fashion-mnist-with-cnn.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "c317af2a",
   "metadata": {
    "id": "7QoLlSiK-PzT",
    "papermill": {
     "duration": 0.020902,
     "end_time": "2022-02-21T07:36:43.923004",
     "exception": false,
     "start_time": "2022-02-21T07:36:43.902102",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Fashion MNIST Classification using CNN Model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6ef068b6",
   "metadata": {
    "id": "BYA4TbWE3P26",
    "papermill": {
     "duration": 0.029372,
     "end_time": "2022-02-21T07:36:43.986373",
     "exception": false,
     "start_time": "2022-02-21T07:36:43.957001",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Prerequisites"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "110fb489",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:36:44.051157Z",
     "iopub.status.busy": "2022-02-21T07:36:44.050182Z",
     "iopub.status.idle": "2022-02-21T07:36:49.350276Z",
     "shell.execute_reply": "2022-02-21T07:36:49.349646Z",
     "shell.execute_reply.started": "2022-02-21T07:33:10.059159Z"
    },
    "id": "baMpbTkJmnkg",
    "papermill": {
     "duration": 5.338434,
     "end_time": "2022-02-21T07:36:49.350434",
     "exception": false,
     "start_time": "2022-02-21T07:36:44.012000",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "90697c4b",
   "metadata": {
    "id": "O9wUBEb84Ypw",
    "papermill": {
     "duration": 0.018267,
     "end_time": "2022-02-21T07:36:49.387866",
     "exception": false,
     "start_time": "2022-02-21T07:36:49.369599",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Data Loading and Preprocess"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "f488afd3",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:36:49.429300Z",
     "iopub.status.busy": "2022-02-21T07:36:49.428490Z",
     "iopub.status.idle": "2022-02-21T07:36:50.184724Z",
     "shell.execute_reply": "2022-02-21T07:36:50.185431Z",
     "shell.execute_reply.started": "2022-02-21T07:33:10.069068Z"
    },
    "id": "FaHs20IQng_D",
    "outputId": "d02ba8f9-c5b3-459a-ff0a-6328473b91f3",
    "papermill": {
     "duration": 0.778645,
     "end_time": "2022-02-21T07:36:50.185588",
     "exception": false,
     "start_time": "2022-02-21T07:36:49.406943",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-labels-idx1-ubyte.gz\n",
      "32768/29515 [=================================] - 0s 0us/step\n",
      "40960/29515 [=========================================] - 0s 0us/step\n",
      "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-images-idx3-ubyte.gz\n",
      "26427392/26421880 [==============================] - 0s 0us/step\n",
      "26435584/26421880 [==============================] - 0s 0us/step\n",
      "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-labels-idx1-ubyte.gz\n",
      "16384/5148 [===============================================================================================] - 0s 0us/step\n",
      "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-images-idx3-ubyte.gz\n",
      "4423680/4422102 [==============================] - 0s 0us/step\n",
      "4431872/4422102 [==============================] - 0s 0us/step\n",
      "Shape of X_train: (60000, 28, 28)\tShape of X_test: (10000, 28, 28)\n",
      "Shape of y_train: (60000,)\t\tShape of y_test: (10000,)\n"
     ]
    }
   ],
   "source": [
    "# Load and divide Fashion MNIST dataset\n",
    "(X_train, y_train), (X_test, y_test) = keras.datasets.fashion_mnist.load_data()\n",
    "\n",
    "print('Shape of X_train: {}\\tShape of X_test: {}'.format(X_train.shape, X_test.shape))\n",
    "print('Shape of y_train: {}\\t\\tShape of y_test: {}'.format(y_train.shape, y_test.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "dde35eaf",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:36:52.536271Z",
     "iopub.status.busy": "2022-02-21T07:36:52.241552Z",
     "iopub.status.idle": "2022-02-21T07:36:52.728487Z",
     "shell.execute_reply": "2022-02-21T07:36:52.728033Z",
     "shell.execute_reply.started": "2022-02-21T07:33:10.480255Z"
