image-cnn_optimizer_analysis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importing DataSet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['src', 'lib', 'input', 'working']\n",
      "['data']\n",
      "['natural_images']\n",
      "['airplane', 'dog', 'fruit', 'cat', 'flower', 'person', 'car', 'motorbike']\n"
     ]
    }
   ],
   "source": [
    "#importing DataSet\n",
    "import os\n",
    "print(os.listdir('../'))\n",
    "print(os.listdir('../input'))\n",
    "print(os.listdir('../input/data'))\n",
    "print(os.listdir('../input/data/natural_images/'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importing Layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Conv2D\n",
    "from keras.layers import MaxPool2D\n",
    "from keras.layers import Dense\n",
    "from keras.layers import Dropout\n",
    "from keras.layers import Flatten"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Declaring variables with Values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "batchSize = 50\n",
    "epochSize = 60\n",
    "#dimensions\n",
    "dim1 = dim2 = 256"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preparing Model and Adding Layers for ADAM optimizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model initializing\n",
    "model = Sequential()\n",
    "\n",
    "#adding conv2d and maxpool2d layers\n",
    "model.add(Conv2D(32, (3,3), activation='relu', input_shape=(dim1, dim2, 3)))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "#2nd layer addition\n",
    "model.add(Conv2D(64, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Conv2D(128, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Flatten())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "#hidden layers\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(rate=0.4))\n",
    "\n",
    "model.add(Dense(512, activation='relu'))\n",
    "model.add(Dropout(rate=0.3))\n",
    "\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(rate=0.2))\n",
    "\n",
    "#output\n",
    "model.add(Dense(8, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model compilation - ADAM\n",
    "model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Image Precprocessing & generation of Training and validarion sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "img_generator = ImageDataGenerator(\n",
    "        validation_split=0.2,\n",
    "        rescale=1./255,\n",
    "        shear_range=0.2,\n",
    "        zoom_range=0.2,\n",
    "        horizontal_flip=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 5522 images belonging to 8 classes.\n",
      "Found 1377 images belonging to 8 classes.\n"
     ]
    }
   ],
   "source": [
    "train_generator = img_generator.flow_from_directory(\n",
    "        '../input/data/natural_images/',\n",
    "        target_size=(dim1, dim2),\n",
    "        batch_size=batchSize,\n",
    "        subset='training'\n",
    "        )\n",
    "\n",
    "validation_generator = img_generator.flow_from_directory(\n",
    "        '../input/data/natural_images/',\n",
    "        target_size=(dim1, dim2),\n",
    "        batch_size=batchSize,\n",
    "        subset='validation'\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Fitting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/60\n",
      "111/110 [==============================] - 106s 953ms/step - loss: 1.4615 - acc: 0.4674 - val_loss: 0.7023 - val_acc: 0.7386\n",
      "Epoch 2/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.5715 - acc: 0.7848 - val_loss: 0.4402 - val_acc: 0.8228\n",
      "Epoch 3/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.4742 - acc: 0.8196 - val_loss: 0.4473 - val_acc: 0.8206\n",
      "Epoch 4/60\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.3742 - acc: 0.8527 - val_loss: 0.4627 - val_acc: 0.8344\n",
      "Epoch 5/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.3282 - acc: 0.8737 - val_loss: 0.3448 - val_acc: 0.8627\n",
      "Epoch 6/60\n",
      "111/110 [==============================] - 85s 761ms/step - loss: 0.3116 - acc: 0.8800 - val_loss: 0.3684 - val_acc: 0.8715\n",
      "Epoch 7/60\n",
      "111/110 [==============================] - 84s 757ms/step - loss: 0.2737 - acc: 0.8915 - val_loss: 0.3288 - val_acc: 0.8903\n",
      "Epoch 8/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.2415 - acc: 0.9041 - val_loss: 0.2945 - val_acc: 0.8853\n",
      "Epoch 9/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 0.2485 - acc: 0.9097 - val_loss: 0.2960 - val_acc: 0.9056\n",
      "Epoch 10/60\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.2077 - acc: 0.9227 - val_loss: 0.2818 - val_acc: 0.8991\n",
      "Epoch 11/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.1948 - acc: 0.9250 - val_loss: 0.2805 - val_acc: 0.9158\n",
      "Epoch 12/60\n",
      "111/110 [==============================] - 86s 771ms/step - loss: 0.1942 - acc: 0.9281 - val_loss: 0.2257 - val_acc: 0.9143\n",
      "Epoch 13/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.1927 - acc: 0.9284 - val_loss: 0.2425 - val_acc: 0.9194\n",
      "Epoch 14/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.1694 - acc: 0.9381 - val_loss: 0.2875 - val_acc: 0.9063\n",
      "Epoch 15/60\n",
      "111/110 [==============================] - 86s 772ms/step - loss: 0.1554 - acc: 0.9390 - val_loss: 0.2738 - val_acc: 0.9107\n",
      "Epoch 16/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.1637 - acc: 0.9413 - val_loss: 0.2561 - val_acc: 0.9121\n",
      "Epoch 17/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.1449 - acc: 0.9473 - val_loss: 0.2496 - val_acc: 0.9216\n",
      "Epoch 18/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.1549 - acc: 0.9414 - val_loss: 0.2332 - val_acc: 0.9194\n",
      "Epoch 19/60\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.1349 - acc: 0.9495 - val_loss: 0.2389 - val_acc: 0.9223\n",
      "Epoch 20/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.1488 - acc: 0.9477 - val_loss: 0.2098 - val_acc: 0.9346\n",
      "Epoch 21/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.1376 - acc: 0.9505 - val_loss: 0.1872 - val_acc: 0.9325\n",
      "Epoch 22/60\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.1173 - acc: 0.9560 - val_loss: 0.2059 - val_acc: 0.9303\n",
      "Epoch 23/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 0.1460 - acc: 0.9526 - val_loss: 0.2215 - val_acc: 0.9245\n",
      "Epoch 24/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.1268 - acc: 0.9553 - val_loss: 0.2358 - val_acc: 0.9267\n",
      "Epoch 25/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.1051 - acc: 0.9630 - val_loss: 0.2552 - val_acc: 0.9259\n",
      "Epoch 26/60\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.1046 - acc: 0.9581 - val_loss: 0.1794 - val_acc: 0.9412\n",
      "Epoch 27/60\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 0.1132 - acc: 0.9600 - val_loss: 0.2113 - val_acc: 0.9368\n",
      "Epoch 28/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.1313 - acc: 0.9548 - val_loss: 0.2118 - val_acc: 0.9310\n",
      "Epoch 29/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.1086 - acc: 0.9590 - val_loss: 0.2210 - val_acc: 0.9361\n",
      "Epoch 30/60\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.0944 - acc: 0.9677 - val_loss: 0.2015 - val_acc: 0.9354\n",
      "Epoch 31/60\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 0.1057 - acc: 0.9625 - val_loss: 0.1975 - val_acc: 0.9368\n",
      "Epoch 32/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.1145 - acc: 0.9643 - val_loss: 0.1746 - val_acc: 0.9361\n",
