: 0.9962 - val_loss: 0.8235 - val_accuracy: 0.8779\n",
      "\n",
      "Epoch 00043: val_accuracy did not improve from 0.90104\n",
      "Epoch 44/50\n",
      "299/299 [==============================] - 990s 3s/step - loss: 0.0099 - accuracy: 0.9967 - val_loss: 0.8701 - val_accuracy: 0.8773\n",
      "\n",
      "Epoch 00044: val_accuracy did not improve from 0.90104\n",
      "Epoch 45/50\n",
      "299/299 [==============================] - 989s 3s/step - loss: 0.0089 - accuracy: 0.9971 - val_loss: 0.8865 - val_accuracy: 0.8757\n",
      "\n",
      "Epoch 00045: val_accuracy did not improve from 0.90104\n",
      "Epoch 46/50\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "299/299 [==============================] - 989s 3s/step - loss: 0.0076 - accuracy: 0.9974 - val_loss: 0.9132 - val_accuracy: 0.8790\n",
      "\n",
      "Epoch 00046: val_accuracy did not improve from 0.90104\n",
      "Epoch 47/50\n",
      "299/299 [==============================] - 989s 3s/step - loss: 0.0101 - accuracy: 0.9965 - val_loss: 0.8028 - val_accuracy: 0.8770\n",
      "\n",
      "Epoch 00047: val_accuracy did not improve from 0.90104\n",
      "Epoch 48/50\n",
      "299/299 [==============================] - 988s 3s/step - loss: 0.0077 - accuracy: 0.9971 - val_loss: 0.8545 - val_accuracy: 0.8800\n",
      "\n",
      "Epoch 00048: val_accuracy did not improve from 0.90104\n",
      "Epoch 49/50\n",
      "299/299 [==============================] - 989s 3s/step - loss: 0.0088 - accuracy: 0.9969 - val_loss: 0.8197 - val_accuracy: 0.8806\n",
      "\n",
      "Epoch 00049: val_accuracy did not improve from 0.90104\n",
      "Epoch 50/50\n",
      "299/299 [==============================] - 990s 3s/step - loss: 0.0086 - accuracy: 0.9969 - val_loss: 0.8090 - val_accuracy: 0.8809\n",
      "\n",
      "Epoch 00050: val_accuracy did not improve from 0.90104\n"
     ]
    }
   ],
   "source": [
    "max_len = 200\n",
    "model = build_model_ACR(lr = 1e-3, lr_d = 1e-10, units = 128, spatial_dr = 0.5, kernel_size1=2, kernel_size2=3, dense_units=32, dr=0.2, conv_size=64)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generating feature vectors\n",
      "Processing 5000/5000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "x_test_new,y_test_new = process_reviews(TEST_PROCESSED_FILE, test_file=True)\n",
    "x_test_new = pad_sequences(x_test_new, maxlen=max_length, padding='post')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Our model loaded\n"
     ]
    }
   ],
   "source": [
    "\n",
    "print('Our model loaded')\n",
    "loaded_acr_model2 = load_model('D:/ACR-SA/Data/IMDB/model/our_model_IMDB.hdf5', custom_objects={'AttentionWithContext': AttentionWithContext , \"tf\": tf})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "our model complete!\n"
     ]
    }
   ],
   "source": [
    "yhat_acr = loaded_acr_model2.predict(x_test_new)\n",
    "print('our model complete!')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 648x576 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "fpr_aclnew, tpr_aclnew, threshold = roc_curve(y_test, yhat_aclupdate)\n",
    "roc_auc_aclnew = auc(fpr_aclnew, tpr_aclnew)\n",
    "\n",
    "fpr_our2, tpr_our2, threshold2 = roc_curve(y_test_new, yhat_acr)\n",
    "roc_auc_our2 = auc(fpr_our2, tpr_our2)\n",
    "\n",
    "fpr_ssbed, tpr_ssbed, threshold = roc_curve(y_test, yhat_ssbed)\n",
    "roc_auc_ssbed = auc(fpr_ssbed, tpr_ssbed)\n",
    "\n",
    "fpr_arc, tpr_arc, threshold = roc_curve(y_test, yhat_arc)\n",
    "roc_auc_arc = auc(fpr_arc, tpr_arc)\n",
    "\n",
    "fpr_abcdm, tpr_abcdm, threshold = roc_curve(y_test, yhat_abcdm)\n",
    "roc_auc_abcdm = auc(fpr_abcdm, tpr_abcdm)\n",
    "\n",
    "fpr_crn, tpr_crn, threshold = roc_curve(y_test, yhat_crnn)\n",
    "roc_auc_crn = auc(fpr_crn, tpr_crn)\n",
    "\n",
    "fpr_iwv, tpr_iwv, threshold = roc_curve(y_test, yhat_iwv)\n",
    "roc_auc_iwv = auc(fpr_iwv, tpr_iwv)\n",
    "\n",
    "fpr_acbilstm, tpr_acbilstm, threshold = roc_curve(y_test, yhat_acbilstm)\n",
    "roc_auc_acbilstm = auc(fpr_acbilstm, tpr_acbilstm)\n",
    "\n",
    "fpr_lstm, tpr_lstm, threshold = roc_curve(y_test, yhat_lstm)\n",
    "roc_auc_lstm = auc(fpr_lstm, tpr_lstm)\n",
    "\n",
    "fpr_cnn, tpr_cnn, threshold = roc_curve(y_test, yhat_cnn)\n",