    },
    "id": "T0T07aa5nt-B",
    "outputId": "e7c0b6af-894a-4487-f6cd-e753cec4a1e7",
    "papermill": {
     "duration": 2.518153,
     "end_time": "2022-02-21T07:36:52.728605",
     "exception": false,
     "start_time": "2022-02-21T07:36:50.210452",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-02-21 07:36:50.293702: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2022-02-21 07:36:50.391925: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2022-02-21 07:36:50.392646: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2022-02-21 07:36:50.395208: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 AVX512F FMA\n",
      "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "2022-02-21 07:36:50.396198: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2022-02-21 07:36:50.396961: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2022-02-21 07:36:50.397641: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2022-02-21 07:36:52.232431: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2022-02-21 07:36:52.233275: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2022-02-21 07:36:52.233933: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero\n",
      "2022-02-21 07:36:52.234520: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1510] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 15403 MB memory:  -> device: 0, name: Tesla P100-PCIE-16GB, pci bus id: 0000:00:04.0, compute capability: 6.0\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "(28, 28)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Generate a random number and use it as index\n",
    "rand_index = tf.random.uniform(shape=(), minval=1, maxval=60000, dtype=tf.int32)\n",
    "\n",
    "plt.imshow(X_train[rand_index])\n",
    "plt.title('Label: {}'.format(y_train[rand_index]))\n",
    "plt.show()\n",
    "X_train[rand_index].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "8b5cbb88",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:36:52.782809Z",
     "iopub.status.busy": "2022-02-21T07:36:52.782196Z",
     "iopub.status.idle": "2022-02-21T07:36:52.786553Z",
     "shell.execute_reply": "2022-02-21T07:36:52.786965Z",
     "shell.execute_reply.started": "2022-02-21T07:33:10.658388Z"
    },
    "id": "apPxd-VKtJL9",
    "outputId": "f1ecfa7f-9ce2-47a9-987f-0cd3808fd33e",
    "papermill": {
     "duration": 0.034477,
     "end_time": "2022-02-21T07:36:52.787100",
     "exception": false,
     "start_time": "2022-02-21T07:36:52.752623",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dtype = uint8\n",
      "min = 0\n",
      "max = 255\n",
      "range = 255\n"
     ]
    }
   ],
   "source": [
    "# Print some properties of X_train and X_test\n",
    "print('dtype =', X_train[rand_index].dtype)\n",
    "print('min =', X_train[rand_index].min())\n",
    "print('max =', X_train[rand_index].max())\n",
    "print('range =', X_train[rand_index].ptp())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "5d9e414f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:36:52.844489Z",
     "iopub.status.busy": "2022-02-21T07:36:52.843399Z",
     "iopub.status.idle": "2022-02-21T07:36:52.964752Z",
     "shell.execute_reply": "2022-02-21T07:36:52.964097Z",
     "shell.execute_reply.started": "2022-02-21T07:33:10.668928Z"
    },
    "id": "xr4hPZottY1L",
    "papermill": {
     "duration": 0.153209,
     "end_time": "2022-02-21T07:36:52.964893",
     "exception": false,
     "start_time": "2022-02-21T07:36:52.811684",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Type casting to floating point \n",
    "X_train = X_train.astype('float32')\n",
    "X_test = X_test.astype('float32')\n",
    "\n",
    "# Scale the data\n",
    "X_train = X_train / 255\n",
    "X_test = X_test / 255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "120b76ad",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:36:53.021647Z",
     "iopub.status.busy": "2022-02-21T07:36:53.016299Z",
     "iopub.status.idle": "2022-02-21T07:36:53.046997Z",
     "shell.execute_reply": "2022-02-21T07:36:53.046563Z",
