Model_Training.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "Import libraries"
      ],
      "metadata": {
        "id": "UJ6jHZVvXWPC"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import tensorflow as tf"
      ],
      "metadata": {
        "id": "f-A7S_mLXVEg"
      },
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Connect to Drive if needed"
      ],
      "metadata": {
        "id": "BFZVfQbSYKRo"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qaJTrxdSYNa4",
        "outputId": "25d348b7-9362-419d-c0cc-86f83872f4eb"
      },
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Load **X_train, y_train, X_val, y_val, X_test** and **y_test** from **source_path**"
      ],
      "metadata": {
        "id": "6jdfRdxDXLgB"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "source_path = \"/content/drive/MyDrive\""
      ],
      "metadata": {
        "id": "gPJN71d_Xqio"
      },
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "id": "nUUApaTbWG_E"
      },
      "outputs": [],
      "source": [
        "X_train = np.load(f\"{source_path}/X_train.npy\")\n",
        "X_test = np.load(f\"{source_path}/X_test.npy\")\n",
        "X_val = np.load(f\"{source_path}/X_val.npy\")\n",
        "y_train = np.load(f\"{source_path}/y_train.npy\")\n",
        "y_test = np.load(f\"{source_path}/y_test.npy\")\n",
        "y_val = np.load(f\"{source_path}/y_val.npy\")"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(X_test[4].shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "41EilsAGYf96",
        "outputId": "563c87a4-841a-4b94-e932-5eca2c8f6d85"
      },
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(64, 64, 1)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Create a *Sequential* **model**"
      ],
      "metadata": {
        "id": "Xg1uAlGggbSh"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model = tf.keras.Sequential()"
      ],
      "metadata": {
        "id": "YhJpD3o4go8p"
      },
      "execution_count": 56,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Add **2D** layers"
      ],
      "metadata": {
        "id": "Wf57J5KEgquA"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.add(tf.keras.layers.Conv2D(64,\n",
        "                                 kernel_size=(4,4),\n",
        "                                 strides=(1,1),\n",
        "                                 padding=\"same\",\n",
        "                                 activation=\"relu\"))\n",
        "model.add(tf.keras.layers.MaxPooling2D(2,2))\n",
        "model.add(tf.keras.layers.Conv2D(128,\n",
        "                                 kernel_size=(2,2),\n",
        "                                 strides=(1,1),\n",
        "                                 padding=\"same\",\n",
        "                                 activation=\"relu\"))\n",
        "model.add(tf.keras.layers.MaxPooling2D(2,2))\n",
        "model.add(tf.keras.layers.Conv2D(128,\n",
        "                                 kernel_size=(2,2),\n",
        "                                 strides=(1,1),\n",
        "                                 padding=\"same\",\n",
        "                                 activation=\"relu\"))\n",
        "model.add(tf.keras.layers.MaxPooling2D(2,2))"
      ],
      "metadata": {
        "id": "33dyO5-cgtrJ"
      },
      "execution_count": 57,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Add **Flatten** layer"
      ],
      "metadata": {
        "id": "7tln3Zekgt1A"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.add(tf.keras.layers.Flatten())"
      ],
      "metadata": {
        "id": "k-tc4hS3gt75"
      },
      "execution_count": 58,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Add **Dense** and **Dropout** layers"
      ],
      "metadata": {
        "id": "LdyYjMh8hMlR"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.add(tf.keras.layers.Dense(128, activation=\"relu\", kernel_regularizer=\"L2\"))\n",
        "model.add(tf.keras.layers.Dropout(0.6))\n",
        "model.add(tf.keras.layers.Dense(64, activation=\"relu\", kernel_regularizer=\"L2\"))\n",
        "model.add(tf.keras.layers.Dropout(0.5))\n",
        "model.add(tf.keras.layers.Dense(10, activation=\"softmax\"))"
      ],
      "metadata": {
        "id": "SkcX0rIchS9g"
      },
      "execution_count": 59,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Compile the **model**"
      ],
      "metadata": {
        "id": "PhSMB-nkhWQA"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "opt = tf.keras.optimizers.Adam(learning_rate=0.001)\n",
        "\n",