      "Epoch 33/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.0904 - acc: 0.9693 - val_loss: 0.2015 - val_acc: 0.9310\n",
      "Epoch 34/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 0.1084 - acc: 0.9641 - val_loss: 0.2020 - val_acc: 0.9296\n",
      "Epoch 35/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.0882 - acc: 0.9699 - val_loss: 0.2164 - val_acc: 0.9412\n",
      "Epoch 36/60\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 0.0974 - acc: 0.9676 - val_loss: 0.2336 - val_acc: 0.9339\n",
      "Epoch 37/60\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.0827 - acc: 0.9713 - val_loss: 0.2078 - val_acc: 0.9325\n",
      "Epoch 38/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.0877 - acc: 0.9722 - val_loss: 0.2075 - val_acc: 0.9434\n",
      "Epoch 39/60\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 0.0748 - acc: 0.9751 - val_loss: 0.2701 - val_acc: 0.9317\n",
      "Epoch 40/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 0.0877 - acc: 0.9697 - val_loss: 0.2248 - val_acc: 0.9397\n",
      "Epoch 41/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 0.0832 - acc: 0.9726 - val_loss: 0.2300 - val_acc: 0.9383\n",
      "Epoch 42/60\n",
      "111/110 [==============================] - 84s 759ms/step - loss: 0.0866 - acc: 0.9688 - val_loss: 0.1779 - val_acc: 0.9470\n",
      "Epoch 43/60\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 0.0968 - acc: 0.9693 - val_loss: 0.2010 - val_acc: 0.9332\n",
      "Epoch 44/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.0711 - acc: 0.9754 - val_loss: 0.2125 - val_acc: 0.9397\n",
      "Epoch 45/60\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 0.0783 - acc: 0.9718 - val_loss: 0.1793 - val_acc: 0.9463\n",
      "Epoch 46/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 0.0820 - acc: 0.9737 - val_loss: 0.2024 - val_acc: 0.9419\n",
      "Epoch 47/60\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.0538 - acc: 0.9800 - val_loss: 0.2398 - val_acc: 0.9405\n",
      "Epoch 48/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.0585 - acc: 0.9792 - val_loss: 0.2998 - val_acc: 0.9281\n",
      "Epoch 49/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 0.0771 - acc: 0.9780 - val_loss: 0.2144 - val_acc: 0.9375\n",
      "Epoch 50/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.0607 - acc: 0.9795 - val_loss: 0.1824 - val_acc: 0.9470\n",
      "Epoch 51/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.0618 - acc: 0.9795 - val_loss: 0.2028 - val_acc: 0.9419\n",
      "Epoch 52/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 0.0896 - acc: 0.9717 - val_loss: 0.2191 - val_acc: 0.9346\n",
      "Epoch 53/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.0627 - acc: 0.9778 - val_loss: 0.1678 - val_acc: 0.9463\n",
      "Epoch 54/60\n",
      " 64/110 [================>.............] - ETA: 28s - loss: 0.0551 - acc: 0.9819"
     ]
    }
   ],
   "source": [
    "history = model.fit_generator(\n",
    "        train_generator,\n",
    "        steps_per_epoch=(5522 / batchSize),\n",
    "        epochs=epochSize,\n",
    "        validation_data=validation_generator,\n",
    "        validation_steps=(1377 / batchSize))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'val_loss': [0.7023057827675957,\n",
       "  0.44024516543618303,\n",
       "  0.4473112775371312,\n",
       "  0.46271779992568296,\n",
       "  0.34480176936394075,\n",
       "  0.36835184678186433,\n",
       "  0.3287662456194906,\n",
       "  0.2945237018292419,\n",
       "  0.29600871073524415,\n",
       "  0.28176382961010016,\n",
       "  0.28054655879497875,\n",
       "  0.2257427384856393,\n",
       "  0.242471623167493,\n",
       "  0.2875208434261268,\n",
       "  0.27377288370307323,\n",
       "  0.2561316513114852,\n",
       "  0.2496296697517366,\n",
       "  0.23318368720787544,\n",
       "  0.2389352623976567,\n",
       "  0.20981251076134375,\n",
       "  0.18723166123649393,\n",
       "  0.20593754878469195,\n",
       "  0.2214725449540603,\n",
       "  0.235845444692852,\n",
       "  0.25519593736087837,\n",
       "  0.1794060069469758,\n",
       "  0.21132433249431845,\n",
       "  0.21178321402381967,\n",
       "  0.22095341240268906,\n",
       "  0.20149231360401462,\n",
       "  0.1975255048872339,\n",
       "  0.1745715279215994,\n",
       "  0.20149319641124364,\n",
       "  0.2019817113183769,\n",
       "  0.21642758246421945,\n",
       "  0.2335743100186925,\n",
       "  0.20780836521345547,\n",
       "  0.20748859744034734,\n",
       "  0.2700817462005473,\n",
       "  0.22483972693234142,\n",
       "  0.23004508778972305,\n",
       "  0.17788659059966744,\n",
       "  0.2010145953429637,\n",
       "  0.21253041311227375,\n",
       "  0.17925062051520413,\n",
       "  0.20244525829918503,\n",
       "  0.23982228479397497,\n",
       "  0.299761893007209,\n",
       "  0.21440770253998997,\n",
       "  0.18236516443126516,\n",
       "  0.20279207240219296,\n",
       "  0.2191443390381362,\n",
       "  0.16776159922974476,\n",
       "  0.2499115825694977,\n",
       "  0.2289481730807029,\n",
       "  0.23002103638490992,\n",
       "  0.267161386958665,\n",
       "  0.2715545079671523,\n",
       "  0.20533722328507736,\n",
       "  0.25846229828001366],\n",
       " 'val_acc': [0.7385620940571344,\n",
       "  0.822803193317779,\n",
       "  0.8206245380202842,\n",
       "  0.8344226539697696,\n",
       "  0.8627450915896209,\n",
       "  0.8714596927382414,\n",
       "  0.8903413158702781,\n",
       "  0.8852578041859961,\n",
       "  0.9055918594937266,\n",
       "  0.8990559143784606,\n",
       "  0.9157588983153466,\n",
       "  0.9143064632827853,\n",
       "  0.9193899754864977,\n",
       "  0.9063180766420773,\n",
       "  0.910675380094898,\n",
       "  0.9121278117944472,\n",
       "  0.9215686226462487,\n",
       "  0.9193899771313608,\n",
       "  0.9222948421753223,\n",
       "  0.9346405164262221,\n",
       "  0.9324618714307648,\n",
       "  0.9302832196827114,\n",
       "  0.9244734831019078,\n",
       "  0.926652140606982,\n",
       "  0.9259259186538995,\n",
       "  0.9411764643983557,\n",
       "  0.9368191715072876,\n",
       "  0.9310094370907774,\n",
       "  0.9360929521946433,\n",
       "  0.93536673790316,\n",
       "  0.936819169342994,\n",
       "  0.9360929514587835,\n",
       "  0.9310094397312156,\n",
       "  0.9295569996342073,\n",
       "  0.9411764704150918,\n",
       "  0.9339143026108835,\n",
       "  0.9324618793520792,\n",
       "  0.9433551216437146,\n",
       "  0.9317356571392815,\n",
       "  0.9397240368109643,\n",
       "  0.9382716009992573,\n",
       "  0.9469862002432996,\n",
       "  0.9331880871506818,\n",
       "  0.9397240368109643,\n",
       "  0.9462599845233826,\n",
       "  0.9419026841871446,\n",
       "  0.9404502565997531,\n",
       "  0.9281045684540886,\n",
       "  0.9375453857987708,\n",
       "  0.9469862031434529,\n",
       "  0.9419026880395871,\n",
       "  0.9346405243475366,\n",
       "  0.946259982359089,\n",
       "  0.9302832201588561,\n",
       "  0.9339143023944542,\n",
       "  0.9368191671787004,\n",
       "  0.9324618671021777,\n",
       "  0.936819169342994,\n",
       "  0.9491648497837735,\n",
       "  0.9346405185905158],\n",
       " 'loss': [1.4647329001053655,\n",
       "  0.571992502971129,\n",
       "  0.47337101954172073,\n",
       "  0.3730867423038904,\n",
       "  0.3277757931439957,\n",
       "  0.311597853208184,\n",
       "  0.2740097369746383,\n",
       "  0.24242116558766588,\n",
       "  0.24943575894339323,\n",