    "roc_auc_cnn = auc(fpr_cnn, tpr_cnn)\n",
    "\n",
    "fpr_gru, tpr_gru, threshold = roc_curve(y_test, yhat_gru)\n",
    "roc_auc_gru = auc(fpr_gru, tpr_gru)\n",
    "\n",
    "fpr_bilstm, tpr_bilstm, threshold = roc_curve(y_test, yhat_bilstm)\n",
    "roc_auc_bilstm = auc(fpr_bilstm, tpr_bilstm)\n",
    "\n",
    "#############################3333\n",
    "fpr_stacked, tpr_stacked, threshold = roc_curve(y_test, yhat_our2)\n",
    "roc_auc_stacked = auc(fpr_stacked, tpr_stacked)\n",
    "#################################3\n",
    "fig = plt.figure(figsize=(9,8))\n",
    "ax = fig.add_subplot(111)\n",
    "plt.gcf().set_facecolor('white')\n",
    "\n",
    "ax.plot(fpr_our2, tpr_our2, label='ACR-SA  = %0.4f' % roc_auc_our2, linewidth=2.3, color = 'red')\n",
    "ax.plot(fpr_aclnew, tpr_aclnew, label='ACL  = %0.4f' % roc_auc_aclnew, linewidth=2.3, color = '#005700')\n",
    "ax.plot(fpr_ssbed, tpr_ssbed, label='SS-BED = %0.4f' % roc_auc_ssbed, linewidth=2)\n",
    "ax.plot(fpr_arc, tpr_arc, label='ARC  = %0.4f' % roc_auc_arc, linewidth=2, color = '#0000cc')\n",
    "ax.plot(fpr_abcdm, tpr_abcdm, label='ABCDM  = %0.4f' % roc_auc_abcdm, linewidth=2, color = 'black')\n",
    "ax.plot(fpr_crn, tpr_crn, label='CRNN = %0.4f' % roc_auc_crn, linewidth=2, color = '#8c1aff')\n",
    "ax.plot(fpr_iwv, tpr_iwv, label='IWV = %0.4f' % roc_auc_iwv, linewidth=2, color = '#39e600')\n",
    "ax.plot(fpr_acbilstm, tpr_acbilstm, label='AC-BiLSTM  = %0.4f' % roc_auc_acbilstm, linewidth=2, color = '#ac3939')\n",
    "ax.plot(fpr_cnn, tpr_cnn, label='Word embeeding-CNN  = %0.4f' % roc_auc_cnn, linewidth=2, color = '#cccc00')\n",
    "ax.plot(fpr_gru, tpr_gru, label='Glove-BiGRU-CNN  = %0.4f' % roc_auc_gru, linewidth=2, color = '#c94f8a')\n",
    "ax.plot(fpr_bilstm, tpr_bilstm, label='Glove-BiLSTM-CNN  = %0.4f' % roc_auc_bilstm, linewidth=2, color = '#c98c4f')\n",
    "\n",
    "#ax.plot([0, 1], [0, 1], 'k--', linewidth=2)\n",
    "ax.set_facecolor('white')\n",
    "ax.set_xlabel('False Positive Rate', fontsize=18)\n",
    "ax.set_ylabel('True Positive Rate')\n",
    "ax.set_title('')\n",
    "ax.legend(loc='lower left',  ncol = 2,bbox_to_anchor=(0.05, 0.0, 0.0, 0.0),framealpha= 0.4)\n",
    "\n",
    "\n",
    "axins = zoomed_inset_axes(ax, 1, loc = 'center right') # zoom-factor: 2.5, location: upper-left\n",
    "\n",
    "\n",
    "axins.plot(fpr_our2, tpr_our2, label='ACR-SA  = %0.4f' % roc_auc_our2, linewidth=2.3, color = 'red')\n",
    "axins.plot(fpr_aclnew, tpr_aclnew, label='ACL (area = %0.3f' % roc_auc_aclnew, linewidth=2, color = '#005700')\n",
    "axins.plot(fpr_ssbed, tpr_ssbed, label='SS-BED = %0.3f' % roc_auc_ssbed, linewidth=2)\n",
    "axins.plot(fpr_arc, tpr_arc, label='ARC (area = %0.3f' % roc_auc_arc, linewidth=2, color = '#0000cc')\n",
    "axins.plot(fpr_abcdm, tpr_abcdm, label='ABCDM = %0.3f' % roc_auc_abcdm, linewidth=2, color = 'black')\n",
    "axins.plot(fpr_crn, tpr_crn, label='CRNN (area = %0.3f' % roc_auc_crn, linewidth=2, color = '#8c1aff')\n",
    "axins.plot(fpr_iwv, tpr_iwv, label='IWV  = %0.3f' % roc_auc_iwv, linewidth=2, color = '#39e600')\n",
    "axins.plot(fpr_acbilstm, tpr_acbilstm, label='AC-BiLSTM  = %0.3f' % roc_auc_acbilstm, linewidth=2, color = '#ac3939')\n",
    "axins.plot(fpr_cnn, tpr_cnn, label='Word embeeding-CNN  = %0.3f' % roc_auc_cnn, linewidth=2, color = '#cccc00')\n",
    "axins.plot(fpr_gru, tpr_gru, label='Glove-BiGRU-CNN  = %0.4f' % roc_auc_gru, linewidth=2, color = '#c94f8a')\n",
    "axins.plot(fpr_bilstm, tpr_bilstm, label='Glove-BiLSTM-CNN  = %0.4f' % roc_auc_bilstm, linewidth=2, color = '#c98c4f')\n",
    "\n",
    "axins.patch.set_facecolor('#d9d9d9')\n",
    "axins.patch.set_alpha(0.3)\n",
    "\n",
    "axins.spines['bottom'].set_color('0.7')\n",
    "axins.spines['bottom'].set_linewidth('1')\n",
    "axins.spines['bottom'].set_linestyle('--')\n",
    "axins.spines['top'].set_color('0.7')\n",
    "axins.spines['top'].set_linewidth('1')\n",
    "axins.spines['top'].set_linestyle('--')\n",
    "axins.spines['right'].set_color('0.7')\n",
    "axins.spines['right'].set_linewidth('1')\n",
    "axins.spines['right'].set_linestyle('--')\n",
    "axins.spines['left'].set_color('0.7')\n",
    "axins.spines['left'].set_linewidth('1') \n",
    "axins.spines['left'].set_linestyle('--')  \n",
    "\n",
