Skip to content

Instantly share code, notes, and snippets.

Embed
What would you like to do?
CNN Untuk Pemula - CIFAR10 v2.ipynb
{
"nbformat": 4,
"nbformat_minor": 0,
"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": "3.5.2"
},
"colab": {
"name": "CNN Untuk Pemula - CIFAR10 v2.ipynb",
"provenance": [],
"collapsed_sections": [],
"include_colab_link": true
}
},
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "view-in-github",
"colab_type": "text"
},
"source": [
"<a href=\"https://colab.research.google.com/gist/datanduth/a864e4328113495e14ad1d187e42b13b/cnn-untuk-pemula-cifar10-v2.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "3nIA0sYxZ4gV",
"colab_type": "text"
},
"source": [
"Model yang kita gunakan ini sengaja saya buat persis seperti [CNN Untuk Pemula - MINIST](https://gist.github.com/datanduth/736e840e98e13a090b6185ef63e5e118) yang saya buat sebelumnya.\n",
"\n",
"Melalui tulisan ini, saya ingin menunjukkan bahwa model CNN yang sama, akan memiliki hasil yang berbeda ketika jenis dataset yang digunakan berbeda.\n",
"\n",
"Sama seperti tulisan sebelumnya, langkah pertama yang perlu anda lakukan adalah import [Keras: The Python Deep Learning library](https://keras.io/).\n"
]
},
{
"cell_type": "code",
"metadata": {
"id": "jBOjUAr53c8P",
"colab_type": "code",
"outputId": "12279ab8-d322-4d1c-e361-47decb724b0a",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 51
}
},
"source": [
"import keras\n",
"keras.__version__"
],
"execution_count": 1,
"outputs": [
{
"output_type": "stream",
"text": [
"Using TensorFlow backend.\n"
],
"name": "stderr"
},
{
"output_type": "execute_result",
"data": {
"text/plain": [
"'2.3.1'"
]
},
"metadata": {
"tags": []
},
"execution_count": 1
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "7AFOnayqcc7t",
"colab_type": "text"
},
"source": [
"Langkah berikutnya adalah menyiapkan dataset yang akan kita gunakan. Pada tulisan ini, dataset yang digunakan adalah [CIFAR10](https://keras.io/datasets/#cifar10-small-image-classification). Dataset CIFAR10 telah tersedia dalam library keras, sehingga kita bisa mengimpor CIFAR10 dari library keras."
]
},
{
"cell_type": "code",
"metadata": {
"id": "aVQG8aFIel7c",
"colab_type": "code",
"colab": {}
},
"source": [
"from keras.datasets import cifar10\n",
"\n",
"(train_images, train_labels), (test_images, test_labels) = cifar10.load_data()"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "ews_oHTDe2z_",
"colab_type": "text"
},
"source": [
"Dataset CIFAR10 berisi gambar `airplane`, `automobile`, `bird`, `cat`, `deer`, `dog`, `frog`, `horse`, `ship`, `truck`.\n",
"\n",
"Masing-masing gambar berukuran 32 x 32 pixel dengan 3 channel warna (RGB).\n",
"\n",
"Channel warna dalam CIFAR10 memiliki rentang nilai antara 0 s.d. 255.\n",
"\n",
"Terdapat 50,000 gambar untuk training dan 10,000 gambar untuk testing dalam dataset tersebut.\n"
]
},
{
"cell_type": "code",
"metadata": {
"id": "19PLMHqN0zVx",
"colab_type": "code",
"outputId": "f5ce4981-c408-4c3f-efcb-fffd0c925136",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 340
}
},
"source": [
"def data_summary():\n",
" print(\"train_images\")\n",
" print(\"shape =\", train_images.shape)\n",
" print(\"max =\", train_images.max())\n",
" print(\"max =\", train_images.min())\n",
"\n",
" print(\"\\ntest_images\")\n",
" print(\"shape =\", test_images.shape)\n",
" print(\"max =\", test_images.max())\n",
" print(\"max =\", test_images.min())\n",
"\n",
" print(\"\\ntrain_labels\")\n",
" print(\"shape =\", train_labels.shape)\n",
" print(\"max =\", train_labels.max())\n",
" print(\"max =\", train_labels.min())\n",
"\n",
" print(\"\\ntest_labels.shape\")\n",