        "model.compile(optimizer=opt,\n",
        "              loss=\"sparse_categorical_crossentropy\",\n",
        "              metrics=\"accuracy\")"
      ],
      "metadata": {
        "id": "KgCVo-uxhYjR"
      },
      "execution_count": 60,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Train the **model**"
      ],
      "metadata": {
        "id": "puvgLwafhZ_p"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "results = model.fit(X_train, y_train,\n",
        "          batch_size=128,\n",
        "          epochs=60,\n",
        "          validation_data=(X_val, y_val))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8ZWcoIrghcXx",
        "outputId": "9799b7d3-f38f-4728-b776-c838da3095ff"
      },
      "execution_count": 61,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/60\n",
            "41/41 [==============================] - 3s 55ms/step - loss: 5.5933 - accuracy: 0.1187 - val_loss: 4.2885 - val_accuracy: 0.1214\n",
            "Epoch 2/60\n",
            "41/41 [==============================] - 2s 47ms/step - loss: 4.0038 - accuracy: 0.1195 - val_loss: 3.7413 - val_accuracy: 0.1694\n",
            "Epoch 3/60\n",
            "41/41 [==============================] - 2s 47ms/step - loss: 3.5468 - accuracy: 0.1466 - val_loss: 3.2701 - val_accuracy: 0.2032\n",
            "Epoch 4/60\n",
            "41/41 [==============================] - 2s 47ms/step - loss: 3.2197 - accuracy: 0.1783 - val_loss: 2.9807 - val_accuracy: 0.2307\n",
            "Epoch 5/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 3.0027 - accuracy: 0.1833 - val_loss: 2.7925 - val_accuracy: 0.3205\n",
            "Epoch 6/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 2.8132 - accuracy: 0.2110 - val_loss: 2.6506 - val_accuracy: 0.2753\n",
            "Epoch 7/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 2.6903 - accuracy: 0.2134 - val_loss: 2.4870 - val_accuracy: 0.3417\n",
            "Epoch 8/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 2.5550 - accuracy: 0.2279 - val_loss: 2.4230 - val_accuracy: 0.3755\n",
            "Epoch 9/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 2.4386 - accuracy: 0.2572 - val_loss: 2.2197 - val_accuracy: 0.4127\n",
            "Epoch 10/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 2.3329 - accuracy: 0.2948 - val_loss: 2.0941 - val_accuracy: 0.4574\n",
            "Epoch 11/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 2.1736 - accuracy: 0.3452 - val_loss: 1.9250 - val_accuracy: 0.4991\n",
            "Epoch 12/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 2.0650 - accuracy: 0.3830 - val_loss: 1.7566 - val_accuracy: 0.5633\n",
            "Epoch 13/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 1.9124 - accuracy: 0.4213 - val_loss: 1.5536 - val_accuracy: 0.6302\n",
            "Epoch 14/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 1.7807 - accuracy: 0.4679 - val_loss: 1.4294 - val_accuracy: 0.6709\n",
            "Epoch 15/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 1.6350 - accuracy: 0.5248 - val_loss: 1.2905 - val_accuracy: 0.7092\n",
            "Epoch 16/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 1.5480 - accuracy: 0.5729 - val_loss: 1.2072 - val_accuracy: 0.7367\n",
            "Epoch 17/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 1.4627 - accuracy: 0.6103 - val_loss: 1.1967 - val_accuracy: 0.7247\n",
            "Epoch 18/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 1.4115 - accuracy: 0.6235 - val_loss: 1.0936 - val_accuracy: 0.7710\n",
            "Epoch 19/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 1.2935 - accuracy: 0.6537 - val_loss: 1.0408 - val_accuracy: 0.7876\n",
            "Epoch 20/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 1.2870 - accuracy: 0.6627 - val_loss: 1.0437 - val_accuracy: 0.7876\n",
            "Epoch 21/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 1.2370 - accuracy: 0.6955 - val_loss: 1.0013 - val_accuracy: 0.8048\n",
            "Epoch 22/60\n",
            "41/41 [==============================] - 2s 51ms/step - loss: 1.1701 - accuracy: 0.7243 - val_loss: 1.0182 - val_accuracy: 0.8048\n",
            "Epoch 23/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 1.1648 - accuracy: 0.7327 - val_loss: 1.0170 - val_accuracy: 0.8208\n",
            "Epoch 24/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 1.1215 - accuracy: 0.7537 - val_loss: 1.0086 - val_accuracy: 0.8203\n",
            "Epoch 25/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 1.0746 - accuracy: 0.7688 - val_loss: 0.9459 - val_accuracy: 0.8432\n",