     "shell.execute_reply.started": "2022-02-21T07:33:10.796516Z"
    },
    "id": "vZ7IjS8cwhL9",
    "outputId": "a87303c6-ee03-4f1f-c551-dcead3ad86d9",
    "papermill": {
     "duration": 0.057519,
     "end_time": "2022-02-21T07:36:53.047110",
     "exception": false,
     "start_time": "2022-02-21T07:36:52.989591",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Label of sample #53068 is [1. 0. 0. 0. 0. 0. 0. 0. 0. 0.].\n"
     ]
    }
   ],
   "source": [
    "# Apply one hot encoder\n",
    "y_train = tf.one_hot(y_train, 10)\n",
    "\n",
    "print('Label of sample #{} is {}.'.format(\n",
    "    rand_index,\n",
    "    y_train[rand_index]\n",
    "))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5e61c6f5",
   "metadata": {
    "id": "N6gDjRI34igk",
    "papermill": {
     "duration": 0.024214,
     "end_time": "2022-02-21T07:36:53.095765",
     "exception": false,
     "start_time": "2022-02-21T07:36:53.071551",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Modeling "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ebf81e5e",
   "metadata": {
    "id": "l_yBsvAw4qa7",
    "papermill": {
     "duration": 0.024487,
     "end_time": "2022-02-21T07:36:53.144838",
     "exception": false,
     "start_time": "2022-02-21T07:36:53.120351",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Custom Callback"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "a521e0a9",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:36:53.201181Z",
     "iopub.status.busy": "2022-02-21T07:36:53.200423Z",
     "iopub.status.idle": "2022-02-21T07:36:53.202879Z",
     "shell.execute_reply": "2022-02-21T07:36:53.202451Z",
     "shell.execute_reply.started": "2022-02-21T07:33:10.806885Z"
    },
    "id": "QFBN0l79pKaL",
    "papermill": {
     "duration": 0.032836,
     "end_time": "2022-02-21T07:36:53.202984",
     "exception": false,
     "start_time": "2022-02-21T07:36:53.170148",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# create class for custom callbacks\n",
    "class ThresholdCallback(keras.callbacks.Callback):\n",
    "    \"\"\"\n",
    "    Threshold Callback for keras.models\n",
    "\n",
    "    Parameters\n",
    "    ------------\n",
    "    point: float (between 0 and 1)\n",
    "    \"\"\"\n",
    "    def __init__(self, point):\n",
    "        super(ThresholdCallback, self).__init__()\n",
    "        # Threshhold for stop training process if it become reached\n",
    "        self.point = point\n",
    "\n",
    "    def on_batch_end(self, epoch, logs=None):\n",
    "        # Check for reaching specified point\n",
    "        if logs['accuracy'] >= self.point:\n",
    "            self.model.stop_training = True"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "606ca993",
   "metadata": {
    "id": "8Aw7J1mG4uLH",
    "papermill": {
     "duration": 0.024805,
     "end_time": "2022-02-21T07:36:53.252783",
     "exception": false,
     "start_time": "2022-02-21T07:36:53.227978",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## CNN Model with RMSProp Optimizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "6d9e7403",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:36:53.308425Z",
     "iopub.status.busy": "2022-02-21T07:36:53.307683Z",
     "iopub.status.idle": "2022-02-21T07:36:53.387648Z",
     "shell.execute_reply": "2022-02-21T07:36:53.387056Z",
     "shell.execute_reply.started": "2022-02-21T07:33:10.815322Z"
    },
    "id": "EA2Y15nUsn8b",
    "outputId": "a1b3b22b-2bf0-4850-e4b6-b99343943665",
    "papermill": {
     "duration": 0.11026,
     "end_time": "2022-02-21T07:36:53.387772",
     "exception": false,
     "start_time": "2022-02-21T07:36:53.277512",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 26, 26, 64)        640       \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 13, 13, 64)        0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 10816)             0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 100)               1081700   \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 10)                1010      \n",