" print(\"shape =\", test_labels.shape)\n",
" print(\"max =\", test_labels.max())\n",
" print(\"max =\", test_labels.min())\n",
"\n",
"\n",
"data_summary()"
],
"execution_count": 3,
"outputs": [
{
"output_type": "stream",
"text": [
"train_images\n",
"shape = (50000, 32, 32, 3)\n",
"max = 255\n",
"max = 0\n",
"\n",
"test_images\n",
"shape = (10000, 32, 32, 3)\n",
"max = 255\n",
"max = 0\n",
"\n",
"train_labels\n",
"shape = (50000, 1)\n",
"max = 9\n",
"max = 0\n",
"\n",
"test_labels.shape\n",
"shape = (10000, 1)\n",
"max = 9\n",
"max = 0\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "_nDv6I8d7J_R",
"colab_type": "text"
},
"source": [
"Bila ingin memvisualisasikan dataset tersebut, kita bisa menggunakan perintah berikut ini."
]
},
{
"cell_type": "code",
"metadata": {
"id": "V1mqNZMH7Qls",
"colab_type": "code",
"outputId": "a5d5b84e-4ba8-4cc7-f736-a9c6dcff9500",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 306
}
},
"source": [
"import matplotlib.pyplot as plt\n",
"\n",
"plt.figure(figsize=(5,5))\n",
"for i in range(9):\n",
" plt.subplot(3,3,i+1)\n",
" plt.xticks([])\n",
" plt.yticks([])\n",
" plt.grid(False)\n",
" plt.imshow(train_images[i], cmap=plt.cm.binary)\n",
"plt.show()\n"
],
"execution_count": 4,
"outputs": [
{
"output_type": "display_data",
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 360x360 with 9 Axes>"
]
},
"metadata": {
"tags": []
}
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "QUYPCbMjuy3S",
"colab_type": "text"
},
"source": [
"# 1. Data Preprocessing\n",
"\n",
"Agar dataset CIFAR10 dapat diproses menggunakan CNN, maka perlu dilakukan penyesuaian format data tersebut, yaitu melakukan normalisasi dengan cara mengubah range data dari 0-255 menjadi 0-1.\n",
"\n",
"Jika sebelumnya kita mengubah data berjenis kategori (data label 0-9) menjadi [binary class matrix](https://keras.io/utils/#to_categorical), maka pada tulisan ini kita tidak melakukannya karena kita akan menggunakan parameter [`loss='sparse_categorical_crossentropy'`](https://jovianlin.io/cat-crossentropy-vs-sparse-cat-crossentropy/) saat melakukan kompilasi CNNN.\n",
"\n"
]
},
{
"cell_type": "code",
"metadata": {
"id": "i319N7Yrubjr",
"colab_type": "code",
"outputId": "5639ef52-e7a4-4c9b-901c-a512d115cd35",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 340
}
},
"source": [
"from keras.utils import to_categorical\n",
"\n",
"train_images = train_images.astype('float32') / 255\n",
"test_images = test_images.astype('float32') / 255\n",
"\n",
"data_summary()"
],
"execution_count": 5,
"outputs": [
{
"output_type": "stream",
"text": [
"train_images\n",
"shape = (50000, 32, 32, 3)\n",
"max = 1.0\n",
"max = 0.0\n",
"\n",
"test_images\n",
"shape = (10000, 32, 32, 3)\n",
"max = 1.0\n",
"max = 0.0\n",
"\n",
"train_labels\n",
"shape = (50000, 1)\n",
"max = 9\n",
"max = 0\n",
"\n",
"test_labels.shape\n",
"shape = (10000, 1)\n",
"max = 9\n",
"max = 0\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "pQBcjhrg0Pmj",
"colab_type": "text"
},
"source": [
"Sekarang kita akan masuk tahap berikutnya, yaitu membuat model CNN.\n",
"\n",
"# 2. Pembuatan Model CNN\n",
"\n",
"Enam layer dalam kode berikut ini adalah model CNN yang paling umum dan mendasar. Model tersebut terdiri dari beberapa layer [`Conv2D`](https://keras.io/layers/convolutional/#conv2d) dan [`MaxPooling2D`](https://keras.io/layers/pooling/#maxpooling2d).\n",
"\n",
"*Convolutional Neural* (*convets*) memerlukan input dalam bentuk `(image_height, image_width, image_channels)` (jumlah `samples` tidak termasuk dalam parameter yang diperhitungkan). Itulah sebabnya kita menuliskan `input_shape=(32, 32, 3)` saat pembuatan `convets` layer."
