Number_Recognition_DL.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Number Recognition DL.ipynb",
      "provenance": [],
      "authorship_tag": "ABX9TyOjYhcAYgJRWHFvXjFqZp4o",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/DataMinati/Super-Notebooks-1K/blob/main/Number_Recognition_DL.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "O3lPyJuTsTaL"
      },
      "source": [
        "### 1. Importing the packages\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HprWMyfcnU8J"
      },
      "source": [
        "import numpy as np # linear algebra\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import confusion_matrix\n",
        "from keras.utils.np_utils import to_categorical # convert to one-hot-encoding\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPool2D, BatchNormalization\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from keras.callbacks import LearningRateScheduler"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tUIt1-yEn8Er"
      },
      "source": [
        "train_file = \"https://raw.githubusercontent.com/MainakRepositor/Datasets-/master/DL-nums.csv\"\n",
        "test_file = \"https://raw.githubusercontent.com/MainakRepositor/Datasets-/master/test.csv\""
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cRoubfI0ofkU"
      },
      "source": [
        "raw_data = np.loadtxt(train_file, skiprows=1, dtype='int', delimiter=',')\n",
        "x_train, x_val, y_train, y_val = train_test_split(\n",
        "    raw_data[:,1:], raw_data[:,0], test_size=0.1)"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 406
        },
        "id": "3n6lwNjXoigo",
        "outputId": "6246ac3e-b8cd-4669-943d-84082f99ae1c"
      },
      "source": [
        "fig, ax = plt.subplots(2, 1, figsize=(12,6))\n",
        "ax[0].plot(x_train[0])\n",
        "ax[0].set_title('784x1 data')\n",
        "ax[1].imshow(x_train[0].reshape(28,28), cmap='gray')\n",
        "ax[1].set_title('28x28 data')"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, '28x28 data')"
            ]
          },
          "metadata": {},
          "execution_count": 6
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x432 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wWWXGBZWotFA"
      },
      "source": [
        "x_train = x_train.reshape(-1, 28, 28, 1)\n",
        "x_val = x_val.reshape(-1, 28, 28, 1)"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xnrsQAIxovvd"
      },
      "source": [
        "x_train = x_train.astype(\"float32\")/255.\n",
        "x_val = x_val.astype(\"float32\")/255."
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "aOuQ37NOoxSx",
        "outputId": "e85ef860-c5c9-4d98-b5eb-7da386220fed"
      },
      "source": [
        "y_train = to_categorical(y_train)\n",
        "y_val = to_categorical(y_val)\n",
        "#example:\n",