      "=================================================================\n",
      "Total params: 1,083,350\n",
      "Trainable params: 1,083,350\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Define CNN model\n",
    "model = keras.models.Sequential(\n",
    "    [\n",
    "     layers.Conv2D(64, (3,3), activation='relu', input_shape=(28, 28, 1)),\n",
    "     layers.MaxPooling2D((2, 2)),\n",
    "     layers.Flatten(),\n",
    "     layers.Dense(100),\n",
    "     layers.Dense(10, activation='sigmoid')\n",
    "    ]\n",
    ")\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "ae54ebcc",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:36:53.445708Z",
     "iopub.status.busy": "2022-02-21T07:36:53.444918Z",
     "iopub.status.idle": "2022-02-21T07:36:53.451321Z",
     "shell.execute_reply": "2022-02-21T07:36:53.450787Z",
     "shell.execute_reply.started": "2022-02-21T07:33:10.855758Z"
    },
    "id": "yHPiazJFucyH",
    "papermill": {
     "duration": 0.038396,
     "end_time": "2022-02-21T07:36:53.451430",
     "exception": false,
     "start_time": "2022-02-21T07:36:53.413034",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Comile model\n",
    "model.compile(\n",
    "    optimizer='rmsprop',\n",
    "    loss='categorical_crossentropy',\n",
    "    metrics=['accuracy'],\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "62a0faab",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:36:53.508620Z",
     "iopub.status.busy": "2022-02-21T07:36:53.507998Z",
     "iopub.status.idle": "2022-02-21T07:37:06.545960Z",
     "shell.execute_reply": "2022-02-21T07:37:06.545497Z",
     "shell.execute_reply.started": "2022-02-21T07:33:10.867191Z"
    },
    "id": "FxHft3xau2C1",
    "outputId": "f7703a2e-dfb9-4c04-b3a0-e4d8ece00bfc",
    "papermill": {
     "duration": 13.069489,
     "end_time": "2022-02-21T07:37:06.546096",
     "exception": false,
     "start_time": "2022-02-21T07:36:53.476607",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-02-21 07:36:53.820262: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:185] None of the MLIR Optimization Passes are enabled (registered 2)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-02-21 07:36:55.114798: I tensorflow/stream_executor/cuda/cuda_dnn.cc:369] Loaded cuDNN version 8005\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "750/750 [==============================] - 9s 4ms/step - loss: 0.4124 - accuracy: 0.8537 - val_loss: 0.3107 - val_accuracy: 0.8889\n",
      "Epoch 2/10\n",
      "750/750 [==============================] - 3s 3ms/step - loss: 0.2843 - accuracy: 0.8999 - val_loss: 0.2812 - val_accuracy: 0.9022\n",
      "Epoch 3/10\n",
      "750/750 [==============================] - 1s 852us/step - loss: 0.1331 - accuracy: 0.9531 - val_loss: 0.2785 - val_accuracy: 0.9013\n"
     ]
    }
   ],
   "source": [
    "# Justify dimension for model use\n",
    "X_train = np.expand_dims(X_train, axis=-1)\n",
    "\n",
    "# Fit model to data with cutstom callbacks\n",
    "history = model.fit(\n",
    "    X_train, y_train,\n",
    "    batch_size=64,\n",
    "    epochs=10,\n",
    "    callbacks=ThresholdCallback(0.95),\n",
    "    validation_split=0.2,\n",
    "    verbose=1\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "06e1f5c0",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:37:06.676251Z",
     "iopub.status.busy": "2022-02-21T07:37:06.675693Z",
     "iopub.status.idle": "2022-02-21T07:37:15.153217Z",
     "shell.execute_reply": "2022-02-21T07:37:15.153724Z",
     "shell.execute_reply.started": "2022-02-21T07:33:29.122802Z"
    },
    "id": "vbQNpMbqDNv8",
    "outputId": "558dd46d-4eb7-42c0-8131-b475ea50f373",
    "papermill": {
     "duration": 8.547483,
     "end_time": "2022-02-21T07:37:15.153900",
     "exception": false,
     "start_time": "2022-02-21T07:37:06.606417",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Ploting accuracy values\n",
    "plt.figure(figsize=(20,8))\n",
    "plt.subplot(1, 2, 1)\n",
    "plt.plot(history.history['accuracy'])\n",