]
},
{
"cell_type": "code",
"metadata": {
"id": "RwMXtpfN6At9",
"colab_type": "code",
"colab": {}
},
"source": [
"from keras import layers\n",
"from keras import models\n",
"\n",
"model = models.Sequential()\n",
"model.add(layers.Conv2D(32, (3, 3), padding='valid', activation='relu', input_shape=(32, 32, 3)))\n",
"model.add(layers.MaxPooling2D((2, 2)))\n",
"model.add(layers.Conv2D(64, (3, 3), padding='valid', activation='relu'))\n",
"model.add(layers.MaxPooling2D((2, 2)))\n",
"model.add(layers.Conv2D(64, (3, 3), padding='valid', activation='relu'))"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "S0nzDVvF3c8c",
"colab_type": "text"
},
"source": [
"Untuk memahami CNN model tersebut, mari kita bongkar baris demi baris.\n",
"\n",
"## Jenis Neural Network\n",
"\n",
"`model = models.Sequential()`\n",
"\n",
"Kode tersebut akan membuat sebuah Neural Network layer yang sifatnya [linier](https://keras.io/getting-started/sequential-model-guide/).\n",
"\n",
"## Convolution\n",
"\n",
"Secara mudah, convolution adalah mengimplementasikan [kernel](https://setosa.io/ev/image-kernels/) atau filter ke dalam sebuah gambar.\n",
"\n",
"`model.add(layers.Conv2D(32, (3, 3), activation='relu', input_shape=(32, 32, 1)))`\n",
"\n",
"1. `32` menandakan jumlah kernel atau filter yang digunakan adalah 32.\n",
"2. `(3, 3)` menandakan ukuran kernal atau filter yang digunakan adalah 3 x 3. \n",
"3. `padding='valid'` berarti kita tidak menggunakan *padding* yang nantinya akan menyebabkan pengurangan ukuran *output* menjadi 26 x 26 (lihat ilustrasi *convolution* pada gambar di bawah.\n",
"4. `activation='relu'` berarti *activation function* yang digunakan adalah [*Rectified Linear Unit*](https://machinelearningmastery.com/rectified-linear-activation-function-for-deep-learning-neural-networks/).\n",
"5. `input_shape=(32, 32, 3)` berarti input layer tersebut adalah 3D tensor`(height, width, channels)`. `height = 32`, `width = 32`, dan `channels = 3` karena MINIST adalah gambar *RGB*. \n",
"\n",
"Berikut ilustrasi proses *convolution* yang saya ambil dari website [a-ydobon](https://medium.com/a-ydobon/tensorflow-2-0-convolutional-neural-network-cnn-4f90ddc3109e)\n",
"\n",
"![alt text](https://miro.medium.com/max/1356/1*SRMHCK0QuV73enkxir5B1Q.png)\n",
"\n",
"Bila masih bingung dengan cara kerja sebuah kernel atau filter, ilustrasi pada [website ini](https://setosa.io/ev/image-kernels/) akan sangat membantu.\n",
"\n",
"Pada `convolutional` kernel selanjutnya\n",
"\n",
"`model.add(layers.Conv2D(64, (3, 3), padding='valid', activation='relu'))`\n",
"\n",
"tidak perlu lagi mendeklarasikan parameter `input_shape` karena secara otomatis akan disesuaikan dengan layer sebelumnya. \n",
"\n",
"Bila anda perhatikan, jumlah kernel atau filter yang digunakan pada layer berikutnya meningkat menjadi 64. Hal ini disebabkan karena setelah melalui layer `convolutional` pertama, dimensi input data telah dikurangi, sehingga memungkinkan untuk memproses dengan lebih banyak kenel.\n",
"\n",
"## MaxPooling\n",
"\n",