        "print(y_train[0])"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_7pMqBBfozKQ"
      },
      "source": [
        "model = Sequential()\n",
        "\n",
        "model.add(Conv2D(filters = 16, kernel_size = (3, 3), activation='relu',\n",
        "                 input_shape = (28, 28, 1)))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Conv2D(filters = 16, kernel_size = (3, 3), activation='relu'))\n",
        "model.add(BatchNormalization())\n",
        "#model.add(Conv2D(filters = 16, kernel_size = (3, 3), activation='relu'))\n",
        "#model.add(BatchNormalization())\n",
        "model.add(MaxPool2D(strides=(2,2)))\n",
        "model.add(Dropout(0.25))\n",
        "\n",
        "model.add(Conv2D(filters = 32, kernel_size = (3, 3), activation='relu'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Conv2D(filters = 32, kernel_size = (3, 3), activation='relu'))\n",
        "model.add(BatchNormalization())\n",
        "#model.add(Conv2D(filters = 32, kernel_size = (3, 3), activation='relu'))\n",
        "#model.add(BatchNormalization())\n",
        "model.add(MaxPool2D(strides=(2,2)))\n",
        "model.add(Dropout(0.25))\n",
        "\n",
        "model.add(Flatten())\n",
        "model.add(Dense(512, activation='relu'))\n",
        "model.add(Dropout(0.25))\n",
        "model.add(Dense(1024, activation='relu'))\n",
        "model.add(Dropout(0.5))\n",
        "model.add(Dense(10, activation='softmax'))\n"
      ],
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pIkT0YfFo3dq"
      },
      "source": [
        "datagen = ImageDataGenerator(zoom_range = 0.1,\n",
        "                            height_shift_range = 0.1,\n",
        "                            width_shift_range = 0.1,\n",
        "                            rotation_range = 10)"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "t4Ll5W1Ko-_X",
        "outputId": "cb5174c7-ed5e-4cd4-97b0-4e75c8251fdf"
      },
      "source": [
        "model.compile(loss='categorical_crossentropy', optimizer = Adam(lr=1e-4), metrics=[\"accuracy\"])"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/keras/optimizer_v2/optimizer_v2.py:356: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead.\n",
            "  \"The `lr` argument is deprecated, use `learning_rate` instead.\")\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "H6ezEwj1pBi1"
      },
      "source": [
        "annealer = LearningRateScheduler(lambda x: 1e-3 * 0.9 ** x)"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "smyRK3V4pDP6",
        "outputId": "631b4df9-fe42-4a1a-ca69-2cb4b4087a08"
      },
      "source": [
        "hist = model.fit_generator(datagen.flow(x_train, y_train, batch_size=16),\n",
        "                           steps_per_epoch=500,\n",
        "                           epochs=20, \n",
        "                           verbose=2,  \n",