       "  0.20874002341576373,\n",
       "  0.19524477323202152,\n",
       "  0.19485069696927407,\n",
       "  0.19186399240223492,\n",
       "  0.16854972438060345,\n",
       "  0.15574058603896088,\n",
       "  0.1642069496364157,\n",
       "  0.144949436779392,\n",
       "  0.1542269546904239,\n",
       "  0.1340984139505103,\n",
       "  0.1477854018366421,\n",
       "  0.13677866099761884,\n",
       "  0.1175634812932004,\n",
       "  0.1460079339935658,\n",
       "  0.12700762427508108,\n",
       "  0.1045685859969812,\n",
       "  0.10454000981557643,\n",
       "  0.1135791164439351,\n",
       "  0.1314945236875209,\n",
       "  0.10757457729522105,\n",
       "  0.09424392406517361,\n",
       "  0.10590470541735429,\n",
       "  0.11490629567153825,\n",
       "  0.09040301877817045,\n",
       "  0.10812683790339962,\n",
       "  0.08837366027218499,\n",
       "  0.09780772858947294,\n",
       "  0.08274145650737087,\n",
       "  0.08770932887648109,\n",
       "  0.07516822433662464,\n",
       "  0.08783322026428916,\n",
       "  0.08358101235643267,\n",
       "  0.08685516130465208,\n",
       "  0.09660225901793187,\n",
       "  0.0710379906473349,\n",
       "  0.07839129288734484,\n",
       "  0.08222629403110562,\n",
       "  0.054036175049395306,\n",
       "  0.05856352711648925,\n",
       "  0.07702746667837379,\n",
       "  0.060956542830551984,\n",
       "  0.06202730015640399,\n",
       "  0.09005072412748269,\n",
       "  0.06297121439783503,\n",
       "  0.06626241130176955,\n",
       "  0.0595926926385278,\n",
       "  0.06503612061689935,\n",
       "  0.051813080102080133,\n",
       "  0.06667387287846872,\n",
       "  0.05831099156771618,\n",
       "  0.043834640424035266],\n",
       " 'acc': [0.46631655303443614,\n",
       "  0.7850416473010794,\n",
       "  0.8196305656122237,\n",
       "  0.8531329178671956,\n",
       "  0.8739587070500665,\n",
       "  0.8801158951573163,\n",
       "  0.8911626167765393,\n",
       "  0.9036580928638767,\n",
       "  0.9092719965937862,\n",
       "  0.9223107533453168,\n",
       "  0.9248460661410768,\n",
       "  0.9279246638538785,\n",
       "  0.9284679442158388,\n",
       "  0.9389713850629282,\n",
       "  0.9389713841562295,\n",
       "  0.9409634165252649,\n",
       "  0.9474827925047552,\n",
       "  0.9420499808543923,\n",
       "  0.9494748251760234,\n",
       "  0.9478449813916345,\n",
       "  0.9507424876293554,\n",
       "  0.955994204631192,\n",
       "  0.9525534196506156,\n",
       "  0.9552698290162396,\n",
       "  0.9632379559234017,\n",
       "  0.9583484229511301,\n",
       "  0.9597971744185037,\n",
       "  0.9545454557975622,\n",
       "  0.9592538938190747,\n",
       "  0.9677653027073019,\n",
       "  0.9625135808481509,\n",
       "  0.9641434264891133,\n",
       "  0.9693951486504971,\n",
       "  0.964143428410451,\n",
       "  0.9699384273933527,\n",
       "  0.9674031151372944,\n",
       "  0.9713871797026606,\n",
       "  0.9722926484117989,\n",
       "  0.9750090574751898,\n",
       "  0.9697573359182717,\n",
       "  0.9724737444419612,\n",
       "  0.9686707716539087,\n",
       "  0.9693951491901988,\n",
       "  0.9755523392187863,\n",
       "  0.9719304616837258,\n",
       "  0.9735603057055834,\n",
       "  0.9798985893680582,\n",
       "  0.9793553109274357,\n",
       "  0.9780876517123133,\n",
       "  0.9793553095457995,\n",
       "  0.9793553095457995,\n",
       "  0.9715682741137183,\n",
       "  0.9777254641423058,\n",
       "  0.9789931225802578,\n",
       "  0.9815284366928897,\n",
       "  0.9795364034819197,\n",
       "  0.9826149981939806,\n",
       "  0.980260777715236,\n",
       "  0.9827960934469727,\n",
       "  0.9842448417409008]}"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history.history"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history.history.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "adam_val_loss = history.history['val_loss']\n",
    "adam_val_acc = history.history['val_acc']\n",
    "adam_loss = history.history['loss']\n",
    "adam_acc = history.history['acc']\n",
    "adam_epochs = range(len(adam_acc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(adam_epochs, adam_acc, 'red', label='Training Acc')\n",
    "plt.plot(adam_epochs, adam_val_acc, 'blue', label='Validation Acc')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('Acc')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(adam_epochs, adam_loss, 'red', label='Training Loss')\n",
    "plt.plot(adam_epochs, adam_val_loss, 'blue', label='Validation Loss')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model preparing for RMSprop optimizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model initializing\n",
    "model = Sequential()\n",
    "\n",
    "#adding conv2d and maxpool2d layers\n",
    "model.add(Conv2D(32, (3,3), activation='relu', input_shape=(dim1, dim2, 3)))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "#2nd layer addition\n",
    "model.add(Conv2D(64, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Conv2D(128, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Flatten())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "#hidden layers\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(rate=0.4))\n",
    "\n",
    "model.add(Dense(512, activation='relu'))\n",
    "model.add(Dropout(rate=0.3))\n",
    "\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(rate=0.2))\n",
    "\n",
    "#output\n",
    "model.add(Dense(8, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model compilation - RMSprop\n",
    "model.compile(optimizer='RMSprop', loss='categorical_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/60\n",
      "111/110 [==============================] - 92s 831ms/step - loss: 1.5297 - acc: 0.5014 - val_loss: 0.7191 - val_acc: 0.7480\n",
      "Epoch 2/60\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.6504 - acc: 0.7665 - val_loss: 0.7111 - val_acc: 0.7749\n",
      "Epoch 3/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 0.5287 - acc: 0.8131 - val_loss: 0.4662 - val_acc: 0.8228\n",
      "Epoch 4/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.4446 - acc: 0.8444 - val_loss: 0.3422 - val_acc: 0.8744\n",
      "Epoch 5/60\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.3639 - acc: 0.8676 - val_loss: 0.3567 - val_acc: 0.8736\n",
      "Epoch 6/60\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 0.3364 - acc: 0.8857 - val_loss: 0.5898 - val_acc: 0.8119\n",
      "Epoch 7/60\n",
      "111/110 [==============================] - 84s 758ms/step - loss: 0.2892 - acc: 0.8939 - val_loss: 0.3573 - val_acc: 0.8925\n",
      "Epoch 8/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.2882 - acc: 0.8999 - val_loss: 0.4697 - val_acc: 0.8489\n",
      "Epoch 9/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.2747 - acc: 0.9120 - val_loss: 0.4298 - val_acc: 0.8845\n",
      "Epoch 10/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 0.2832 - acc: 0.9076 - val_loss: 0.2574 - val_acc: 0.9208\n",
      "Epoch 11/60\n",
      "111/110 [==============================] - 86s 772ms/step - loss: 0.2708 - acc: 0.9167 - val_loss: 0.3387 - val_acc: 0.9121\n",
      "Epoch 12/60\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.2401 - acc: 0.9264 - val_loss: 0.2701 - val_acc: 0.9201\n",
      "Epoch 13/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 0.2373 - acc: 0.9251 - val_loss: 0.5628 - val_acc: 0.8627\n",
      "Epoch 14/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.2539 - acc: 0.9275 - val_loss: 0.3551 - val_acc: 0.9056\n",