"Sesuai namanya, layer ini akan melakukan \"pooling\" untuk mencari nilai maksimal dari hasil *convolution*. Dengan kata lain ini adalah proses *downsampling* untuk mengurasi dimensi data. `MaxPooling` kan melakukan *scanning* gambar dengan dimensi tertentu dan memilih data dengan nilai maksimal.\n",
"\n",
"`model.add(layers.MaxPooling2D((2, 2)))`\n",
"\n",
"Perintah tersebut akan membuat sebuah layer `MaxPooling`. Parameter `(2, 2)` akan membuat sebuah kernel berukuran 2 dan kenel tersebut akan bergeser sepanjang 2.\n",
"\n",
"Berikut ilustrasi proses `MaxPooling` yang saya ambil dari website [a-ydobon](https://medium.com/a-ydobon/tensorflow-2-0-convolutional-neural-network-cnn-4f90ddc3109e)\n",
"\n",
"![alt text](https://miro.medium.com/max/1072/1*1ToopU-sgt-iGlKfE8j6vA.png)\n",
"\n",
"Sekarang kita bisa melihat rangkuman dari hasil pembuatan layer tersebut dengan perintah berikut.\n"
]
},
{
"cell_type": "code",
"metadata": {
"id": "8fcM_9uQ6QMO",
"colab_type": "code",
"outputId": "9cb0d959-007f-4c57-e4dc-9ceeee0241d2",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 323
}
},
"source": [
"model.summary()"
],
"execution_count": 7,
"outputs": [
{
"output_type": "stream",
"text": [
"Model: \"sequential_1\"\n",
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
"conv2d_1 (Conv2D) (None, 30, 30, 32) 896 \n",
"_________________________________________________________________\n",
"max_pooling2d_1 (MaxPooling2 (None, 15, 15, 32) 0 \n",
"_________________________________________________________________\n",
"conv2d_2 (Conv2D) (None, 13, 13, 64) 18496 \n",
"_________________________________________________________________\n",
"max_pooling2d_2 (MaxPooling2 (None, 6, 6, 64) 0 \n",
"_________________________________________________________________\n",
"conv2d_3 (Conv2D) (None, 4, 4, 64) 36928 \n",
"=================================================================\n",
"Total params: 56,320\n",
"Trainable params: 56,320\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "markdown",
"metadata": {
"collapsed": true,
"id": "dwv0jy0S3c8l",
"colab_type": "text"
},
"source": [
"## Layer convolution pertama\n",
"\n",
"`model.add(layers.Conv2D(32, (3, 3), padding='valid', activation='relu', input_shape=(32, 32, 3)))`\n",
"\n",
"akan menghasilkan output dengan dimensi\n",
"\n",
"`(None, 30, 30, 32)` dengan jumlah parameter `896`.\n",
"\n",
"`30` didapatkan dari dimensi input `32 - 2` *(dimensi input - padding)*.\n",
"\n",
"`32` didapatkan dari jumlah kernel yang digunakan dalam satu layer.\n",
"\n",
"`896` didapatkan dari `((3 x 3 x 3) + 1) x 32` *((kernel height x kenel weight x image_channels) + [bias parameter](https://towardsdatascience.com/convolutional-neural-networks-for-beginners-practical-guide-with-python-and-keras-dc688ea90dca)) x jumlah kenel)*\n",
"\n",
"## Layer MaxPooling pertama\n",
"\n",
"`model.add(layers.MaxPooling2D((2, 2)))`\n",
"\n",
"akan menghasilkan output dengan dimensi\n",
"\n",
"`(None, 15, 15, 32)`\n",
"\n",
"`15` didapatkan dari `30 / 2` *(dimensi output layer convolutional pertama / dimensi kernel `MaxPooling2D`)*.\n",