    "x1, x2, y1, y2 = 0, 0.3, 0.7, 1 # specify the limits\n",
    "#x1, x2, y1, y2 = 0.2, 0.5, 0.6, 0.9 # specify the limits\n",
    "\n",
    "axins.set_xlim(x1, x2) # apply the x-limits\n",
    "axins.set_ylim(y1, y2) # apply the y-limits\n",
    "\n",
    "plt.yticks(visible=False)\n",
    "plt.xticks(visible=False)\n",
    "plt.grid(visible=False)\n",
    "\n",
    "from mpl_toolkits.axes_grid1.inset_locator import mark_inset\n",
    "mark_inset(ax, axins, loc1=1, loc2=3, fc=\"#d9d9d9\", ec=\"0.3\", alpha = 0.3, ls = '-.')\n",
    "\n",
    "plt.savefig(path+ dataset_name + '-AUC7_ibdm12PJ2forusereal300.png',dpi=300,bbox_inches = 'tight',pad_inches = 0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\t Recall, Specifity, Precision, F1, Accuracy\n",
      "ACR - :  [0.8967, 0.9045, 0.9028, 0.8997, 0.9006]\n",
      "ACR + :  [0.9045, 0.8967, 0.8984, 0.9014, 0.9006]\n"
     ]
    }
   ],
   "source": [
    "classifiers = [yhat_acr]\n",
    "classifiers_name = [ 'ACR -','ACR +']#SS-BED', 'CNN']#\n",
    "dataset_name = 'airline'\n",
    "\n",
    "y_test_list = y_test_new.tolist()\n",
    "results = []\n",
    "for c,i in zip(classifiers,range(len(classifiers))):\n",
    "  labels = (c > 0.5).astype(np.int)\n",
    "  labels = labels.flatten().tolist()\n",
    "  \n",
    "  data = []\n",
    "  cnf_matrix = confusion_matrix(y_test_list, labels)\n",
    "\n",
    "  FP = cnf_matrix.sum(axis=0) - np.diag(cnf_matrix)  \n",
    "  FN = cnf_matrix.sum(axis=1) - np.diag(cnf_matrix)\n",
    "  TP = np.diag(cnf_matrix)\n",
    "  TN = cnf_matrix.sum() - (FP + FN + TP)\n",
    "  \n",
    "\n",
    "  FP = FP.astype(float)\n",
    "  FN = FN.astype(float)\n",
    "  TP = TP.astype(float)\n",
    "  TN = TN.astype(float)\n",
    "\n",
    "  # Sensitivity, hit rate, recall, or true positive rate\n",
    "  TPR = TP/(TP+FN)\n",
    "  # Specificity or true negative rate\n",
    "  TNR = TN/(TN+FP) \n",
    "  # Precision or positive predictive value\n",
    "  PPV = TP/(TP+FP)\n",
    "  # Negative predictive value\n",
    "  NPV = TN/(TN+FN)\n",
    "  # Fall out or false positive rate\n",
    "  FPR = FP/(FP+TN)\n",
    "  # False negative rate\n",
    "  FNR = FN/(TP+FN)\n",
    "  # False discovery rate\n",
    "  FDR = FP/(TP+FP)\n",
    "  # Overall accuracy\n",
    "  ACC = (TP+TN)/(TP+FP+FN+TN)\n",
    "  F1 = (2 * TPR * PPV)/(TPR + PPV)\n",
    "  data.append(TPR)\n",
    "  data.append(TNR)\n",
    "  data.append(PPV)\n",
    "  data.append(F1)\n",
    "  data.append(ACC)\n",
    "  \n",
    "  results.append([round(d[0],4) for d in data])\n",
    "  results.append([round(d[1],4) for d in data])\n",
    "\n",
    "print(\"\\t Recall, Specifity, Precision, F1, Accuracy\")\n",
    "for r,name in zip(results,classifiers_name):\n",
    "  print(name,': ',r)\n",
    "with open(path+\"Results10new2002.txt\", \"w\") as filehandle:  \n",
    "    filehandle.writelines(f\"{name} %s\\n\" %r for r,name in zip(results,classifiers_name))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {},
   "outputs": [],
   "source": [
    "path = 'D:/ACR-SA/Data/IMDB/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {},
   "outputs": [],
   "source": [
    "SEED = 2000\n",
    "\n",
    "csv = path+'model/result_IMDB.csv' \n",
    "\n",
    "\n",
    "my_df = pd.read_csv(csv, encoding = 'latin-1')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 203,
   "metadata": {},
   "outputs": [],
   "source": [
    "embedding_dim = 200\n",
    "\n",
    "def build_model_ACR(lr=0.0, lr_d=0.0, units=0, spatial_dr=0.0, kernel_size1=3, kernel_size2=2, dense_units=128, dr=0.1, conv_size=64):\n",
    "    file_path = path + \"model/our_model_IMDB-GN.hdf5\"\n",
    "    check_point = ModelCheckpoint(file_path, monitor = \"val_accuracy\", verbose = 1,\n",
    "                                  save_best_only = True, mode = \"max\")\n",
    "    early_stop = EarlyStopping(monitor = \"val_accuracy\", mode = \"max\", patience = 5)\n",
    "   \n",
    "    \n",
    "    inp = Input(shape = (reviews.shape[1],), name = 'our-inp')\n",
    "    inp = Input(shape = (reviews.shape[1],), name = 'our-inp')\n",
    "    embd1 = Embedding(embedding_matrix.shape[0], dim, weights = [embedding_matrix], trainable =True, name = 'our-embd_1')(inp)\n",