    "plt.plot(history.history['val_accuracy'])\n",
    "plt.title('Accuracy')\n",
    "plt.legend(['Train', 'Validation'])\n",
    "plt.xlabel('Epoch')\n",
    "\n",
    "# Ploting loss values\n",
    "plt.subplot(1, 2, 2)\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('Loss')\n",
    "plt.legend(['Train', 'Validation'])\n",
    "plt.xlabel('Epoch')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1ef6ee99",
   "metadata": {
    "id": "sGM1KFg44-Wp",
    "papermill": {
     "duration": 0.056909,
     "end_time": "2022-02-21T07:37:15.269557",
     "exception": false,
     "start_time": "2022-02-21T07:37:15.212648",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## CNN Model with Adam Optimizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "e2e06143",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:37:15.394509Z",
     "iopub.status.busy": "2022-02-21T07:37:15.393655Z",
     "iopub.status.idle": "2022-02-21T07:37:15.423638Z",
     "shell.execute_reply": "2022-02-21T07:37:15.424209Z",
     "shell.execute_reply.started": "2022-02-21T07:33:29.442998Z"
    },
    "id": "EN1Lmuie5QHu",
    "outputId": "efa81915-bc40-49c4-8da6-b1709fe7325c",
    "papermill": {
     "duration": 0.098418,
     "end_time": "2022-02-21T07:37:15.424409",
     "exception": false,
     "start_time": "2022-02-21T07:37:15.325991",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 26, 26, 64)        640       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 13, 13, 64)        0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 10816)             0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 100)               1081700   \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 10)                1010      \n",
      "=================================================================\n",
      "Total params: 1,083,350\n",
      "Trainable params: 1,083,350\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Define CNN model\n",
    "model = keras.models.Sequential(\n",
    "    [\n",
    "     layers.Conv2D(64, (3,3), activation='relu', input_shape=(28, 28, 1)),\n",
    "     layers.MaxPooling2D((2, 2)),\n",
    "     layers.Flatten(),\n",
    "     layers.Dense(100),\n",
    "     layers.Dense(10, activation='sigmoid')\n",
    "    ]\n",
    ")\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "39d80db3",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:37:15.554292Z",
     "iopub.status.busy": "2022-02-21T07:37:15.553510Z",
     "iopub.status.idle": "2022-02-21T07:37:15.561735Z",
     "shell.execute_reply": "2022-02-21T07:37:15.561314Z",
     "shell.execute_reply.started": "2022-02-21T07:33:29.481637Z"
    },
    "id": "pjMg7DDK5QHv",
    "papermill": {
     "duration": 0.074816,
     "end_time": "2022-02-21T07:37:15.561856",
     "exception": false,
     "start_time": "2022-02-21T07:37:15.487040",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Comile model\n",
    "model.compile(\n",
    "    optimizer='adam',\n",
    "    loss='categorical_crossentropy',\n",
    "    metrics=['accuracy'],\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "f45bc65b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:37:15.686114Z",
     "iopub.status.busy": "2022-02-21T07:37:15.685246Z",
     "iopub.status.idle": "2022-02-21T07:37:25.436435Z",
     "shell.execute_reply": "2022-02-21T07:37:25.435874Z",
     "shell.execute_reply.started": "2022-02-21T07:33:29.492226Z"
    },
    "id": "o3Cdxbac5QHw",
    "outputId": "08aee039-ddbb-4ebb-f109-6bf7907a9267",
    "papermill": {
     "duration": 9.816237,
     "end_time": "2022-02-21T07:37:25.436574",
     "exception": false,
     "start_time": "2022-02-21T07:37:15.620337",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "750/750 [==============================] - 4s 5ms/step - loss: 0.4028 - accuracy: 0.8587 - val_loss: 0.3124 - val_accuracy: 0.8895\n",
      "Epoch 2/10\n",
      "750/750 [==============================] - 2s 3ms/step - loss: 0.2860 - accuracy: 0.8991 - val_loss: 0.2858 - val_accuracy: 0.8979\n",
      "Epoch 3/10\n",