      "Epoch 15/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.2192 - acc: 0.9310 - val_loss: 0.1845 - val_acc: 0.9397\n",
      "Epoch 16/60\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.2299 - acc: 0.9274 - val_loss: 0.2819 - val_acc: 0.9274\n",
      "Epoch 17/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.2223 - acc: 0.9341 - val_loss: 0.3957 - val_acc: 0.9012\n",
      "Epoch 18/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.2260 - acc: 0.9295 - val_loss: 0.2724 - val_acc: 0.9165\n",
      "Epoch 19/60\n",
      "111/110 [==============================] - 86s 771ms/step - loss: 0.2770 - acc: 0.9334 - val_loss: 0.2446 - val_acc: 0.9288\n",
      "Epoch 20/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.2186 - acc: 0.9307 - val_loss: 0.2416 - val_acc: 0.9245\n",
      "Epoch 21/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.2417 - acc: 0.9288 - val_loss: 0.2373 - val_acc: 0.9267\n",
      "Epoch 22/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.2177 - acc: 0.9353 - val_loss: 0.2140 - val_acc: 0.9179\n",
      "Epoch 23/60\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.2259 - acc: 0.9367 - val_loss: 0.2529 - val_acc: 0.9230\n",
      "Epoch 24/60\n",
      "111/110 [==============================] - 84s 759ms/step - loss: 0.1906 - acc: 0.9476 - val_loss: 0.3130 - val_acc: 0.8969\n",
      "Epoch 25/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 0.2318 - acc: 0.9413 - val_loss: 0.2579 - val_acc: 0.9208\n",
      "Epoch 26/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.2480 - acc: 0.9362 - val_loss: 0.2228 - val_acc: 0.9361\n",
      "Epoch 27/60\n",
      "111/110 [==============================] - 86s 771ms/step - loss: 0.2155 - acc: 0.9391 - val_loss: 0.2567 - val_acc: 0.9317\n",
      "Epoch 28/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.2455 - acc: 0.9445 - val_loss: 0.3091 - val_acc: 0.9310\n",
      "Epoch 29/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 0.2841 - acc: 0.9364 - val_loss: 0.2322 - val_acc: 0.9354\n",
      "Epoch 30/60\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.2544 - acc: 0.9384 - val_loss: 0.2212 - val_acc: 0.9354\n",
      "Epoch 31/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.2766 - acc: 0.9323 - val_loss: 0.3247 - val_acc: 0.9121\n",
      "Epoch 32/60\n",
      "111/110 [==============================] - 86s 771ms/step - loss: 0.2552 - acc: 0.9407 - val_loss: 0.2174 - val_acc: 0.9368\n",
      "Epoch 33/60\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.2759 - acc: 0.9356 - val_loss: 0.2356 - val_acc: 0.9288\n",
      "Epoch 34/60\n",
      "111/110 [==============================] - 86s 772ms/step - loss: 0.2847 - acc: 0.9345 - val_loss: 0.2486 - val_acc: 0.9339\n",
      "Epoch 35/60\n",
      "111/110 [==============================] - 85s 761ms/step - loss: 0.2274 - acc: 0.9436 - val_loss: 0.2177 - val_acc: 0.9426\n",
      "Epoch 36/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.2348 - acc: 0.9451 - val_loss: 0.3654 - val_acc: 0.9201\n",
      "Epoch 37/60\n",
      "111/110 [==============================] - 86s 770ms/step - loss: 0.2314 - acc: 0.9398 - val_loss: 0.4254 - val_acc: 0.9085\n",
      "Epoch 38/60\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.2341 - acc: 0.9406 - val_loss: 0.2795 - val_acc: 0.9237\n",
      "Epoch 39/60\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.2751 - acc: 0.9406 - val_loss: 0.2392 - val_acc: 0.9339\n",
      "Epoch 40/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.2550 - acc: 0.9407 - val_loss: 0.2417 - val_acc: 0.9288\n",
      "Epoch 41/60\n",
      "111/110 [==============================] - 85s 770ms/step - loss: 0.2331 - acc: 0.9421 - val_loss: 0.2086 - val_acc: 0.9434\n",
      "Epoch 42/60\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.2905 - acc: 0.9394 - val_loss: 0.4219 - val_acc: 0.8991\n",
      "Epoch 43/60\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.2837 - acc: 0.9389 - val_loss: 0.3059 - val_acc: 0.9267\n",
      "Epoch 44/60\n",
      "111/110 [==============================] - 86s 776ms/step - loss: 0.2934 - acc: 0.9409 - val_loss: 0.5908 - val_acc: 0.8773\n",
      "Epoch 45/60\n",
      "111/110 [==============================] - 87s 786ms/step - loss: 0.2539 - acc: 0.9369 - val_loss: 0.2579 - val_acc: 0.9339\n",
      "Epoch 46/60\n",
      "111/110 [==============================] - 89s 798ms/step - loss: 0.2466 - acc: 0.9409 - val_loss: 0.3338 - val_acc: 0.9085\n",
      "Epoch 47/60\n",
      "111/110 [==============================] - 85s 770ms/step - loss: 0.2512 - acc: 0.9491 - val_loss: 0.5452 - val_acc: 0.9092\n",
      "Epoch 48/60\n",
      "111/110 [==============================] - 86s 777ms/step - loss: 0.2811 - acc: 0.9427 - val_loss: 0.2344 - val_acc: 0.9274\n",
      "Epoch 49/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.3077 - acc: 0.9402 - val_loss: 0.2988 - val_acc: 0.9303\n",
      "Epoch 50/60\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.3004 - acc: 0.9401 - val_loss: 0.3131 - val_acc: 0.9136\n",
      "Epoch 51/60\n",
      "111/110 [==============================] - 85s 769ms/step - loss: 0.2486 - acc: 0.9448 - val_loss: 0.3421 - val_acc: 0.9317\n",
      "Epoch 52/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.3351 - acc: 0.9345 - val_loss: 0.3203 - val_acc: 0.9020\n",
      "Epoch 53/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.2476 - acc: 0.9470 - val_loss: 0.2318 - val_acc: 0.9405\n",
      "Epoch 54/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.3119 - acc: 0.9388 - val_loss: 0.2212 - val_acc: 0.9426\n",
      "Epoch 55/60\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.3405 - acc: 0.9349 - val_loss: 0.3199 - val_acc: 0.9223\n",
      "Epoch 56/60\n",
      "111/110 [==============================] - 86s 771ms/step - loss: 0.2812 - acc: 0.9381 - val_loss: 0.3072 - val_acc: 0.9354\n",
      "Epoch 57/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.3351 - acc: 0.9344 - val_loss: 0.3498 - val_acc: 0.9194\n",
      "Epoch 58/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 0.2966 - acc: 0.9343 - val_loss: 0.3305 - val_acc: 0.9150\n",
      "Epoch 59/60\n",
      "111/110 [==============================] - 86s 772ms/step - loss: 0.3121 - acc: 0.9428 - val_loss: 0.3398 - val_acc: 0.9230\n",
      "Epoch 60/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 0.3544 - acc: 0.9353 - val_loss: 0.2402 - val_acc: 0.9296\n"
     ]
    }
   ],
   "source": [
    "RMSprop_history = model.fit_generator(\n",
    "        train_generator,\n",
    "        steps_per_epoch=(5522 / batchSize),\n",
    "        epochs=epochSize,\n",
    "        validation_data=validation_generator,\n",
    "        validation_steps=(1377 / batchSize))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "RMSprop_history.history.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "RMSprop_loss = RMSprop_history.history['loss']\n",
    "RMSprop_val_loss = RMSprop_history.history['val_loss']\n",
    "RMSprop_acc = RMSprop_history.history['acc']\n",
    "RMSprop_val_acc = RMSprop_history.history['val_acc']\n",
    "RMSprop_epochs = range(len(RMSprop_acc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(RMSprop_epochs, RMSprop_acc, 'red', label='Training Acc')\n",
    "plt.plot(RMSprop_epochs, RMSprop_val_acc, 'blue', label='Validation Acc')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('Acc')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(RMSprop_epochs, RMSprop_loss, 'red', label='Training Loss')\n",