    "    embd2 = Embedding(embedding_matrix.shape[0], dim, weights = [embedding_matrix], trainable = True, name = 'our-embd_2')(inp)\n",
    "    first=GaussianNoise(0.3)(embd1)\n",
    "    second=GaussianNoise(0.3)(embd2)\n",
    "    x1 = SpatialDropout1D(spatial_dr, name = 'our-spdrop_1')(first)\n",
    "    \n",
    "    x4 = SpatialDropout1D(spatial_dr, name = 'our-spdrop_2')(second)\n",
    "    \n",
    "    x1_cnn = Conv1D(conv_size, kernel_size=2, padding='valid', kernel_initializer='he_uniform')(embd1)\n",
    "\n",
    "    max_pool1_cnn = MaxPooling1D(pool_size=2)(x1_cnn)\n",
    "\n",
    "    x2_cnn = Conv1D(conv_size, kernel_size=3, padding='valid', kernel_initializer='he_uniform')(embd1)\n",
    "    max_pool2_cnn = MaxPooling1D(pool_size=2)(x2_cnn)\n",
    "\n",
    "    x1_cnn_c = concatenate([max_pool1_cnn, max_pool2_cnn])#, x1])\n",
    "    x_lstm1 = Bidirectional(LSTM(units, return_sequences = True))(x1_cnn_c)\n",
    "    x_lstm11 = Dropout(dr)(x_lstm1)\n",
    "    x_lstm111 = AttentionWithContext()(x_lstm11)\n",
    "\n",
    "\n",
    "    x3_cnn = Conv1D(conv_size, kernel_size=2, padding='valid', kernel_initializer='he_uniform')(embd2)\n",
    "    max_pool3_cnn = MaxPooling1D(pool_size=2)(x3_cnn)\n",
    "\n",
    "    x4_cnn = Conv1D(conv_size, kernel_size=3, padding='valid', kernel_initializer='he_uniform')(embd2)\n",
    "    max_pool4_cnn = MaxPooling1D(pool_size=2)(x4_cnn)\n",
    "\n",
    "    x2_cnn_c = concatenate([max_pool3_cnn, max_pool4_cnn])#, x1])\n",
    "    x_gru2 = Bidirectional(GRU(units, return_sequences = True))(x2_cnn_c)\n",
    "    x_gru22 = Dropout(dr)(x_gru2)\n",
    "    x_gru222 = AttentionWithContext()(x_gru22)\n",
    "\n",
    "\n",
    "    x = concatenate([x_gru222, x_lstm111])#, x1])\n",
    "    x = BatchNormalization()(x)\n",
    "   \n",
    "    x = Dropout(dr)(Dense(dense_units, activation='relu') (x))\n",
    "    x = BatchNormalization()(x)\n",
    "    x = Dropout(dr)(x)\n",
    "    x = Dense(int(dense_units / 2), activation='relu') (x)\n",
    "    x = Dense(1, activation = \"sigmoid\")(x)\n",
    "    model = Model(inputs = inp, outputs = x)\n",
    "    model.compile(loss = \"binary_crossentropy\", optimizer = Adam(lr = lr, decay = lr_d), metrics = [\"accuracy\"])\n",
    "  \n",
    "    \n",
    "    print(model.summary())\n",
    "    plot_model(model, show_shapes=True, to_file= path + 'model/our_model.png')\n",
    "    history = model.fit(reviews, labels, batch_size = 128, epochs = 50,  validation_split=0.10, \n",
    "                        verbose = 1, callbacks = [check_point])\n",
    "    hist_df = pd.DataFrame(history.history)\n",
    "    hist_csv_file = path+'model/our_model_result_IMDB-GN.csv' \n",
    "    with open(hist_csv_file, mode='w') as f:\n",
    "        hist_df.to_csv(f)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\kamyab\\anaconda3\\lib\\site-packages\\keras\\optimizer_v2\\optimizer_v2.py:355: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_5\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "our-inp (InputLayer)            [(None, 200)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "our-embd_2 (Embedding)          (None, 200, 200)     20000200    our-inp[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "our-embd_1 (Embedding)          (None, 200, 200)     20000200    our-inp[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_26 (Conv1D)              (None, 199, 64)      25664       our-embd_2[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_27 (Conv1D)              (None, 198, 64)      38464       our-embd_2[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_24 (Conv1D)              (None, 199, 64)      25664       our-embd_1[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_25 (Conv1D)              (None, 198, 64)      38464       our-embd_1[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_26 (MaxPooling1D) (None, 99, 64)       0           conv1d_26[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_27 (MaxPooling1D) (None, 99, 64)       0           conv1d_27[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_24 (MaxPooling1D) (None, 99, 64)       0           conv1d_24[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_25 (MaxPooling1D) (None, 99, 64)       0           conv1d_25[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_16 (Concatenate)    (None, 99, 128)      0           max_pooling1d_26[0][0]           \n",