      "750/750 [==============================] - 2s 3ms/step - loss: 0.2509 - accuracy: 0.9109 - val_loss: 0.2806 - val_accuracy: 0.9023\n",
      "Epoch 4/10\n",
      "750/750 [==============================] - 0s 547us/step - loss: 0.1027 - accuracy: 0.9531 - val_loss: 0.2901 - val_accuracy: 0.9007\n"
     ]
    }
   ],
   "source": [
    "# Fit model to data with cutstom callbacks\n",
    "history = model.fit(\n",
    "    X_train, y_train,\n",
    "    batch_size=64,\n",
    "    epochs=10,\n",
    "    callbacks=ThresholdCallback(0.95),\n",
    "    validation_split=0.2,\n",
    "    verbose=1\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "6be50a83",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:37:25.636659Z",
     "iopub.status.busy": "2022-02-21T07:37:25.635804Z",
     "iopub.status.idle": "2022-02-21T07:37:25.951819Z",
     "shell.execute_reply": "2022-02-21T07:37:25.952506Z",
     "shell.execute_reply.started": "2022-02-21T07:33:38.329770Z"
    },
    "id": "ePKP5btP5QHw",
    "outputId": "43935c9e-c01f-4190-d2ff-11819aea74ed",
    "papermill": {
     "duration": 0.420211,
     "end_time": "2022-02-21T07:37:25.952660",
     "exception": false,
     "start_time": "2022-02-21T07:37:25.532449",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Ploting accuracy values\n",
    "plt.figure(figsize=(20,8))\n",
    "plt.subplot(1, 2, 1)\n",
    "plt.plot(history.history['accuracy'])\n",
    "plt.plot(history.history['val_accuracy'])\n",
    "plt.title('Accuracy')\n",
    "plt.legend(['Train', 'Validation'])\n",
    "plt.xlabel('Epoch')\n",
    "\n",
    "# Ploting loss values\n",
    "plt.subplot(1, 2, 2)\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('Loss')\n",
    "plt.legend(['Train', 'Validation'])\n",
    "plt.xlabel('Epoch')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "de57c07b",
   "metadata": {
    "id": "uGdGCyWV8LcO",
    "papermill": {
     "duration": 0.097379,
     "end_time": "2022-02-21T07:37:26.148121",
     "exception": false,
     "start_time": "2022-02-21T07:37:26.050742",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "As we can see in the above figures, model with Adam optimizer reached 0.95 accuracy sooner than RMSProp optimizer. On the other hand, the difference between train accuracy and validation accuracy is larger using Adam which could be a reason for tendention of this model to overfitting the training data."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "88ba72b6",
   "metadata": {
    "id": "QOzqcb8K1eTY",
    "papermill": {
     "duration": 0.096491,
     "end_time": "2022-02-21T07:37:26.343644",
     "exception": false,
     "start_time": "2022-02-21T07:37:26.247153",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "# Test Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "8f7c7a6a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-02-21T07:37:26.545384Z",
     "iopub.status.busy": "2022-02-21T07:37:26.544409Z",
     "iopub.status.idle": "2022-02-21T07:37:27.311577Z",
     "shell.execute_reply": "2022-02-21T07:37:27.310967Z",
     "shell.execute_reply.started": "2022-02-21T07:33:40.906093Z"
    },
    "id": "oAUGchck1iHM",
    "outputId": "53bc4d49-8b47-4c00-8bdf-9f38e751eb66",
    "papermill": {
     "duration": 0.871283,
     "end_time": "2022-02-21T07:37:27.311712",
     "exception": false,
     "start_time": "2022-02-21T07:37:26.440429",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test set: 0.8937\n"
     ]
    }
   ],
   "source": [
    "output = model.predict(np.expand_dims(X_test, axis=-1))\n",
    "preds = np.argmax(output, axis=1)\n",
    "\n",
    "print('Accuracy on test set: {}'.format(\n",
    "    np.sum(preds==y_test) / len(y_test)\n",
    "    )\n",
    ")"
   ]
  }
 ],
 "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": "3.7.12"
  },
  "papermill": {
   "default_parameters": {},
   "duration": 55.400587,
   "end_time": "2022-02-21T07:37:31.209553",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2022-02-21T07:36:35.808966",
   "version": "2.3.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}