    "plt.plot(RMSprop_epochs, RMSprop_val_loss, 'blue', label='Validation Loss')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model preperation for Adagrad"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model initializing\n",
    "model = Sequential()\n",
    "\n",
    "#adding conv2d and maxpool2d layers\n",
    "model.add(Conv2D(32, (3,3), activation='relu', input_shape=(dim1, dim2, 3)))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "#2nd layer addition\n",
    "model.add(Conv2D(64, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Conv2D(128, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Flatten())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "#hidden layers\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(rate=0.4))\n",
    "\n",
    "model.add(Dense(512, activation='relu'))\n",
    "model.add(Dropout(rate=0.3))\n",
    "\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(rate=0.2))\n",
    "\n",
    "#output\n",
    "model.add(Dense(8, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model compilation - Adagrad\n",
    "model.compile(optimizer='Adagrad', loss='categorical_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/60\n",
      "111/110 [==============================] - 92s 832ms/step - loss: 13.6940 - acc: 0.1432 - val_loss: 13.8239 - val_acc: 0.1423\n",
      "Epoch 2/60\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 13.8737 - acc: 0.1392 - val_loss: 13.8122 - val_acc: 0.1431\n",
      "Epoch 3/60\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 13.8040 - acc: 0.1436 - val_loss: 13.8239 - val_acc: 0.1423\n",
      "Epoch 4/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 13.7806 - acc: 0.1450 - val_loss: 13.8473 - val_acc: 0.1409\n",
      "Epoch 5/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 13.8272 - acc: 0.1421 - val_loss: 13.8356 - val_acc: 0.1416\n",
      "Epoch 6/60\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 13.7858 - acc: 0.1447 - val_loss: 13.7888 - val_acc: 0.1445\n",
      "Epoch 7/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 13.8613 - acc: 0.1400 - val_loss: 13.8356 - val_acc: 0.1416\n",
      "Epoch 8/60\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 13.7721 - acc: 0.1456 - val_loss: 13.7302 - val_acc: 0.1481\n",
      "Epoch 9/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 13.8054 - acc: 0.1435 - val_loss: 13.8941 - val_acc: 0.1380\n",
      "Epoch 10/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 13.8919 - acc: 0.1381 - val_loss: 13.7536 - val_acc: 0.1467\n",
      "Epoch 11/60\n",
      "111/110 [==============================] - 84s 759ms/step - loss: 13.7176 - acc: 0.1489 - val_loss: 13.7888 - val_acc: 0.1445\n",
      "Epoch 12/60\n",
      "111/110 [==============================] - 84s 759ms/step - loss: 13.8404 - acc: 0.1413 - val_loss: 13.8473 - val_acc: 0.1409\n",
      "Epoch 13/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 13.8476 - acc: 0.1409 - val_loss: 13.8005 - val_acc: 0.1438\n",
      "Epoch 14/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 13.7887 - acc: 0.1445 - val_loss: 13.8005 - val_acc: 0.1438\n",
      "Epoch 15/60\n",
      "111/110 [==============================] - 84s 759ms/step - loss: 13.7895 - acc: 0.1445 - val_loss: 13.8941 - val_acc: 0.1380\n",
      "Epoch 16/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 13.8317 - acc: 0.1419 - val_loss: 13.7419 - val_acc: 0.1474\n",
      "Epoch 17/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 13.8178 - acc: 0.1427 - val_loss: 13.8590 - val_acc: 0.1402\n",
      "Epoch 18/60\n",
      "111/110 [==============================] - 84s 758ms/step - loss: 13.8454 - acc: 0.1410 - val_loss: 13.7302 - val_acc: 0.1481\n",
      "Epoch 19/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 13.7931 - acc: 0.1442 - val_loss: 13.8707 - val_acc: 0.1394\n",
      "Epoch 20/60\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 13.8288 - acc: 0.1420 - val_loss: 13.8239 - val_acc: 0.1423\n",
      "Epoch 21/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 13.8077 - acc: 0.1433 - val_loss: 13.8239 - val_acc: 0.1423\n",
      "Epoch 22/60\n",
      "111/110 [==============================] - 84s 759ms/step - loss: 13.7974 - acc: 0.1440 - val_loss: 13.8122 - val_acc: 0.1431\n",
      "Epoch 23/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 13.8476 - acc: 0.1409 - val_loss: 13.8005 - val_acc: 0.1438\n",
      "Epoch 24/60\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 13.8317 - acc: 0.1419 - val_loss: 13.7888 - val_acc: 0.1445\n",
      "Epoch 25/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 13.8149 - acc: 0.1429 - val_loss: 13.8473 - val_acc: 0.1409\n",
      "Epoch 26/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 13.8193 - acc: 0.1426 - val_loss: 13.7771 - val_acc: 0.1452\n",
      "Epoch 27/60\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 13.7808 - acc: 0.1450 - val_loss: 13.8473 - val_acc: 0.1409\n",
      "Epoch 28/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 13.8454 - acc: 0.1410 - val_loss: 13.8122 - val_acc: 0.1431\n",
      "Epoch 29/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 13.8606 - acc: 0.1401 - val_loss: 13.7888 - val_acc: 0.1445\n",
      "Epoch 30/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 13.8048 - acc: 0.1435 - val_loss: 13.8122 - val_acc: 0.1431\n",
      "Epoch 31/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 13.7859 - acc: 0.1447 - val_loss: 13.8824 - val_acc: 0.1387\n",
      "Epoch 32/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 13.8243 - acc: 0.1423 - val_loss: 13.7302 - val_acc: 0.1481\n",
      "Epoch 33/60\n",
      "111/110 [==============================] - 84s 759ms/step - loss: 13.7728 - acc: 0.1455 - val_loss: 13.8239 - val_acc: 0.1423\n",
      "Epoch 34/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 13.8404 - acc: 0.1413 - val_loss: 13.7419 - val_acc: 0.1474\n",
      "Epoch 35/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 13.8236 - acc: 0.1424 - val_loss: 13.8707 - val_acc: 0.1394\n",
      "Epoch 36/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 13.7589 - acc: 0.1464 - val_loss: 13.8590 - val_acc: 0.1402\n",
      "Epoch 37/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 13.8991 - acc: 0.1377 - val_loss: 13.7888 - val_acc: 0.1445\n",
      "Epoch 38/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 13.7750 - acc: 0.1454 - val_loss: 13.7302 - val_acc: 0.1481\n",
      "Epoch 39/60\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 13.8570 - acc: 0.1403 - val_loss: 13.8590 - val_acc: 0.1402\n",
      "Epoch 40/60\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 13.8534 - acc: 0.1405 - val_loss: 13.8473 - val_acc: 0.1409\n",
      "Epoch 41/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 13.7517 - acc: 0.1468 - val_loss: 13.7653 - val_acc: 0.1460\n",
      "Epoch 42/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 13.8106 - acc: 0.1432 - val_loss: 13.8356 - val_acc: 0.1416\n",
      "Epoch 43/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 13.8315 - acc: 0.1419 - val_loss: 13.8590 - val_acc: 0.1402\n",
      "Epoch 44/60\n",
      "111/110 [==============================] - 84s 758ms/step - loss: 13.8171 - acc: 0.1428 - val_loss: 13.7068 - val_acc: 0.1496\n",
      "Epoch 45/60\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 13.8091 - acc: 0.1433 - val_loss: 13.9058 - val_acc: 0.1373\n",
      "Epoch 46/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 13.8375 - acc: 0.1415 - val_loss: 13.8005 - val_acc: 0.1438\n",
      "Epoch 47/60\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 13.8142 - acc: 0.1429 - val_loss: 13.7771 - val_acc: 0.1452\n",