      "                                                                 max_pooling1d_27[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_15 (Concatenate)    (None, 99, 128)      0           max_pooling1d_24[0][0]           \n",
      "                                                                 max_pooling1d_25[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "bidirectional_15 (Bidirectional (None, 99, 256)      198144      concatenate_16[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bidirectional_14 (Bidirectional (None, 99, 256)      263168      concatenate_15[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dropout_25 (Dropout)            (None, 99, 256)      0           bidirectional_15[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dropout_24 (Dropout)            (None, 99, 256)      0           bidirectional_14[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "attention_with_context_11 (Atte (None, 256)          66048       dropout_25[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "attention_with_context_10 (Atte (None, 256)          66048       dropout_24[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_17 (Concatenate)    (None, 512)          0           attention_with_context_11[0][0]  \n",
      "                                                                 attention_with_context_10[0][0]  \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_10 (BatchNo (None, 512)          2048        concatenate_17[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dense_23 (Dense)                (None, 32)           16416       batch_normalization_10[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "dropout_26 (Dropout)            (None, 32)           0           dense_23[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_11 (BatchNo (None, 32)           128         dropout_26[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dropout_27 (Dropout)            (None, 32)           0           batch_normalization_11[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "dense_24 (Dense)                (None, 16)           528         dropout_27[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dense_25 (Dense)                (None, 1)            17          dense_24[0][0]                   \n",
      "==================================================================================================\n",
      "Total params: 40,741,201\n",
      "Trainable params: 40,740,113\n",
      "Non-trainable params: 1,088\n",
      "__________________________________________________________________________________________________\n",
      "None\n",
      "('You must install pydot (`pip install pydot`) and install graphviz (see instructions at https://graphviz.gitlab.io/download/) ', 'for plot_model/model_to_dot to work.')\n",
      "Epoch 1/50\n",
      "299/299 [==============================] - 1404s 5s/step - loss: 0.3569 - accuracy: 0.8419 - val_loss: 0.3575 - val_accuracy: 0.8621\n",
      "\n",
      "Epoch 00001: val_accuracy improved from -inf to 0.86207, saving model to D:/ACR-SA/Data/IMDB/model\\our_model_IMDBpeerjoutGN.hdf5\n",
      "Epoch 2/50\n",
      "299/299 [==============================] - 1413s 5s/step - loss: 0.2084 - accuracy: 0.9180 - val_loss: 0.2601 - val_accuracy: 0.8930\n",
      "\n",
      "Epoch 00002: val_accuracy improved from 0.86207 to 0.89304, saving model to D:/ACR-SA/Data/IMDB/model\\our_model_IMDBpeerjoutGN.hdf5\n",
      "Epoch 3/50\n",
      "299/299 [==============================] - 1419s 5s/step - loss: 0.1005 - accuracy: 0.9651 - val_loss: 0.3501 - val_accuracy: 0.8681\n",
      "\n",
      "Epoch 00003: val_accuracy did not improve from 0.89304\n",
      "Epoch 4/50\n",
      "299/299 [==============================] - 1429s 5s/step - loss: 0.0457 - accuracy: 0.9850 - val_loss: 0.7128 - val_accuracy: 0.7999\n",
      "\n",
      "Epoch 00004: val_accuracy did not improve from 0.89304\n",
      "Epoch 5/50\n",
      "299/299 [==============================] - 1415s 5s/step - loss: 0.0283 - accuracy: 0.9905 - val_loss: 0.6722 - val_accuracy: 0.8473\n",
      "\n",
      "Epoch 00005: val_accuracy did not improve from 0.89304\n",
      "Epoch 6/50\n",