            "Epoch 26/60\n",
            "41/41 [==============================] - 2s 51ms/step - loss: 1.0464 - accuracy: 0.7911 - val_loss: 0.9461 - val_accuracy: 0.8426\n",
            "Epoch 27/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 1.0125 - accuracy: 0.7967 - val_loss: 0.9446 - val_accuracy: 0.8409\n",
            "Epoch 28/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 1.0005 - accuracy: 0.8060 - val_loss: 0.9464 - val_accuracy: 0.8449\n",
            "Epoch 29/60\n",
            "41/41 [==============================] - 2s 50ms/step - loss: 0.9454 - accuracy: 0.8146 - val_loss: 0.8926 - val_accuracy: 0.8569\n",
            "Epoch 30/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.9287 - accuracy: 0.8272 - val_loss: 0.9208 - val_accuracy: 0.8500\n",
            "Epoch 31/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.9001 - accuracy: 0.8360 - val_loss: 0.8937 - val_accuracy: 0.8569\n",
            "Epoch 32/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.8785 - accuracy: 0.8393 - val_loss: 0.8899 - val_accuracy: 0.8592\n",
            "Epoch 33/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.8358 - accuracy: 0.8505 - val_loss: 0.9083 - val_accuracy: 0.8620\n",
            "Epoch 34/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 0.8425 - accuracy: 0.8536 - val_loss: 0.8982 - val_accuracy: 0.8643\n",
            "Epoch 35/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.8688 - accuracy: 0.8519 - val_loss: 0.8786 - val_accuracy: 0.8603\n",
            "Epoch 36/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.8130 - accuracy: 0.8654 - val_loss: 0.8519 - val_accuracy: 0.8741\n",
            "Epoch 37/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 0.7678 - accuracy: 0.8763 - val_loss: 0.8891 - val_accuracy: 0.8672\n",
            "Epoch 38/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.7653 - accuracy: 0.8723 - val_loss: 0.8869 - val_accuracy: 0.8712\n",
            "Epoch 39/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.8092 - accuracy: 0.8627 - val_loss: 0.8710 - val_accuracy: 0.8598\n",
            "Epoch 40/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.7447 - accuracy: 0.8855 - val_loss: 0.8590 - val_accuracy: 0.8724\n",
            "Epoch 41/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.7285 - accuracy: 0.8881 - val_loss: 0.8557 - val_accuracy: 0.8712\n",
            "Epoch 42/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.7064 - accuracy: 0.8942 - val_loss: 0.8899 - val_accuracy: 0.8632\n",
            "Epoch 43/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.7635 - accuracy: 0.8835 - val_loss: 0.8836 - val_accuracy: 0.8718\n",
            "Epoch 44/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.7288 - accuracy: 0.8897 - val_loss: 0.8423 - val_accuracy: 0.8792\n",
            "Epoch 45/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.7310 - accuracy: 0.8877 - val_loss: 0.8443 - val_accuracy: 0.8775\n",
            "Epoch 46/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.6803 - accuracy: 0.9074 - val_loss: 0.8230 - val_accuracy: 0.8792\n",
            "Epoch 47/60\n",
            "41/41 [==============================] - 2s 50ms/step - loss: 0.6798 - accuracy: 0.9051 - val_loss: 0.9208 - val_accuracy: 0.8598\n",
            "Epoch 48/60\n",
            "41/41 [==============================] - 2s 51ms/step - loss: 0.6994 - accuracy: 0.8956 - val_loss: 0.8317 - val_accuracy: 0.8890\n",
            "Epoch 49/60\n",
            "41/41 [==============================] - 2s 50ms/step - loss: 0.6551 - accuracy: 0.9097 - val_loss: 0.9037 - val_accuracy: 0.8735\n",
            "Epoch 50/60\n",
            "41/41 [==============================] - 2s 51ms/step - loss: 0.6671 - accuracy: 0.9059 - val_loss: 0.8451 - val_accuracy: 0.8804\n",
            "Epoch 51/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.6624 - accuracy: 0.9086 - val_loss: 0.8265 - val_accuracy: 0.8849\n",
            "Epoch 52/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.6924 - accuracy: 0.9072 - val_loss: 0.9086 - val_accuracy: 0.8603\n",
            "Epoch 53/60\n",
            "41/41 [==============================] - 2s 51ms/step - loss: 0.7006 - accuracy: 0.9017 - val_loss: 0.8506 - val_accuracy: 0.8861\n",
            "Epoch 54/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.6511 - accuracy: 0.9168 - val_loss: 0.9043 - val_accuracy: 0.8775\n",
            "Epoch 55/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 0.6872 - accuracy: 0.9005 - val_loss: 0.9703 - val_accuracy: 0.8443\n",