      "Epoch 48/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 13.8142 - acc: 0.1429 - val_loss: 13.8122 - val_acc: 0.1431\n",
      "Epoch 49/60\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 13.8410 - acc: 0.1413 - val_loss: 13.8005 - val_acc: 0.1438\n",
      "Epoch 50/60\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 13.7517 - acc: 0.1468 - val_loss: 13.9175 - val_acc: 0.1365\n",
      "Epoch 51/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 13.8346 - acc: 0.1417 - val_loss: 13.7536 - val_acc: 0.1467\n",
      "Epoch 52/60\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 13.8462 - acc: 0.1410 - val_loss: 13.7888 - val_acc: 0.1445\n",
      "Epoch 53/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 13.8460 - acc: 0.1410 - val_loss: 13.8122 - val_acc: 0.1431\n",
      "Epoch 54/60\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 13.7438 - acc: 0.1473 - val_loss: 13.7888 - val_acc: 0.1445\n",
      "Epoch 55/60\n",
      "111/110 [==============================] - 84s 759ms/step - loss: 13.8309 - acc: 0.1419 - val_loss: 13.8590 - val_acc: 0.1402\n",
      "Epoch 56/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 13.8222 - acc: 0.1424 - val_loss: 13.7771 - val_acc: 0.1452\n",
      "Epoch 57/60\n",
      "111/110 [==============================] - 84s 759ms/step - loss: 13.8541 - acc: 0.1405 - val_loss: 13.8824 - val_acc: 0.1387\n",
      "Epoch 58/60\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 13.7779 - acc: 0.1452 - val_loss: 13.7419 - val_acc: 0.1474\n",
      "Epoch 59/60\n",
      "111/110 [==============================] - 85s 761ms/step - loss: 13.8751 - acc: 0.1392 - val_loss: 13.8005 - val_acc: 0.1438\n",
      "Epoch 60/60\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 13.7931 - acc: 0.1442 - val_loss: 13.8005 - val_acc: 0.1438\n"
     ]
    }
   ],
   "source": [
    "Adagrad_history = model.fit_generator(\n",
    "        train_generator,\n",
    "        steps_per_epoch=(5522 / batchSize),\n",
    "        epochs=epochSize,\n",
    "        validation_data=validation_generator,\n",
    "        validation_steps=(1377 / batchSize))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Adagrad_history.history.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "Adagrad_loss = Adagrad_history.history['loss']\n",
    "Adagrad_val_loss = Adagrad_history.history['val_loss']\n",
    "Adagrad_acc = Adagrad_history.history['acc']\n",
    "Adagrad_val_acc = Adagrad_history.history['val_acc']\n",
    "Adagrad_epochs = range(len(Adagrad_acc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(Adagrad_epochs, Adagrad_acc, 'red', label='Training Acc')\n",
    "plt.plot(Adagrad_epochs, Adagrad_val_acc, 'blue', label='Validation Acc')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('Acc')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(Adagrad_epochs, Adagrad_loss, 'red', label='Training Loss')\n",
    "plt.plot(Adagrad_epochs, Adagrad_val_loss, 'blue', label='Validation Loss')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model preperation for Adadelta"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model initializing\n",
    "model = Sequential()\n",
    "\n",
    "#adding conv2d and maxpool2d layers\n",
    "model.add(Conv2D(32, (3,3), activation='relu', input_shape=(dim1, dim2, 3)))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "#2nd layer addition\n",
    "model.add(Conv2D(64, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Conv2D(128, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Flatten())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "#hidden layers\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(rate=0.4))\n",
    "\n",
    "model.add(Dense(512, activation='relu'))\n",
    "model.add(Dropout(rate=0.3))\n",
    "\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(rate=0.2))\n",
    "\n",
    "#output\n",
    "model.add(Dense(8, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model compilation - Adadelta\n",
    "model.compile(optimizer='Adadelta', loss='categorical_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "111/110 [==============================] - 93s 836ms/step - loss: 1.8100 - acc: 0.3185 - val_loss: 1.1904 - val_acc: 0.6144\n",
      "Epoch 2/30\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 0.9589 - acc: 0.6637 - val_loss: 0.9814 - val_acc: 0.6115\n",
      "Epoch 3/30\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.5906 - acc: 0.7847 - val_loss: 0.8425 - val_acc: 0.7117\n",
      "Epoch 4/30\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 0.4721 - acc: 0.8285 - val_loss: 0.4160 - val_acc: 0.8373\n",
      "Epoch 5/30\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.4098 - acc: 0.8457 - val_loss: 0.4283 - val_acc: 0.8301\n",
      "Epoch 6/30\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.3669 - acc: 0.8646 - val_loss: 0.3535 - val_acc: 0.8729\n",
      "Epoch 7/30\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.3360 - acc: 0.8729 - val_loss: 0.4423 - val_acc: 0.8519\n",
      "Epoch 8/30\n",
      "111/110 [==============================] - 85s 761ms/step - loss: 0.3080 - acc: 0.8819 - val_loss: 0.3159 - val_acc: 0.8787\n",
      "Epoch 9/30\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 0.2874 - acc: 0.8947 - val_loss: 0.2744 - val_acc: 0.8991\n",
      "Epoch 10/30\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.2352 - acc: 0.9095 - val_loss: 0.3688 - val_acc: 0.8671\n",
      "Epoch 11/30\n",
      "111/110 [==============================] - 84s 759ms/step - loss: 0.2295 - acc: 0.9123 - val_loss: 0.2840 - val_acc: 0.9034\n",
      "Epoch 12/30\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.2192 - acc: 0.9198 - val_loss: 0.2746 - val_acc: 0.8991\n",
      "Epoch 13/30\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 0.2129 - acc: 0.9205 - val_loss: 0.2596 - val_acc: 0.9165\n",
      "Epoch 14/30\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.1939 - acc: 0.9290 - val_loss: 0.2471 - val_acc: 0.9129\n",
      "Epoch 15/30\n",
      "111/110 [==============================] - 85s 761ms/step - loss: 0.1692 - acc: 0.9376 - val_loss: 0.2552 - val_acc: 0.9179\n",
      "Epoch 16/30\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.1695 - acc: 0.9406 - val_loss: 0.2510 - val_acc: 0.9187\n",
      "Epoch 17/30\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.1603 - acc: 0.9432 - val_loss: 0.2361 - val_acc: 0.9259\n",
      "Epoch 18/30\n",
      "111/110 [==============================] - 84s 759ms/step - loss: 0.1471 - acc: 0.9486 - val_loss: 0.1923 - val_acc: 0.9317\n",
      "Epoch 19/30\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.1538 - acc: 0.9448 - val_loss: 0.2898 - val_acc: 0.9143\n",
      "Epoch 20/30\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 0.1363 - acc: 0.9548 - val_loss: 0.2134 - val_acc: 0.9296\n",
      "Epoch 21/30\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.1348 - acc: 0.9543 - val_loss: 0.2169 - val_acc: 0.9252\n",
      "Epoch 22/30\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 0.1147 - acc: 0.9586 - val_loss: 0.2703 - val_acc: 0.9194\n",
      "Epoch 23/30\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 0.1214 - acc: 0.9545 - val_loss: 0.2947 - val_acc: 0.9208\n",
      "Epoch 24/30\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.1012 - acc: 0.9677 - val_loss: 0.2196 - val_acc: 0.9303\n",
      "Epoch 25/30\n",
      "111/110 [==============================] - 84s 758ms/step - loss: 0.1076 - acc: 0.9614 - val_loss: 0.2634 - val_acc: 0.9237\n",
      "Epoch 26/30\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 0.1074 - acc: 0.9639 - val_loss: 0.2088 - val_acc: 0.9368\n",