      "299/299 [==============================] - 1401s 5s/step - loss: 0.0186 - accuracy: 0.9943 - val_loss: 0.8081 - val_accuracy: 0.8419\n",
      "\n",
      "Epoch 00006: val_accuracy did not improve from 0.89304\n",
      "Epoch 7/50\n",
      "299/299 [==============================] - 1416s 5s/step - loss: 0.0212 - accuracy: 0.9930 - val_loss: 0.5716 - val_accuracy: 0.8750\n",
      "\n",
      "Epoch 00007: val_accuracy did not improve from 0.89304\n",
      "Epoch 8/50\n",
      "299/299 [==============================] - 1441s 5s/step - loss: 0.0155 - accuracy: 0.9948 - val_loss: 0.5770 - val_accuracy: 0.8804\n",
      "\n",
      "Epoch 00008: val_accuracy did not improve from 0.89304\n",
      "Epoch 9/50\n",
      "299/299 [==============================] - 1440s 5s/step - loss: 0.0133 - accuracy: 0.9956 - val_loss: 0.5765 - val_accuracy: 0.8750\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Epoch 00009: val_accuracy did not improve from 0.89304\n",
      "Epoch 10/50\n",
      "299/299 [==============================] - 1445s 5s/step - loss: 0.0127 - accuracy: 0.9955 - val_loss: 0.6395 - val_accuracy: 0.8807\n",
      "\n",
      "Epoch 00010: val_accuracy did not improve from 0.89304\n",
      "Epoch 11/50\n",
      "299/299 [==============================] - 1438s 5s/step - loss: 0.0107 - accuracy: 0.9963 - val_loss: 0.7747 - val_accuracy: 0.8582\n",
      "\n",
      "Epoch 00011: val_accuracy did not improve from 0.89304\n",
      "Epoch 12/50\n",
      "299/299 [==============================] - 1454s 5s/step - loss: 0.0089 - accuracy: 0.9971 - val_loss: 0.7087 - val_accuracy: 0.8766\n",
      "\n",
      "Epoch 00012: val_accuracy did not improve from 0.89304\n",
      "Epoch 13/50\n",
      "299/299 [==============================] - 1433s 5s/step - loss: 0.0091 - accuracy: 0.9970 - val_loss: 0.7389 - val_accuracy: 0.8756\n",
      "\n",
      "Epoch 00013: val_accuracy did not improve from 0.89304\n",
      "Epoch 14/50\n",
      "299/299 [==============================] - 1439s 5s/step - loss: 0.0077 - accuracy: 0.9974 - val_loss: 0.8640 - val_accuracy: 0.8584\n",
      "\n",
      "Epoch 00014: val_accuracy did not improve from 0.89304\n",
      "Epoch 15/50\n",
      "299/299 [==============================] - 1447s 5s/step - loss: 0.0071 - accuracy: 0.9975 - val_loss: 0.7228 - val_accuracy: 0.8779\n",
      "\n",
      "Epoch 00015: val_accuracy did not improve from 0.89304\n",
      "Epoch 16/50\n",
      "299/299 [==============================] - 1417s 5s/step - loss: 0.0085 - accuracy: 0.9972 - val_loss: 1.2364 - val_accuracy: 0.8212\n",
      "\n",
      "Epoch 00016: val_accuracy did not improve from 0.89304\n",
      "Epoch 17/50\n",
      "299/299 [==============================] - 1407s 5s/step - loss: 0.0074 - accuracy: 0.9974 - val_loss: 0.8958 - val_accuracy: 0.8637\n",
      "\n",
      "Epoch 00017: val_accuracy did not improve from 0.89304\n",
      "Epoch 18/50\n",
      "299/299 [==============================] - 1424s 5s/step - loss: 0.0078 - accuracy: 0.9973 - val_loss: 0.8711 - val_accuracy: 0.8702\n",
      "\n",
      "Epoch 00018: val_accuracy did not improve from 0.89304\n",
      "Epoch 19/50\n",
      "299/299 [==============================] - 1430s 5s/step - loss: 0.0093 - accuracy: 0.9970 - val_loss: 0.8361 - val_accuracy: 0.8681\n",
      "\n",
      "Epoch 00019: val_accuracy did not improve from 0.89304\n",
      "Epoch 20/50\n",
      "299/299 [==============================] - 1427s 5s/step - loss: 0.0060 - accuracy: 0.9980 - val_loss: 0.8441 - val_accuracy: 0.8745\n",
      "\n",
      "Epoch 00020: val_accuracy did not improve from 0.89304\n",
      "Epoch 21/50\n",
      "299/299 [==============================] - 1428s 5s/step - loss: 0.0039 - accuracy: 0.9988 - val_loss: 0.8997 - val_accuracy: 0.8707\n",
      "\n",
      "Epoch 00021: val_accuracy did not improve from 0.89304\n",
      "Epoch 22/50\n",
      "299/299 [==============================] - 1428s 5s/step - loss: 0.0051 - accuracy: 0.9984 - val_loss: 1.1519 - val_accuracy: 0.8465\n",
      "\n",
      "Epoch 00022: val_accuracy did not improve from 0.89304\n",
      "Epoch 23/50\n",
      "299/299 [==============================] - 1425s 5s/step - loss: 0.0056 - accuracy: 0.9983 - val_loss: 0.8584 - val_accuracy: 0.8726\n",