"\n",
"`32` didapatkan dari jumlah kernel dari output layer *convolutional* pertama.\n",
"\n",
"## Layer convolution kedua\n",
"\n",
"Dari penjelasan dua layer sebelunya, sebenarnya anda sudah bisa menebak output dari layer *convolution* kedua ini.\n",
"\n",
"`model.add(layers.Conv2D(64, (3, 3), padding='valid', activation='relu'))`\n",
"\n",
"yaitu\n",
"\n",
"`(None, 13, 13, 64)` dengan jumlah parameter `18496`.\n",
"\n",
"`13` didapatkan dari `15 - 2` *(dimensi output layer `MaxPooling` pertama - padding)*.\n",
"\n",
"`64` didapatkan dari jumlah kernel yang digunakan dalam satu layer.\n",
"\n",
"`18496` didapatkan dari `((3 x 3 x 32) + 1) x 64` *((kernel height x kenel weight x feature map layer `MaxPooling` pertama) + bias parameter) x jumlah kenel)*\n",
"\n",
"## Layer MaxPooling kedua\n",
"\n",
"`model.add(layers.MaxPooling2D((2, 2)))`\n",
"\n",
"akan menghasilkan output dengan dimensi\n",
"\n",
"`(None, 6, 6, 64)`\n",
"\n",
"`6` didapatkan dari `13 / 2` *(dimensi output layer convolutional kedua / dimensi kernel `MaxPooling2D`)*.\n",
"\n",
"`64` didapatkan dari jumlah kernel dari output layer *convolutional* edua.\n",
"\n",
"## Layer convolution ketiga\n",
"\n",
"`model.add(layers.Conv2D(64, (3, 3), padding='valid', activation='relu'))`\n",
"\n",
"yaitu\n",
"\n",
"`(None, 4, 4, 64)` dengan jumlah parameter `36928`.\n",
"\n",
"`4` didapatkan dari `6 - 2` *(dimensi output layer `MaxPooling` kedua - padding)*.\n",
"\n",
"`64` didapatkan dari jumlah kernel yang digunakan dalam satu layer.\n",
"\n",
"`36928` didapatkan dari `((3 x 3 x 64) + 1) x 64` *((kernel height x kenel weight x feature map layer `MaxPooling` kedua) + bias parameter) x jumlah kenel)*"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "Rqel1gIPfQA7",
"colab_type": "text"
},
"source": [
"# Dense Layer\n",
"\n",
"Layer penutup setelah *convolutional* layer adalah *Dense* layer. Sekali lagi ini adalah model CNN yang sangat umum digunakan. *Dense* layer ini akan kita gunakan untuk melakukan klasifikasi dari hasil proses *convolutional* yang telah dilakukan.\n",
"\n",
"Karena *Dense* layer hanya memerlukan data 1D, maka kita harus melakukan menambah satu layer untuk melakukan transformasi data tersebut dengan perintah: "
]
},
{
"cell_type": "code",
"metadata": {
"id": "v1RtQBrs3c8l",
"colab_type": "code",
"colab": {}
},
"source": [
"model.add(layers.Flatten())"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "VSe0IOqGg_k_",
"colab_type": "text"
},
"source": [
"Lalu tambahkan dua *Dense* layer dengan parameter berikut"
]
},
{
"cell_type": "code",
"metadata": {
"id": "5oERxsI3hAx5",
"colab_type": "code",
"colab": {}
},
"source": [
"model.add(layers.Dense(64, activation='relu'))\n",
"model.add(layers.Dense(10, activation='softmax'))"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "l4cnlmhb3c8q",
"colab_type": "text"
},
"source": [
"Tampilkan hasil akhir dari CNN model yang telah dibuat dengan perintah berikut"
]
},
{
"cell_type": "code",
"metadata": {
"id": "i3y0Y_U53c8q",
"colab_type": "code",