      "Epoch 27/30\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.1199 - acc: 0.9603 - val_loss: 0.2450 - val_acc: 0.9259\n",
      "Epoch 28/30\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.0990 - acc: 0.9663 - val_loss: 0.2207 - val_acc: 0.9288\n",
      "Epoch 29/30\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.0912 - acc: 0.9708 - val_loss: 0.3458 - val_acc: 0.9049\n",
      "Epoch 30/30\n",
      "111/110 [==============================] - 84s 761ms/step - loss: 0.0841 - acc: 0.9719 - val_loss: 0.2208 - val_acc: 0.9383\n"
     ]
    }
   ],
   "source": [
    "#epochSize edited\n",
    "epochSize = 30\n",
    "\n",
    "Adadelta_history = model.fit_generator(\n",
    "        train_generator,\n",
    "        steps_per_epoch=(5522 / batchSize),\n",
    "        epochs=epochSize,\n",
    "        validation_data=validation_generator,\n",
    "        validation_steps=(1377 / batchSize))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Adadelta_history.history.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "Adadelta_loss = Adadelta_history.history['loss']\n",
    "Adadelta_val_loss = Adadelta_history.history['val_loss']\n",
    "Adadelta_acc = Adadelta_history.history['acc']\n",
    "Adadelta_val_acc = Adadelta_history.history['val_acc']\n",
    "Adadelta_epochs = range(len(Adadelta_acc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(Adadelta_epochs, Adadelta_acc, 'red', label='Training Acc')\n",
    "plt.plot(Adadelta_epochs, Adadelta_val_acc, 'blue', label='Validation Acc')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('Acc')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(Adadelta_epochs, Adadelta_loss, 'red', label='Training Loss')\n",
    "plt.plot(Adadelta_epochs, Adadelta_val_loss, 'blue', label='Validation Loss')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model preperation for Adamax"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model initializing\n",
    "model = Sequential()\n",
    "\n",
    "#adding conv2d and maxpool2d layers\n",
    "model.add(Conv2D(32, (3,3), activation='relu', input_shape=(dim1, dim2, 3)))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "#2nd layer addition\n",
    "model.add(Conv2D(64, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Conv2D(128, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Flatten())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "#hidden layers\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(rate=0.4))\n",
    "\n",
    "model.add(Dense(512, activation='relu'))\n",
    "model.add(Dropout(rate=0.3))\n",
    "\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(rate=0.2))\n",
    "\n",
    "#output\n",
    "model.add(Dense(8, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model compilation - Adamax\n",
    "model.compile(optimizer='Adamax', loss='categorical_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "111/110 [==============================] - 93s 834ms/step - loss: 1.5000 - acc: 0.4779 - val_loss: 0.7960 - val_acc: 0.7052\n",
      "Epoch 2/30\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 0.6150 - acc: 0.7752 - val_loss: 0.5974 - val_acc: 0.7894\n",
      "Epoch 3/30\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 0.4711 - acc: 0.8200 - val_loss: 0.5404 - val_acc: 0.8097\n",
      "Epoch 4/30\n",
      "111/110 [==============================] - 84s 758ms/step - loss: 0.3913 - acc: 0.8477 - val_loss: 0.3746 - val_acc: 0.8577\n",
      "Epoch 5/30\n",
      "111/110 [==============================] - 85s 762ms/step - loss: 0.3295 - acc: 0.8695 - val_loss: 0.3278 - val_acc: 0.8853\n",
      "Epoch 6/30\n",
      "111/110 [==============================] - 84s 757ms/step - loss: 0.3114 - acc: 0.8782 - val_loss: 0.4094 - val_acc: 0.8664\n",
      "Epoch 7/30\n",
      "111/110 [==============================] - 84s 756ms/step - loss: 0.2850 - acc: 0.8932 - val_loss: 0.3091 - val_acc: 0.8925\n",
      "Epoch 8/30\n",
      "111/110 [==============================] - 84s 760ms/step - loss: 0.2506 - acc: 0.9017 - val_loss: 0.2581 - val_acc: 0.9085\n",
      "Epoch 9/30\n",
      "111/110 [==============================] - 96s 861ms/step - loss: 0.2333 - acc: 0.9098 - val_loss: 0.3023 - val_acc: 0.9034\n",
      "Epoch 10/30\n",
      "111/110 [==============================] - 117s 1s/step - loss: 0.2109 - acc: 0.9154 - val_loss: 0.2649 - val_acc: 0.9056\n",
      "Epoch 11/30\n",
      "111/110 [==============================] - 88s 793ms/step - loss: 0.1870 - acc: 0.9281 - val_loss: 0.2348 - val_acc: 0.9143\n",
      "Epoch 12/30\n",
      "111/110 [==============================] - 87s 783ms/step - loss: 0.1845 - acc: 0.9309 - val_loss: 0.2548 - val_acc: 0.9179\n",
      "Epoch 13/30\n",
      "111/110 [==============================] - 87s 781ms/step - loss: 0.1680 - acc: 0.9358 - val_loss: 0.2571 - val_acc: 0.9121\n",
      "Epoch 14/30\n",
      "111/110 [==============================] - 85s 769ms/step - loss: 0.1688 - acc: 0.9369 - val_loss: 0.2763 - val_acc: 0.9143\n",
      "Epoch 15/30\n",
      "111/110 [==============================] - 86s 771ms/step - loss: 0.1581 - acc: 0.9424 - val_loss: 0.2221 - val_acc: 0.9296\n",
      "Epoch 16/30\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.1274 - acc: 0.9517 - val_loss: 0.2352 - val_acc: 0.9216\n",
      "Epoch 17/30\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.1337 - acc: 0.9470 - val_loss: 0.2381 - val_acc: 0.9267\n",
      "Epoch 18/30\n",
      "111/110 [==============================] - 85s 763ms/step - loss: 0.1238 - acc: 0.9540 - val_loss: 0.1920 - val_acc: 0.9390\n",
      "Epoch 19/30\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.1322 - acc: 0.9499 - val_loss: 0.1682 - val_acc: 0.9455\n",
      "Epoch 20/30\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.1258 - acc: 0.9513 - val_loss: 0.1685 - val_acc: 0.9455\n",
      "Epoch 21/30\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.1092 - acc: 0.9607 - val_loss: 0.2851 - val_acc: 0.9172\n",
      "Epoch 22/30\n",
      "111/110 [==============================] - 102s 916ms/step - loss: 0.1028 - acc: 0.9623 - val_loss: 0.2498 - val_acc: 0.9208\n",
      "Epoch 23/30\n",
      "111/110 [==============================] - 109s 978ms/step - loss: 0.1153 - acc: 0.9555 - val_loss: 0.2058 - val_acc: 0.9259\n",
      "Epoch 24/30\n",
      "111/110 [==============================] - 88s 792ms/step - loss: 0.1014 - acc: 0.9605 - val_loss: 0.1846 - val_acc: 0.9346\n",
      "Epoch 25/30\n",
      "111/110 [==============================] - 87s 780ms/step - loss: 0.0971 - acc: 0.9639 - val_loss: 0.2101 - val_acc: 0.9346\n",
      "Epoch 26/30\n",
      "111/110 [==============================] - 85s 769ms/step - loss: 0.1003 - acc: 0.9618 - val_loss: 0.2376 - val_acc: 0.9223\n",
      "Epoch 27/30\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.0988 - acc: 0.9645 - val_loss: 0.1985 - val_acc: 0.9325\n",
      "Epoch 28/30\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.0801 - acc: 0.9721 - val_loss: 0.2157 - val_acc: 0.9390\n",
      "Epoch 29/30\n",
      "111/110 [==============================] - 86s 776ms/step - loss: 0.0732 - acc: 0.9732 - val_loss: 0.2253 - val_acc: 0.9325\n",
      "Epoch 30/30\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.0779 - acc: 0.9726 - val_loss: 0.1708 - val_acc: 0.9448\n"
     ]
    }
   ],
   "source": [
    "Adamax_history = model.fit_generator(\n",
    "        train_generator,\n",
    "        steps_per_epoch=(5522 / batchSize),\n",
    "        epochs=epochSize,\n",
    "        validation_data=validation_generator,\n",