        "                           validation_data=(x_val[:400,:], y_val[:400,:]), #For speed\n",
        "                           callbacks=[annealer])"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/keras/engine/training.py:1972: UserWarning: `Model.fit_generator` is deprecated and will be removed in a future version. Please use `Model.fit`, which supports generators.\n",
            "  warnings.warn('`Model.fit_generator` is deprecated and '\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/20\n",
            "500/500 - 27s - loss: 0.8228 - accuracy: 0.7454 - val_loss: 0.2152 - val_accuracy: 0.9300\n",
            "Epoch 2/20\n",
            "500/500 - 25s - loss: 0.3211 - accuracy: 0.9040 - val_loss: 0.0782 - val_accuracy: 0.9775\n",
            "Epoch 3/20\n",
            "500/500 - 25s - loss: 0.2318 - accuracy: 0.9321 - val_loss: 0.0274 - val_accuracy: 0.9900\n",
            "Epoch 4/20\n",
            "500/500 - 26s - loss: 0.1936 - accuracy: 0.9409 - val_loss: 0.0535 - val_accuracy: 0.9850\n",
            "Epoch 5/20\n",
            "500/500 - 26s - loss: 0.1975 - accuracy: 0.9439 - val_loss: 0.0323 - val_accuracy: 0.9900\n",
            "Epoch 6/20\n",
            "500/500 - 25s - loss: 0.1500 - accuracy: 0.9555 - val_loss: 0.0277 - val_accuracy: 0.9925\n",
            "Epoch 7/20\n",
            "500/500 - 25s - loss: 0.1497 - accuracy: 0.9572 - val_loss: 0.0199 - val_accuracy: 0.9950\n",
            "Epoch 8/20\n",
            "500/500 - 25s - loss: 0.1218 - accuracy: 0.9661 - val_loss: 0.0353 - val_accuracy: 0.9875\n",
            "Epoch 9/20\n",
            "500/500 - 25s - loss: 0.1149 - accuracy: 0.9656 - val_loss: 0.0268 - val_accuracy: 0.9900\n",
            "Epoch 10/20\n",
            "500/500 - 25s - loss: 0.1125 - accuracy: 0.9675 - val_loss: 0.0150 - val_accuracy: 0.9950\n",
            "Epoch 11/20\n",
            "500/500 - 25s - loss: 0.1011 - accuracy: 0.9693 - val_loss: 0.0176 - val_accuracy: 0.9975\n",
            "Epoch 12/20\n",
            "500/500 - 25s - loss: 0.0961 - accuracy: 0.9707 - val_loss: 0.0112 - val_accuracy: 0.9975\n",
            "Epoch 13/20\n",
            "500/500 - 25s - loss: 0.0918 - accuracy: 0.9745 - val_loss: 0.0123 - val_accuracy: 0.9925\n",
            "Epoch 14/20\n",
            "500/500 - 24s - loss: 0.0903 - accuracy: 0.9714 - val_loss: 0.0111 - val_accuracy: 0.9975\n",
            "Epoch 15/20\n",
            "500/500 - 25s - loss: 0.0789 - accuracy: 0.9761 - val_loss: 0.0092 - val_accuracy: 0.9975\n",
            "Epoch 16/20\n",
            "500/500 - 25s - loss: 0.0801 - accuracy: 0.9768 - val_loss: 0.0105 - val_accuracy: 0.9950\n",
            "Epoch 17/20\n",
            "500/500 - 25s - loss: 0.0709 - accuracy: 0.9811 - val_loss: 0.0128 - val_accuracy: 0.9975\n",
            "Epoch 18/20\n",
            "500/500 - 25s - loss: 0.0751 - accuracy: 0.9800 - val_loss: 0.0067 - val_accuracy: 0.9975\n",