      "\n",
      "Epoch 00023: val_accuracy did not improve from 0.89304\n",
      "Epoch 24/50\n",
      "299/299 [==============================] - 1425s 5s/step - loss: 0.0064 - accuracy: 0.9978 - val_loss: 0.8937 - val_accuracy: 0.8686\n",
      "\n",
      "Epoch 00024: val_accuracy did not improve from 0.89304\n",
      "Epoch 25/50\n",
      "299/299 [==============================] - 1420s 5s/step - loss: 0.0087 - accuracy: 0.9971 - val_loss: 0.8392 - val_accuracy: 0.8744\n",
      "\n",
      "Epoch 00025: val_accuracy did not improve from 0.89304\n",
      "Epoch 26/50\n",
      "299/299 [==============================] - 1418s 5s/step - loss: 0.0038 - accuracy: 0.9988 - val_loss: 0.8369 - val_accuracy: 0.8796\n",
      "\n",
      "Epoch 00026: val_accuracy did not improve from 0.89304\n",
      "Epoch 27/50\n",
      "299/299 [==============================] - 1423s 5s/step - loss: 0.0050 - accuracy: 0.9984 - val_loss: 1.1099 - val_accuracy: 0.8572\n",
      "\n",
      "Epoch 00027: val_accuracy did not improve from 0.89304\n",
      "Epoch 28/50\n",
      "299/299 [==============================] - 1421s 5s/step - loss: 0.0061 - accuracy: 0.9977 - val_loss: 0.9095 - val_accuracy: 0.8719\n",
      "\n",
      "Epoch 00028: val_accuracy did not improve from 0.89304\n",
      "Epoch 29/50\n",
      "299/299 [==============================] - 1421s 5s/step - loss: 0.0054 - accuracy: 0.9982 - val_loss: 0.8669 - val_accuracy: 0.8674\n",
      "\n",
      "Epoch 00029: val_accuracy did not improve from 0.89304\n",
      "Epoch 30/50\n",
      "299/299 [==============================] - 1423s 5s/step - loss: 0.0045 - accuracy: 0.9986 - val_loss: 1.5203 - val_accuracy: 0.8157\n",
      "\n",
      "Epoch 00030: val_accuracy did not improve from 0.89304\n",
      "Epoch 31/50\n",
      "299/299 [==============================] - 1421s 5s/step - loss: 0.0046 - accuracy: 0.9985 - val_loss: 1.0722 - val_accuracy: 0.8668\n",
      "\n",
      "Epoch 00031: val_accuracy did not improve from 0.89304\n",
      "Epoch 32/50\n",
      "299/299 [==============================] - 1413s 5s/step - loss: 0.0049 - accuracy: 0.9984 - val_loss: 1.1717 - val_accuracy: 0.8556\n",
      "\n",
      "Epoch 00032: val_accuracy did not improve from 0.89304\n",
      "Epoch 33/50\n",
      "299/299 [==============================] - 1412s 5s/step - loss: 0.0044 - accuracy: 0.9985 - val_loss: 0.9501 - val_accuracy: 0.8695\n",
      "\n",
      "Epoch 00033: val_accuracy did not improve from 0.89304\n",
      "Epoch 34/50\n",
      "299/299 [==============================] - 1413s 5s/step - loss: 0.0042 - accuracy: 0.9989 - val_loss: 0.8882 - val_accuracy: 0.8799\n",
      "\n",
      "Epoch 00034: val_accuracy did not improve from 0.89304\n",
      "Epoch 35/50\n",
      "299/299 [==============================] - 1414s 5s/step - loss: 0.0031 - accuracy: 0.9990 - val_loss: 0.9923 - val_accuracy: 0.8619\n",
      "\n",
      "Epoch 00035: val_accuracy did not improve from 0.89304\n",
      "Epoch 36/50\n",
      "299/299 [==============================] - 1410s 5s/step - loss: 0.0031 - accuracy: 0.9990 - val_loss: 1.0723 - val_accuracy: 0.8641\n",
      "\n",
      "Epoch 00036: val_accuracy did not improve from 0.89304\n",
      "Epoch 37/50\n",
      "299/299 [==============================] - 1414s 5s/step - loss: 0.0049 - accuracy: 0.9983 - val_loss: 1.2353 - val_accuracy: 0.8539\n",
      "\n",
      "Epoch 00037: val_accuracy did not improve from 0.89304\n",
      "Epoch 38/50\n",
      "299/299 [==============================] - 1411s 5s/step - loss: 0.0051 - accuracy: 0.9985 - val_loss: 0.8990 - val_accuracy: 0.8680\n",
      "\n",
      "Epoch 00038: val_accuracy did not improve from 0.89304\n",
      "Epoch 39/50\n",
      "299/299 [==============================] - 1409s 5s/step - loss: 0.0047 - accuracy: 0.9983 - val_loss: 0.9213 - val_accuracy: 0.8759\n",
      "\n",
      "Epoch 00039: val_accuracy did not improve from 0.89304\n",
      "Epoch 40/50\n",
      "299/299 [==============================] - 1408s 5s/step - loss: 0.0027 - accuracy: 0.9991 - val_loss: 0.9988 - val_accuracy: 0.8744\n",
      "\n",
      "Epoch 00040: val_accuracy did not improve from 0.89304\n",
      "Epoch 41/50\n",