    "        validation_steps=(1377 / batchSize))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Adamax_history.history.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "Adamax_loss = Adamax_history.history['loss']\n",
    "Adamax_val_loss = Adamax_history.history['val_loss']\n",
    "Adamax_acc = Adamax_history.history['acc']\n",
    "Adamax_val_acc = Adamax_history.history['val_acc']\n",
    "Adamax_epochs = range(len(Adamax_acc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(Adamax_epochs, Adamax_acc, 'red', label='Training Acc')\n",
    "plt.plot(Adamax_epochs, Adamax_val_acc, 'blue', label='Validation Acc')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('Acc')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(Adamax_epochs, Adamax_loss, 'red', label='Training Loss')\n",
    "plt.plot(Adamax_epochs, Adamax_val_loss, 'blue', label='Validation Loss')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model preperation for Nadam"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model initializing\n",
    "model = Sequential()\n",
    "\n",
    "#adding conv2d and maxpool2d layers\n",
    "model.add(Conv2D(32, (3,3), activation='relu', input_shape=(dim1, dim2, 3)))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "#2nd layer addition\n",
    "model.add(Conv2D(64, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Conv2D(128, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Flatten())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "#hidden layers\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(rate=0.4))\n",
    "\n",
    "model.add(Dense(512, activation='relu'))\n",
    "model.add(Dropout(rate=0.3))\n",
    "\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(rate=0.2))\n",
    "\n",
    "#output\n",
    "model.add(Dense(8, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model compilation - Nadam\n",
    "model.compile(optimizer='Nadam', loss='categorical_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "111/110 [==============================] - 93s 840ms/step - loss: 2.0312 - acc: 0.3260 - val_loss: 1.7757 - val_acc: 0.4459\n",
      "Epoch 2/30\n",
      "111/110 [==============================] - 86s 772ms/step - loss: 1.3049 - acc: 0.5545 - val_loss: 0.8245 - val_acc: 0.7124\n",
      "Epoch 3/30\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.8390 - acc: 0.6945 - val_loss: 0.6714 - val_acc: 0.7778\n",
      "Epoch 4/30\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.6113 - acc: 0.7774 - val_loss: 0.5293 - val_acc: 0.8061\n",
      "Epoch 5/30\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.5673 - acc: 0.7959 - val_loss: 0.6892 - val_acc: 0.7393\n",
      "Epoch 6/30\n",
      "111/110 [==============================] - 86s 774ms/step - loss: 0.5245 - acc: 0.8090 - val_loss: 0.5358 - val_acc: 0.8257\n",
      "Epoch 7/30\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.3940 - acc: 0.8513 - val_loss: 0.3647 - val_acc: 0.8656\n",
      "Epoch 8/30\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.3542 - acc: 0.8660 - val_loss: 0.3340 - val_acc: 0.8816\n",
      "Epoch 9/30\n",
      "111/110 [==============================] - 86s 774ms/step - loss: 0.3290 - acc: 0.8752 - val_loss: 0.3986 - val_acc: 0.8577\n",
      "Epoch 10/30\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.3030 - acc: 0.8885 - val_loss: 0.3000 - val_acc: 0.8903\n",
      "Epoch 11/30\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.3115 - acc: 0.8936 - val_loss: 0.3906 - val_acc: 0.8722\n",
      "Epoch 12/30\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.2993 - acc: 0.8910 - val_loss: 0.2692 - val_acc: 0.9063\n",
      "Epoch 13/30\n",
      "111/110 [==============================] - 86s 774ms/step - loss: 0.2643 - acc: 0.9060 - val_loss: 0.3837 - val_acc: 0.8744\n",
      "Epoch 14/30\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 1.3284 - acc: 0.6675 - val_loss: 0.3880 - val_acc: 0.8649\n",
      "Epoch 15/30\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.3812 - acc: 0.8679 - val_loss: 0.3829 - val_acc: 0.8860\n",
      "Epoch 16/30\n",
      "111/110 [==============================] - 86s 771ms/step - loss: 0.2756 - acc: 0.9047 - val_loss: 0.3660 - val_acc: 0.8809\n",
      "Epoch 17/30\n",
      "111/110 [==============================] - 85s 770ms/step - loss: 0.2784 - acc: 0.9069 - val_loss: 0.3220 - val_acc: 0.9012\n",
      "Epoch 18/30\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.2496 - acc: 0.9156 - val_loss: 0.2616 - val_acc: 0.9099\n",
      "Epoch 19/30\n",
      "111/110 [==============================] - 85s 767ms/step - loss: 0.1938 - acc: 0.9276 - val_loss: 0.2553 - val_acc: 0.9165\n",
      "Epoch 20/30\n",
      "111/110 [==============================] - 86s 772ms/step - loss: 0.2218 - acc: 0.9258 - val_loss: 0.2707 - val_acc: 0.9129\n",
      "Epoch 21/30\n",
      "111/110 [==============================] - 85s 769ms/step - loss: 0.1920 - acc: 0.9328 - val_loss: 0.2516 - val_acc: 0.9267\n",
      "Epoch 22/30\n",
      "111/110 [==============================] - 85s 769ms/step - loss: 0.1927 - acc: 0.9294 - val_loss: 0.2177 - val_acc: 0.9310\n",
      "Epoch 23/30\n",
      "111/110 [==============================] - 85s 764ms/step - loss: 0.1760 - acc: 0.9381 - val_loss: 0.3670 - val_acc: 0.8809\n",
      "Epoch 24/30\n",
      "111/110 [==============================] - 86s 774ms/step - loss: 0.1894 - acc: 0.9352 - val_loss: 0.2496 - val_acc: 0.9201\n",
      "Epoch 25/30\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.1643 - acc: 0.9464 - val_loss: 0.2791 - val_acc: 0.9078\n",
      "Epoch 26/30\n",
      "111/110 [==============================] - 85s 765ms/step - loss: 0.1428 - acc: 0.9490 - val_loss: 0.2334 - val_acc: 0.9208\n",
      "Epoch 27/30\n",
      "111/110 [==============================] - 85s 769ms/step - loss: 0.1807 - acc: 0.9416 - val_loss: 0.2230 - val_acc: 0.9332\n",
      "Epoch 28/30\n",
      "111/110 [==============================] - 85s 766ms/step - loss: 0.1448 - acc: 0.9486 - val_loss: 0.2203 - val_acc: 0.9317\n",
      "Epoch 29/30\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.1463 - acc: 0.9526 - val_loss: 0.2399 - val_acc: 0.9259\n",
      "Epoch 30/30\n",
      "111/110 [==============================] - 85s 768ms/step - loss: 0.1483 - acc: 0.9497 - val_loss: 0.2904 - val_acc: 0.8954\n"
     ]
    }
   ],
   "source": [
    "Nadam_history = model.fit_generator(\n",
    "        train_generator,\n",
    "        steps_per_epoch=(5522 / batchSize),\n",
    "        epochs=epochSize,\n",
    "        validation_data=validation_generator,\n",
    "        validation_steps=(1377 / batchSize))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Nadam_history.history.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "Nadam_loss = Nadam_history.history['loss']\n",
    "Nadam_val_loss = Nadam_history.history['val_loss']\n",
    "Nadam_acc = Nadam_history.history['acc']\n",
    "Nadam_val_acc = Nadam_history.history['val_acc']\n",
    "Nadam_epochs = range(len(Nadam_acc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(Nadam_epochs, Nadam_acc, 'red', label='Training Acc')\n",
    "plt.plot(Nadam_epochs, Nadam_val_acc, 'blue', label='Validation Acc')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('Acc')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(Nadam_epochs, Nadam_loss, 'red', label='Training Loss')\n",
    "plt.plot(Nadam_epochs, Nadam_val_loss, 'blue', label='Validation Loss')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# END-OF-CODE"
   ]
  }
 ],
 "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}