            "Epoch 19/20\n",
            "500/500 - 25s - loss: 0.0686 - accuracy: 0.9806 - val_loss: 0.0081 - val_accuracy: 0.9975\n",
            "Epoch 20/20\n",
            "500/500 - 24s - loss: 0.0708 - accuracy: 0.9800 - val_loss: 0.0158 - val_accuracy: 0.9975\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xXTa59jBpF2o",
        "outputId": "0e68c0e4-daa8-485a-b33b-8e1e971e162a"
      },
      "source": [
        "final_loss, final_acc = model.evaluate(x_val, y_val, verbose=0)\n",
        "print(\"Final loss: {0:.4f}, final accuracy: {1:.4f}\".format(final_loss, final_acc))"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Final loss: 0.0203, final accuracy: 0.9938\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 513
        },
        "id": "5ml8J2hsrzKE",
        "outputId": "3a4d78d1-4911-40a6-9d8b-894b5b425d77"
      },
      "source": [
        "plt.plot(hist.history['loss'], color='b')\n",
        "plt.plot(hist.history['val_loss'], color='r')\n",
        "plt.show()\n",
        "plt.plot(hist.history['accuracy'], color='b')\n",
        "plt.plot(hist.history['val_accuracy'], color='r')\n",
        "plt.show()"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "display_data",
          "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": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAD4CAYAAADiry33AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3de3RU5b3/8fdXrhUEUeKFS7goolhFbURbrLiO9Xpaaa2nBX+1XrqWx7Z4arX+jrY92tJlW3tVWtBDj1TtUdFqL/QnrbWVeqsgQUUFCyIJkIASRJA7hHx/fzx7OpPJJBnIZGay9+e11qzZsy8z30wmn9l59rOfbe6OiIjE1wGlLkBERDqXgl5EJOYU9CIiMaegFxGJOQW9iEjMdS91AdkGDhzow4cPL3UZIiJdyqJFiza4e0WuZWUX9MOHD6e6urrUZYiIdClmtqq1ZWq6ERGJOQW9iEjMKehFRGKu3aA3s1lmtt7MXm9luZnZNDNbYWavmtkpGcsuN7M3o9vlhSxcRETyk88e/b3A+W0svwAYFd2uBu4CMLNDgFuB04BxwK1mNqAjxYqIyL5rN+jd/RlgYxurTATu92A+cLCZHQmcBzzp7hvd/T3gSdr+whARkU5QiDb6wcCajMd10bzW5rdgZlebWbWZVTc0NBSgJBERSSmLfvTuPhOYCVBVVaVxk6VraGqCt9+G2lpYtSrctm/v2HOOGAETJoR7s4KU2a6mJnjtNXjmGejojtbBB8OwYTB8eLg/9NDO+zncYf368L6nfgdbtnTOaxXLkCFw9dUFf9pCBH09MDTj8ZBoXj1wVtb8vxXg9USKo7ER6urSIZ4Z6LW1sGYN7N7dfJuOhFrmtSEGDw6Bf+aZ4X706MIFZmMjvPxyCPann4Znn4VNm8KyQtWfcuCB6dDP/AJITR9+OBzQSsPC3r2wbl3L9z01vWoV7NzZfJtifTl2ltNOK9ugnwNMMbPZhAOvm919nZk9AXw34wDsucDNBXg9KRc7dsDq1ek/wLo6OOEE+MQnoHfvUleXv5074cUXQ+AtW5YOk/r6EDaZjjwyBNSpp8IllzQPrmHDoE+f/a/DHd54I4Tv00/DU0/Bgw+GZYcdlg79CRPg+ONbD8hsu3dDdXX6eZ9/HrZuDcuOOSb8HKkvlcrKjtW/aVPuL8VVq2DBAtiYdbivZ8/wmqn3ce/e5l+kjY3N16+oCOuecAJ8/OPp7VL3/frtf/0xZu1dYcrMHiLsmQ8E3iH0pOkB4O53m5kBPyccaN0OXOnu1dG2VwFfj57qNnf/ZXsFVVVVuYZAKBPvv9987yn7j3f9+tzb9e8Pn/0sXHYZjB9ffntZ27bB/Pnp4FuwAHbtCnVWVra+9zl0KPTqVbw63WHFilBjau979eqw7JBD4KMfTYf/SSdBt25h2Y4d4WdKbfPCC2EehC+IVKifeWb44iqmrVtb/yJYtSp8eWWHd2q6sjL8hyA5mdkid6/KuazcLiWooC8S97B3lf2vcObj995rvk2vXs33vnL9G/7ss3D//fDYYyFQR44MgX/ZZXDUUUX/MYHwhfX88+ngW7gw7Cl26wannJIOyzPOgAFl3gO4tjb9czz9NLz1Vpjfr1/4Ut2yJfx3snt3+OI66aTmP19FzjGvJAYU9EnU1BT2uHPtOaXut21rvk2fPs0DPDvIDzss/+aCrVvhN78Jof/UU+GLZfx4+Pzn4TOfCQftOsvGjfDcc+k94ZdeCu9Hjx6hySW1Rzt+PBx0UOfVUQz19eFnfOaZ8CXbp0/65zvjjM59n6WsKOiLYf16uOOOsEc4YQKcfHIIlmJYuzaE2nPPwZtvhiBfvTo0R2QaMKD1Zolhw0JzQGc0s6xZAw88EEL/jTfCfwYXXRRC/7zz9u99amoKB+qyv8Dmzw89SNzD65x+erpd+/TT9a+/xJaCvjPt3QszZ8LXvx6aCJqawvw+fcIeY2rv6tRTC9e+29q/7wcdBMcdlzvMy+FAlTssWhQC/6GHYMOG0JRw6aUh9E8+Of1Fs2dP6z1eVq0KX2R79jR//oEDQ1NFKtjHjStum7pICSnoO8tLL8E114Q233/5F5gxIxyIfPbZdAi/Hg0R1Ls3fPjD6fbS00+HD3yg/ddo64DcgAHpg2oTJsDYsdC9LE6NaN/u3fCnP4XQ/8MfwuMxY8J/FakeL6kvzZRBg1r/j6SysmM9XkS6OAV9oW3eDN/8Zgj2igr4yU9g8uTczR7vvts8+F95JYR3jx5hjzO19/mRj0Dfvs272KWCfd268FwVFen1zzwTPvjB/NvMy9nGjfDII/DwwyHcczUrFbvHi0gXo6AvFPfQ5HD99eEMwi99Cb7znX074LVpU+gBkgry6urQ/NOtG5x4YmjP3rAhrDtoUDrYC33SjIjESltB30X+zy8Dy5bBl78Mf/0rVFXB44/Dhz60789z8MHwr/8abhB6p/z97yH0X3ghhH2qKWbkSAW7iHSYgr49O3bAd78LP/hBaFOfMSOcopw6OaWj+vaFc88NNxGRTqCgb8vcuTBlCtTUhBN+fvjDcFKQiJSFpqbwz/aCBeE8sc2bQ+ey/v3Dfa5b5rLevfP7p9k9nGO3a1cYMaO1G4T9wd69W78Vah9xXyjoc1mzBr7yFfjtb0N3xXnz4KyzSl2VSOK980461BcsCB3eNm8Oy/r1Cz1st2wJPZ2zTyPJpXv35l8C3bu3HuLZncD2V/fu6dDP/lIYOxZ+8YvCvE6z1yz8U3Zhe/aEk56+/e3wW/3ud+GGG8LASyJSVNu3h0E2FyxI31atCstSfRcmTQoDPp52Ghx7bPNOaLt2hcBv77Z5c/q+qantvfFevdpeBm3v8e/cGVqDW1uWT4/r/aGgT9m5M5zg9NJLYfTFadNC1z4RKZimptzNH6nwW748HeqvvpoePLSyMoT5tdeG+1NOaf8k5169Qo9kDe+joE+bPTuE/C9/CVdcUepqRLqEvXtDID//fOg8tnZt23uu2cP359KvXziR/D//M4T6uHFwxBGd/7PEmYIewpGWO+8MJyBdfnmpqxHZJ1u2hLB99dXQ/jt6dGjGqKgofO/crVvDcELPPx9uL7yQHtp+0KAwQGn//iGYW2viaOtg5dChLZtgpOMU9BAGA3vllTBmjfqtS5lyD6NfLF4cPq6LF4dbaqijbAcfHEI/dTv22HB/9NH5n2RcXx/+PFLBvnhx2Is3C9f+SF1yYPz4cAKz/nzKk86MBfi3fwsnQtXVaXRDKQu7dsGSJS1DPfOKf0cfHcZwGzs2fdu7N3Q3XLYM/vGP9HR9ffq5U9f2yP4CGD06nPD9/PPpcE8d/DzwwNCMkgr1D3847LlL+dCZsW1ZvTp0o7zhBoW8dDr30NTy7rthiJ+NG9PT774bQnnx4hDSqavo9emT7mEydmwI9w9+MJxrl8vw4WH050xbt4YDnZnhv2xZGIkj1/XMjzwyBPp114Vh7ceOLd6o21J4Cvq77gp/fV/6UqkrkS6mqSlchGv9+vRtw4bcAZ6at3Fjy8ugZhoyJAT5xInpUD/qqI63WfftG3qqnHJKy5+hri6E/vLlYS99/PjwZaFmmPhIdtDv2BHa5T/1qdDAKInmHvZ8Gxqah3euW0NDuGVfOzylT58w4vKhh4b7E04I95nzUvep6QEDin/KxgEHhK6LlZVwzjnFfW0pnmQH/YMPhl2s//iPUlcinWjbNnj77XBWZXv3qWtoZzvooHAlxcMOC2PNnX56+nFFRfq+oiIEt0ZUlnKS3KB3DydFjR0LH/1oqatJtB07wpUGZ84MTR89eza/9ejRcl6u+T16hDMcs8M71f0vk1k4Xf7ww0NXwPHjw/Thh6cDPDPIO+uMRZFiSG7QP/106Hh8zz1qjCyRurowGOjMmaEd+8QTw4G/3bvDbc+e9PT27aHHSa5lmbd+/dLhPW5cejr7vqKi61yMS6SjkvtRnzYtNIxOnlzqShLFPZxkc+ed8Nhj4fHEiWEMuTPP1HeuSGdIZtDX1sLvfx/Osdb/5EWxa1e4WuC0aeGiWv37h657U6ZoSCGRzpbMoJ8xI+w6fvGLpa4k9t55B+6+O/RifeedcHLOjBnhjMrW+oGLSGElL+i3bQsDPl98cRhYQzrFokVh73327NB2fuGFoXPTOedoHBORYkte0D/wQDiqpy6VBdfYGE4yvvPOcPp8nz7hqovXXgvHHFPq6kSSK1lBn+pSefLJoT+dtCvX2Z+5Th5avz4MUbtlC4wYAT/5CVx1lcZDESkHyQr6efPCSFH33qvuHVk2bYKf/xzeeKNliOc6+9MsdFpK9TUfOzY0y3zsY/Dxj5fmupgikluygv7OO0MH6s9+ttSVlI3GxtCP/dZbQ1/2kSNDcI8YEfqhZ588lLodeqj6oYt0FXn9qZrZ+cCdQDfgf9z9+1nLhwGzgApgI/A5d6+Llu0FXotWXe3uFxWo9n2zciX84Q/wjW+EKxwIf/pTGLRz6VKYMCE0t2QPeiUiXV+7/R/MrBswHbgAGANMNrMxWav9CLjf3U8EpgLfy1i2w91Pim6lCXmA6dNDe8I115SshHKxZAlccEG47doVDqDOm6eQF4mrfDq6jQNWuPtKd98NzAYmZq0zBngqmp6XY3lpbd0ahjq45BIYPLjU1ZRMQ0MYjXns2HB26o9/HPbmP/lJHbIQibN8gn4wsCbjcV00L9Ni4OJo+lPAQWZ2aPS4t5lVm9l8M/tkrhcws6ujdaobGhr2ofw8/epXsHlzYrtU7toFP/oRjBoV2uO/+EVYsQKuv774w+KKSPEV6tSVrwETzOxlYAJQD6T6agyLLm91KXCHmR2VvbG7z3T3KnevqqioKFBJ/3zy0KWyqiqMLZsg7mE8mTFj4MYbQ4/S116Dn/0sjNwoIsmQz8HYeiDzFNIh0bx/cve1RHv0ZtYX+LS7b4qW1Uf3K83sb8DJQCuXM+4Ef/lLuH7a/fcnqn1i0aKwx/7MM3D88fDEE3DuuaWuSkRKIZ89+oXAKDMbYWY9gUnAnMwVzGygmaWe62ZCDxzMbICZ9UqtA4wHlhaq+LxMmxb6A37mM0V92VKpr4crroBTTw194u+6K1xcWiEvklzt7tG7e6OZTQGeIHSvnOXuS8xsKlDt7nOAs4DvmZkDzwBfjjY/DvhvM2sifKl8392LF/QrVsDjj8N//VdsL/mzbVu41ufy5WFUyBkzQt/4G2+Er39dZ6aKCJi7l7qGZqqqqry6urowT3bddSH5Vq0Kl7Xvohobw8jKy5enL+Kcuq/PaEQzg09/Gm6/PZz4JCLJYWaLouOhLcT33MYtW2DWrNBk00VCfuPG0Mc9O9DfeitcUSnl4INh9Gg4++wwWNjo0eH+6KPhwANLV7+IlKf4Bv1994WwL/Mule4wf344lPDoo2HvHUK3x6OPhuOOC1dgSoX56NFh+IEEHVcWkQ6KZ9A3NYU+hKedFgZsKUO7d6evuLRwYWhLv/bacND0mGNg2DANDCYihRHPoP/zn0O7xwMPlLqSFlJXXLr7bnj77bCHPn06fP7zuuKSiHSOeAb9tGlwxBFhyIMysWhRGDzz4YfD3vwFF4QLYuuKSyLS2eIX9MuXwx//CN/+dsnP79+zJ33Fpb//Peyx64pLIlJs8Qv6n/0MevSAf//3kpWwYUO4LO2MGVBXF7o6/vSncOWV6tcuIsUXr6DfvDlcPWrSJDj88KK//Ouvwx13hEMDO3eGqy3NmBEujK0DqyJSKvEK+nvvDUMSl6BL5YsvwhlnhKsuXX55aJ45/viilyEi0kJ8gj7VpfIjHwkjVRbR++/D5MnhvKyFC8PQOiIi5SI+QV9bCzt2FH1v3j2M775qFTz9tEJeRMpPfIJ+5MgQ9kU+ZfS+++DBB+E73wnjvYuIlJv4BD2E3jZFtGwZTJkCZ50FN99c1JcWEcmbTtXZT7t2hXb53r3hf/9XvWpEpHzFa4++iG66CV5+GebMSfT1xkWkC9Ae/X54/PHQX/7aa+ETnyh1NSIibVPQ76O1a8Ol+saOhR/8oNTViIi0T0G/D/buhc99DrZvh9mzQ/u8iEi5Uxv9Prj9dpg3D+65B449ttTViIjkR3v0eXrhBbjlljCMzpVXlroaEZH8KejzsGlT6EpZWRkuGKLL+IlIV6Kmm3a4hzHk6+vhuec0zLCIdD0K+nbccw/8+tfw/e+HS9CKiHQ1arppw9KlYYy0j30Mbryx1NWIiOwfBX0rdu4MB1779oX779d1XUWk61LTTSu+9jV47TWYOzeMMy8i0lVpPzWH3/0Opk+H66+HCy4odTUiIh2joM+yZg1cdRV86EPwve+VuhoRkY5T0GdIDXGwZw889BD07FnqikREOk5t9Bluuw2eeSYcfB01qtTViIgUhvboI5s3w9SpcOmlcNllpa5GRKRw8gp6MzvfzJaZ2QozuynH8mFm9lcze9XM/mZmQzKWXW5mb0a3ywtZfCG99VZournkklJXIiJSWO0GvZl1A6YDFwBjgMlmNiZrtR8B97v7icBU4HvRtocAtwKnAeOAW81sQOHKL5yamnA/YkRp6xARKbR89ujHASvcfaW77wZmAxOz1hkDPBVNz8tYfh7wpLtvdPf3gCeB8zteduEp6EUkrvIJ+sHAmozHddG8TIuBi6PpTwEHmdmheW6LmV1tZtVmVt3Q0JBv7QVVUwMDBmjQMhGJn0IdjP0aMMHMXgYmAPXA3nw3dveZ7l7l7lUVFRUFKmnf1NRob15E4imfoK8HhmY8HhLN+yd3X+vuF7v7ycA3onmb8tm2XCjoRSSu8gn6hcAoMxthZj2BScCczBXMbKCZpZ7rZmBWNP0EcK6ZDYgOwp4bzSsr7lBbC8OHl7oSEZHCazfo3b0RmEII6DeAR9x9iZlNNbOLotXOApaZ2XLgcOC2aNuNwHcIXxYLganRvLLy9tthtErt0YtIHOV1Zqy7zwXmZs27JWP6UeDRVradRXoPvyypx42IxJnOjEVBLyLxpqAnHfRqoxeROFLQE4L+iCPgAx8odSUiIoWnoEddK0Uk3hT0KOhFJN4SH/SNjeGqUgp6EYmrxAf9mjVheGIFvYjEVeKDXl0rRSTuFPQKehGJOQV9DXTrBkOHtr+uiEhXpKCvCSHfXZdJF5GYUtCra6WIxJyCXkEvIjGX6KDfsSMMUaygF5E4S3TQ19aGewW9iMRZooNeXStFJAkU9CjoRSTeEh/0vXuHIYpFROIq8UE/fDiYlboSEZHOk/igV7ONiMRdooO+tlZBLyLxl9ig37wZ3ntPQS8i8ZfYoFePGxFJCgW9gl5EYk5Br6AXkZhLdND37w8DBpS6EhGRzpXooNfevIgkQaKDfvjwUlchItL5Ehn07upDLyLJkcigX78etm9X0ItIMuQV