"outputId": "f5acdddf-7b61-4586-c1f4-bcb7a30ddd0e",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 425
}
},
"source": [
"model.summary()"
],
"execution_count": 10,
"outputs": [
{
"output_type": "stream",
"text": [
"Model: \"sequential_1\"\n",
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
"conv2d_1 (Conv2D) (None, 30, 30, 32) 896 \n",
"_________________________________________________________________\n",
"max_pooling2d_1 (MaxPooling2 (None, 15, 15, 32) 0 \n",
"_________________________________________________________________\n",
"conv2d_2 (Conv2D) (None, 13, 13, 64) 18496 \n",
"_________________________________________________________________\n",
"max_pooling2d_2 (MaxPooling2 (None, 6, 6, 64) 0 \n",
"_________________________________________________________________\n",
"conv2d_3 (Conv2D) (None, 4, 4, 64) 36928 \n",
"_________________________________________________________________\n",
"flatten_1 (Flatten) (None, 1024) 0 \n",
"_________________________________________________________________\n",
"dense_1 (Dense) (None, 64) 65600 \n",
"_________________________________________________________________\n",
"dense_2 (Dense) (None, 10) 650 \n",
"=================================================================\n",
"Total params: 122,570\n",
"Trainable params: 122,570\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "5rUOkfcf1qaP",
"colab_type": "text"
},
"source": [
"## Flatten Layer\n",
"\n",
"`model.add(layers.Flatten())`\n",
"\n",
"Flatten layer akan mengubah dimensi layer sebelumnya dari `(None, 4, 4, 64)` menjadi `(None, 1024) `.\n",
"\n",
"`1024` didapatkan dari `4 x 4 x 64` *(kernel height x kenel weight x feature map layer `convolutional` ketiga)*\n",
"\n",
"## Dense Layer\n",
"\n",
"`Dense` layer pertama akan mengurangi dimensi feature map dari `1024` menjadi `64`. Sedangkan `Dense` layer kedua akan mengurangi dimensi feature map dari `64` menjadi `10` sesuai dengan klasifikasi kelas yang diinginkan.\n",
"\n",
"# Compile\n",
"\n",
"Langkah berikutnya adalah melakukan kompilasi."
]
},
{
"cell_type": "code",
"metadata": {
"id": "fy60pIsY3c80",
"colab_type": "code",
"colab": {}
},
"source": [
"model.compile(optimizer='rmsprop',\n",
" loss='sparse_categorical_crossentropy',\n",
" metrics=['accuracy'])"
],
"execution_count": 0,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "kWZ8r0Ra3eXy",
"colab_type": "text"
},
"source": [
"`optimizer='rmsprop'` menandakan jenis [`optimizer`](https://keras.io/optimizers/) yang kita gunakan adalah [`rmsprop`](http://www.cs.toronto.edu/~tijmen/csc321/slides/lecture_slides_lec6.pdf)\n",
"\n",
"`loss='categorical_crossentropy'` menandakan jenis [`loss`](https://keras.io/losses/) yang kita gunakan adalah [`categorical_crossentropy`](https://keras.io/backend/#categorical_crossentropy).\n",
"\n",
"`metrics=['accuracy']` menandakan jenis [`metrics`](https://keras.io/metrics/) yang kita gunakan adalah [`accuracy`](https://keras.io/metrics/#available-metrics).\n",
"\n",
"# Training\n",
"\n",
"Kemudian melakukan training model yang telah kita buat."