      "299/299 [==============================] - 1411s 5s/step - loss: 0.0026 - accuracy: 0.9990 - val_loss: 0.9162 - val_accuracy: 0.8776\n",
      "\n",
      "Epoch 00041: val_accuracy did not improve from 0.89304\n",
      "Epoch 42/50\n",
      "299/299 [==============================] - 1410s 5s/step - loss: 0.0042 - accuracy: 0.9987 - val_loss: 0.9502 - val_accuracy: 0.8769\n",
      "\n",
      "Epoch 00042: val_accuracy did not improve from 0.89304\n",
      "Epoch 43/50\n",
      "299/299 [==============================] - 1412s 5s/step - loss: 0.0030 - accuracy: 0.9990 - val_loss: 1.1087 - val_accuracy: 0.8714\n",
      "\n",
      "Epoch 00043: val_accuracy did not improve from 0.89304\n",
      "Epoch 44/50\n",
      "299/299 [==============================] - 1413s 5s/step - loss: 0.0046 - accuracy: 0.9986 - val_loss: 0.9274 - val_accuracy: 0.8791\n",
      "\n",
      "Epoch 00044: val_accuracy did not improve from 0.89304\n",
      "Epoch 45/50\n",
      "299/299 [==============================] - 1411s 5s/step - loss: 0.0031 - accuracy: 0.9990 - val_loss: 0.9064 - val_accuracy: 0.8801\n",
      "\n",
      "Epoch 00045: val_accuracy did not improve from 0.89304\n",
      "Epoch 46/50\n",
      "299/299 [==============================] - 1411s 5s/step - loss: 0.0042 - accuracy: 0.9988 - val_loss: 0.9632 - val_accuracy: 0.8767\n",
      "\n",
      "Epoch 00046: val_accuracy did not improve from 0.89304\n",
      "Epoch 47/50\n",
      "299/299 [==============================] - 1409s 5s/step - loss: 0.0039 - accuracy: 0.9988 - val_loss: 0.9998 - val_accuracy: 0.8711\n",
      "\n",
      "Epoch 00047: val_accuracy did not improve from 0.89304\n",
      "Epoch 48/50\n",
      "299/299 [==============================] - 1410s 5s/step - loss: 0.0020 - accuracy: 0.9994 - val_loss: 1.0293 - val_accuracy: 0.8785\n",
      "\n",
      "Epoch 00048: val_accuracy did not improve from 0.89304\n",
      "Epoch 49/50\n",
      "299/299 [==============================] - 1410s 5s/step - loss: 0.0029 - accuracy: 0.9991 - val_loss: 0.9959 - val_accuracy: 0.8799\n",
      "\n",
      "Epoch 00049: val_accuracy did not improve from 0.89304\n",
      "Epoch 50/50\n",
      "299/299 [==============================] - 1410s 5s/step - loss: 0.0020 - accuracy: 0.9994 - val_loss: 1.1119 - val_accuracy: 0.8769\n",
      "\n",
      "Epoch 00050: val_accuracy did not improve from 0.89304\n"
     ]
    }
   ],
   "source": [
    "max_len = 200\n",
    "model = build_model_ACR(lr = 1e-3, lr_d = 1e-10, units = 128, spatial_dr = 0.5, kernel_size1=2, kernel_size2=3, dense_units=32, dr=0.2, conv_size=64)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 213,
   "metadata": {},
   "outputs": [],
   "source": [
    "SEED = 2000\n",
    "csv = path+'model/our_model_result_IMDB-GN.csv' \n",
    "\n",
    "\n",
    "my_df3 = pd.read_csv(csv, encoding = 'latin-1')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Exception ignored in: <function WeakKeyDictionary.__init__.<locals>.remove at 0x000001DE3F3018B0>\n",
      "Traceback (most recent call last):\n",
      "  File \"C:\\Users\\kamyab\\anaconda3\\lib\\weakref.py\", line 345, in remove\n",
      "    def remove(k, selfref=ref(self)):\n",
      "KeyboardInterrupt: \n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(8,7))\n",
    "ax.set_facecolor('white')\n",
    "\n",
    "\n",
    "ax.plot(my_df3.accuracy, label=\"training without GN\",linewidth=2.3, color = '#c98c4f')\n",
    "if my_df3.accuracy is not None:\n",
    "    ax.plot(my_df3.val_accuracy, label=\"validation without GN\",linewidth=2.3, color = '#39e600')\n",
    "ax.set_title('Model Accuracy')\n",
    "ax.set_xlabel('number of Epoch')\n",
    "ax.legend(loc='lower right')\n",
    "\n",
    "ax.plot(my_df.accuracy, label=\"training\",linewidth=2.3)\n",
    "if my_df.accuracy is not None:\n",
    "    ax.plot(my_df.val_accuracy, label=\"validation\",linewidth=2.3)\n",
    "ax.set_title('Model Accuracy')\n",
    "ax.set_xlabel('number of Epoch')\n",
    "ax.legend(loc='lower right')\n",
    "\n",
    "ax.set_ylim(bottom=0.5)\n",
    "ax.set_ylim(top=1.01)\n",
    "from mpl_toolkits.axes_grid1.inset_locator import mark_inset\n",
    "\n",
    "plt.savefig(path+'our_model_result.png',dpi=300,bbox_inches = 'tight',pad_inches = 0.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}