9GZ2vpktM7MVZnZTjuWVZjbPzF42s1fN7MJo/nAz22Fmr0S3uwv9A+wP9bgRkSRp95LYZtYNmA6cA9QBC81sjrsvzVjtm8Aj7n6XmY0B5gLDo2VvuftJhS27YxT0IpIk+ezRjwNWuPtKd98NzAYmZq3jQL9ouj+wtnAlFl4q6HUwVkSSIJ+gHwysyXhcF83L9C3gc2ZWR9ibvzZj2YioSedpM/torhcws6vNrNrMqhsaGvKvfj/V1MBhh0GfPp3+UiIiJVeog7GTgXvdfQhwIfArMzsAWAdUuvvJwPXAg2bWL3tjd5/p7lXuXlVRUVGgklqnPvQikiT5BH09MDTj8ZBoXqYvAI8AuPsLQG9goLvvcvd3o/mLgLeAYzpadEcp6EUkSfIJ+oXAKDMbYWY9gUnAnKx1VgNnA5jZcYSgbzCziuhgLmY2EhgFrCxU8ftj715YvVpBLyLJ0W6vG3dvNLMpwBNAN2CWuy8xs6lAtbvPAW4AfmFmXyUcmL3C3d3MzgSmmtkeoAm4xt03dtpPk4e6OmhsVNCLSHK0G/QA7j6XcJA1c94tGdNLgfE5tnsMeKyDNRaUulaKSNIk7sxYBb2IJE0ig/6AA6CystSViIgURyKDfsgQ6NGj1JWIiBRHIoNezTYikiQKehGRmEtU0O/cCWvXKuhFJFkSFfSrVoV7Bb2IJEmigl5dK0UkiRT0IiIxl7ig79ULjjyy1JWIiBRP4oJ+2LBwwpSISFIkKvLUtVJEkkhBLyISc4kJ+vffh40bFfQikjyJCXr1uBGRpFLQi4jEnIJeRCTmEhX0ffvCIYeUuhIRkeJKTNDX1oa9ebNSVyIiUlyJCXp1rRSRpEpE0Lsr6EUkuRIR9Bs2wLZtCnoRSaZEBL163IhIkinoRURiTkEvIhJziQn6gQNDP3oRkaRJTNBrb15EkkpBLyISc7EP+r17YdUqBb2IJFfsg37tWtizR0EvIsmVV9Cb2flmtszMVpjZTTmWV5rZPDN72cxeNbMLM5bdHG23zMzOK2Tx+VCPGxFJuu7trWBm3YDpwDlAHbDQzOa4+9KM1b4JPOLud5nZGGAuMDyangQcDwwC/mJmx7j73kL/IK1R0ItI0uWzRz8OWOHuK919NzAbmJi1jgP9oun+wNpoeiIw2913uXsNsCJ6vqKpqQkjVlZWFvNVRUTKRz5BPxhYk/G4LpqX6VvA58ysjrA3f+0+bIuZXW1m1WZW3dDQkGfp+ampgcGDoVevgj6tiEiXUaiDsZOBe919CHAh8Cszy/u53X2mu1e5e1VFRUWBSgrUtVJEki6fMK4HhmY8HhLNy/QF4BEAd38B6A0MzHPbTqWgF5GkyyfoFwKjzGyEmfUkHFydk7XOauBsADM7jhD0DdF6k8ysl5mNAEYBLxaq+Pbs2gX19Qp6EUm2dnvduHujmU0BngC6AbPcfYmZTQWq3X0OcAPwCzP7KuHA7BXu7sASM3sEWAo0Al8uZo+b1avDRUcU9CKSZO0GPYC7zyUcZM2cd0vG9FJgfCvb3gbc1oEa95u6VoqIxPzMWAW9iEgCgr5HDxg0qNSViIiUTuyDftgw6Nat1JWIiJRO7INezTYiknQKehGRmItt0G/dChs2KOhFRGIb9OpxIyISKOhFRGIu9kE/fHhJyxARKblYB/2BB0KBB8MUEelyYh30I0aEi46IiCRZ7INeRCTpYhn07gp6EZGUWAb9u++GfvQKehGRmAa9ulaKiKQp6EVEYi6WQV9bG+4V9CIiMQ36mho45BDo16/UlYiIlF5sg1578yIigYJeRCTmYhf0TU2hjV5BLyISxC7o162D3bsV9CIiKbELenWtFBFpTkEvIhJzsQ36YcNKW4eISLmIZdAPGgS9e5e6EhGR8hDLoFezjYhImoJeRCTmYhX0e/ZAXZ2CXkQkU6yCfvXqcMKUgl5EJC2voDez881smZmtMLObciz/qZm9Et2Wm9mmjGV7M5bNKWTx2dS1UkSkpe7trWBm3YDpwDlAHbDQzOa4+9LUOu7+1Yz1rwVOzniKHe5+UuFKbp2CXkSkpXz26McBK9x9pbvvBmYDE9tYfzLwUCGK21c1NdC9OwwZUopXFxEpT/kE/WBgTcbjumheC2Y2DBgBPJUxu7eZVZvZfDP7ZCvbXR2tU93Q0JBn6S3V1EBlJXTrtt9PISISO4U+GDsJeNTd92bMG+buVcClwB1mdlT2Ru4+092r3L2qoqJiv19cXStFRFrKJ+jrgaEZj4dE83KZRFazjbvXR/crgb/RvP2+oBT0IiIt5RP0C4FRZjbCzHoSwrxF7xkzOxYYALyQMW+AmfWKpgcC44Gl2dsWwrZtsH69gl5EJFu7vW7cvdHMpgBPAN2AWe6+xMymAtXungr9ScBsd/eMzY8D/tvMmghfKt/P7K1TSNu3w+TJcOqpnfHsIiJdlzXP5dKrqqry6urqUpchItKlmNmi6HhoC7E6M1ZERFpS0IuIxJyCXkQk5hT0IiIxp6AXEYk5Bb2ISMwp6EVEYk5BLyISc2V3wpSZNQCrOvAUA4ENBSqnM6i+jlF9HaP6Oqac6xvm7jlHhSy7oO8oM6tu7eywcqD6Okb1dYzq65hyr681aroREYk5Bb2ISMzFMehnlrqAdqi+jlF9HaP6Oqbc68spdm30IiLSXBz36EVEJIOCXkQk5rpk0JvZ+Wa2zMxWmNlNOZb3MrOHo+ULzGx4EWsbambzzGypmS0xs6/kWOcsM9tsZq9Et1uKVV9GDbVm9lr0+i2u9GLBtOg9fNXMTilibaMz3ptXzOx9M7sua52ivodmNsvM1pvZ6xnzDjGzJ83szeh+QCvbXh6t86aZXV7E+n5oZv+Ifn+/NbODW9m2zc9CJ9b3LTOrz/gdXtjKtm3+vXdifQ9n1FZrZq+0sm2nv38d5u5d6ka4nOFbwEigJ7AYGJO1zpeAu6PpScDDRazvSOCUaPogYHmO+s4C/l+J38daYGAbyy8E/ggYcDqwoIS/77cJJ4OU7D0EzgROAV7PmPcD4KZo+ibg9hzbHQKsjO4HRNMDilTfuUD3aPr2XPXl81noxPq+BXwtj99/m3/vnVVf1vIfA7eU6v3r6K0r7tGPA1a4+0p33w3MBiZmrTMRuC+afhQ428ysGMW5+zp3fyma3gK8AQwuxmsX2ETgfg/mAweb2ZElqONs4C1378jZ0h3m7s8AG7NmZ37O7gM+mWPT84An3X2ju78HPAmcX4z63P3P7t4YPZwPDCn06+arlfcvH/n8vXdYW/VF2fEZ4KFCv26xdMWgHwysyXhcR8sg/ec60Qd9M3BoUarLEDUZnQwsyLH4w2a22Mz+aGbHF7WwwIE/m9kiM7s6x/J83udimETrf2Clfg8Pd/d10fTbwOE51imX9/Eqwn9oubT3WehMU6KmpVmtNH2Vw/v3UeAdd3+zleWlfP/y0hWDvksws77AY8B17v5+1uKXCE0RY4GfAb8rdn3AGe5+CnAB8GUzO7MENbTJzHoCFwG/zrG4HN7Df/LwP3xZ9lU2s28AjcADraxSqs/CXcBRwEnAOkLzSDmaTNt782X/t9QVg74eGJrxeEg0L+c6ZtYd6A+8W5Tqwmv2IIT8A+7+m+zl7v6+u2+NpucCPcxsYLHqi163PrpfD/yW8C9ypnze5852AfCSu7+TvaAc3kPgnVRzVnS/Psc6JX0fzewK4OPA/4m+jFrI47PQKdz9HXff61b9zh4AAAGZSURBVO5NwC9aed1Sv3/dgYuBh1tbp1Tv377oikG/EBhlZiOiPb5JwJysdeYAqd4NlwBPtfYhL7SoPe8e4A13/0kr6xyROmZgZuMIv4difhH1MbODUtOEg3avZ602B/h81PvmdGBzRjNFsbS6J1Xq9zCS+Tm7HPh9jnWeAM41swFR08S50bxOZ2bnA/8XuMjdt7eyTj6fhc6qL/OYz6daed18/t4708eAf7h7Xa6FpXz/9kmpjwbvz43QI2Q54Wj8N6J5UwkfaIDehH/3VwAvAiOLWNsZhH/hXwVeiW4XAtcA10TrTAGWEHoQzAc+UuT3b2T02oujOlLvYWaNBkyP3uPXgKoi19iHENz9M+aV7D0kfOGsA/YQ2om/QDju81fgTeAvwCHRulXA/2Rse1X0WVwBXFnE+lYQ2rdTn8NUT7RBwNy2PgtFqu9X0WfrVUJ4H5ldX/S4xd97MeqL5t+b+sxlrFv096+jNw2BICISc12x6UZERPaBgl5EJOYU9CIiMaegFxGJOQW9iEjMKehFRGJOQS8iEnP/H6hbVGZ0KNKTAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "s43bOiGMr2iN",
        "outputId": "98805459-0e12-4ae8-ce40-346496342d32"
      },
      "source": [
        "y_hat = model.predict(x_val)\n",
        "y_pred = np.argmax(y_hat, axis=1)\n",
        "y_true = np.argmax(y_val, axis=1)\n",
        "cm = confusion_matrix(y_true, y_pred)\n",
        "print(cm)"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[404   0   0   0   0   0   1   0   3   0]\n",
            " [  0 463   0   0   0   1   0   1   0   0]\n",
            " [  0   0 379   1   0   0   0   1   1   0]\n",
            " [  0   0   0 431   0   0   0   0   0   0]\n",
            " [  0   0   0   0 434   0   0   1   0   0]\n",
            " [  0   0   0   0   0 393   0   0   1   0]\n",
            " [  0   0   0   0   0   0 405   0   0   0]\n",
            " [  0   3   2   0   2   0   0 451   0   1]\n",
            " [  0   0   0   0   0   0   0   0 406   0]\n",
            " [  0   0   0   0   3   0   0   1   3 408]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9Nxj51NOr7Cl"
      },
      "source": [
        "mnist_testset = np.loadtxt(test_file, skiprows=1, dtype='int', delimiter=',')\n",
        "x_test = mnist_testset.astype(\"float32\")\n",
        "x_test = x_test.reshape(-1, 28, 28, 1)/255.\n",
        "y_hat = model.predict(x_test, batch_size=64)"
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Oblfur4CsPyO"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}