            "Epoch 56/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.6689 - accuracy: 0.9072 - val_loss: 0.8401 - val_accuracy: 0.8741\n",
            "Epoch 57/60\n",
            "41/41 [==============================] - 2s 48ms/step - loss: 0.6407 - accuracy: 0.9158 - val_loss: 0.8632 - val_accuracy: 0.8804\n",
            "Epoch 58/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 0.6335 - accuracy: 0.9223 - val_loss: 0.8200 - val_accuracy: 0.8924\n",
            "Epoch 59/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 0.5819 - accuracy: 0.9278 - val_loss: 0.8659 - val_accuracy: 0.8804\n",
            "Epoch 60/60\n",
            "41/41 [==============================] - 2s 49ms/step - loss: 0.6100 - accuracy: 0.9244 - val_loss: 0.8187 - val_accuracy: 0.8849\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Test the **model**"
      ],
      "metadata": {
        "id": "FdFwsGHQhchx"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.evaluate(X_test, y_test)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IW6wKXF2hcrK",
        "outputId": "160d8f81-7633-4c4e-9215-0c1b2c147ba6"
      },
      "execution_count": 62,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "55/55 [==============================] - 0s 7ms/step - loss: 0.8058 - accuracy: 0.8769\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.8058210015296936, 0.87693190574646]"
            ]
          },
          "metadata": {},
          "execution_count": 62
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Visualize **loss** graph "
      ],
      "metadata": {
        "id": "5uCWE8kKiWhL"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Plot the the training loss\n",
        "plt.plot(results.history[\"loss\"], label=\"loss\")\n",
        "\n",
        "# Plot the the validation loss\n",
        "plt.plot(results.history[\"val_loss\"], label=\"val_loss\")\n",
        "\n",
        "# Name the x and y axises\n",
        "plt.xlabel(\"Epoch\")\n",
        "plt.ylabel(\"loss\")\n",
        "\n",
        "# Put legend table\n",
        "plt.legend()\n",
        "\n",
        "# Show the plot\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "5b6CxyEhid45",
        "outputId": "a1ec24fc-0a21-43ac-d866-59454f61e87a"
      },
      "execution_count": 63,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Visualize **accuracy** graph"
      ],
      "metadata": {
        "id": "qbpEv_TiieA4"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Plot the the training accuracy\n",
        "plt.plot(results.history[\"accuracy\"], label=\"accuracy\")\n",
        "\n",
        "# Plot the the validation accuracy\n",
        "plt.plot(results.history[\"val_accuracy\"], label=\"val_accuracy\")\n",
        "\n",
        "# Name the x and y axises\n",
        "plt.xlabel(\"Epoch\")\n",
        "plt.ylabel(\"accuracy\")\n",
        "\n",
        "# Put legend table\n",
        "plt.legend()\n",
        "\n",
        "# Show the plot\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "ABKdh2ckieIA",
        "outputId": "1ff05454-439a-4984-9c04-b7d80f039fa3"
      },
      "execution_count": 64,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Try a prediction"
      ],
      "metadata": {
        "id": "VxewhZqarYF5"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_index = 4\n",
        "# Make prediction on the reshaped sample\n",
        "prediction_result = model.predict(X_test[test_index].reshape(1,64,64,1))\n",
        "correct_result = y_test[test_index]\n",
        "\n",
        "# Print the prediction result\n",
        "print(prediction_result)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mJKvwG-fraRB",
        "outputId": "953e9718-d602-46e4-9371-e8dabd30cb7b"
      },
      "execution_count": 67,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[1.5675084e-03 4.0395471e-06 1.2680747e-05 5.6565998e-05 1.8196512e-05\n",
            "  9.9818039e-01 5.9109343e-08 5.4513817e-05 8.5635424e-05 2.0404290e-05]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(correct_result)\n",
        "# Find the predicted class\n",
        "print(prediction_result.argmax())\n",
        "\n",
        "# Find the prediction probability\n",
        "print(prediction_result.max())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "flaw94-wrenh",
        "outputId": "48bf6f03-1e57-49e1-c34a-c3002d4d58f4"
      },
      "execution_count": 68,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "5.0\n",
            "5\n",
            "0.9981804\n"
          ]
        }
      ]
    }
  ]
}