]
},
{
"cell_type": "code",
"metadata": {
"id": "IUQxRcxf3iXt",
"colab_type": "code",
"outputId": "59076c84-4cf1-4479-9f1b-421ea902b678",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 204
}
},
"source": [
"model.fit(train_images, train_labels, epochs=5, batch_size=64)"
],
"execution_count": 12,
"outputs": [
{
"output_type": "stream",
"text": [
"Epoch 1/5\n",
"50000/50000 [==============================] - 62s 1ms/step - loss: 1.6095 - accuracy: 0.4207\n",
"Epoch 2/5\n",
"50000/50000 [==============================] - 61s 1ms/step - loss: 1.2244 - accuracy: 0.5690\n",
"Epoch 3/5\n",
"50000/50000 [==============================] - 61s 1ms/step - loss: 1.0506 - accuracy: 0.6331\n",
"Epoch 4/5\n",
"50000/50000 [==============================] - 61s 1ms/step - loss: 0.9268 - accuracy: 0.6762\n",
"Epoch 5/5\n",
"50000/50000 [==============================] - 61s 1ms/step - loss: 0.8355 - accuracy: 0.7102\n"
],
"name": "stdout"
},
{
"output_type": "execute_result",
"data": {
"text/plain": [
"<keras.callbacks.callbacks.History at 0x7fa262da65f8>"
]
},
"metadata": {
"tags": []
},
"execution_count": 12
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "abLnmjxx3c84",
"colab_type": "text"
},
"source": [
"[`epochs`](https://keras.io/models/model/)`='5'` menandakan training akan dilakukan sebanyak 5 kali.\n",
"\n",
"[`batch_size`](https://keras.io/models/model/)`='64'` menandakan jumlah sample tiap gradient update.\n",
"\n",
"# Evaluasi Model\n",
"\n",
"Setelah traning, langkah terakhir adalah mengevaluasi model yang telah dibuat dengan perintah berikut "
]
},
{
"cell_type": "code",
"metadata": {
"id": "tBS4IOLy3c85",
"colab_type": "code",
"outputId": "f746a213-9812-4cef-b8b8-594e068a63ce",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 34
}
},
"source": [
"test_loss, test_acc = model.evaluate(test_images, test_labels)"
],
"execution_count": 13,
"outputs": [
{
"output_type": "stream",
"text": [
"10000/10000 [==============================] - 4s 373us/step\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "x8GCIuUx3c88",
"colab_type": "code",
"outputId": "73bb65f1-e146-43e6-f1e1-7be3904e499a",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 51
}
},
"source": [
"print('test_acc =', test_acc)\n",
"print('test_loss =', test_loss)"
],
"execution_count": 14,
"outputs": [
{
"output_type": "stream",
"text": [
"test_acc = 0.6101999878883362\n",
"test_loss = 1.1348929668426513\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "wetNtCf63c9A",
"colab_type": "text"
},
"source": [
"Hasil evaluasi menunjukkan bahwa CNN model yang kita buat memiliki akurasi sebesar **0.61** dan loss sebesar **1.13**. \n",
"\n",
"Hasil tersebut sangat berbeda dengan [CNN Untuk Pemula - MINIST](https://gist.github.com/datanduth/736e840e98e13a090b6185ef63e5e118) yang saya buat sebelumnya, dimana kita mendapatkan akurasi sebesar **0.99** dan loss sebesar **0.03**.\n",
"\n",
"Sehingga kesimpulan yang bisa ambil adalah, tiap model CNN akan efektif untuk satu jenis dataset / task saja. Dengan teknologi saat ini, cukup sulit membuat model CNN yang bisa efektif untuk semua jenis dataset / task."
]
}
]
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment