From b7d772d1153191e23c90cce387574bdffa939b67 Mon Sep 17 00:00:00 2001
From: Mirko Birbaumer <mirko.birbaumer@hslu.ch>
Date: Thu, 3 Mar 2022 01:13:53 +0000
Subject: [PATCH] Adapted MNIST example
---
... Notebook Block 2 - Neural Networks .ipynb | 1155 ++++++-----------
1 file changed, 381 insertions(+), 774 deletions(-)
diff --git a/notebooks/Block_2/Jupyter Notebook Block 2 - Neural Networks .ipynb b/notebooks/Block_2/Jupyter Notebook Block 2 - Neural Networks .ipynb
index 842df71..cd230a5 100644
--- a/notebooks/Block_2/Jupyter Notebook Block 2 - Neural Networks .ipynb
+++ b/notebooks/Block_2/Jupyter Notebook Block 2 - Neural Networks .ipynb
@@ -3323,182 +3323,312 @@
"# Part 3 : Keras for MNIST"
]
},
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "Let’s look at a concrete example of a neural network that uses the Python library\n",
+ "Keras to learn to classify handwritten digits. \n",
+ "\n",
+ "The problem we’re trying to solve here is to classify grayscale images of handwritten\n",
+ "digits (28x28 pixels) into their 10 categories (0 through 9). We’ll use the MNIST\n",
+ "dataset, a classic in the machine learning community, which has been around almost\n",
+ "as long as the field itself and has been intensively studied. It’s a set of 60'000 training images, plus 10'000 test images, assembled by the National Institute of Standards and Technology (the NIST in MNIST) in the 1980s. You can think of \n",
+ "“solving†MNIST as the “Hello World†of deep learning—it’s what you do to verify \n",
+ "that your algorithms are working as expected. As you become a machine learning practitioner, you’ll see MNIST come up over and over again in scientific papers, blog posts, and so on."
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "## Loading the MNIST dataset in Keras\n",
+ "The MNIST dataset comes preloaded in Keras, in the form of a set of four NumPy\n",
+ "arrays."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 72,
+ "metadata": {},
+ "outputs": [],
+ "source": [
+ "from tensorflow.keras.datasets import mnist\n",
+ "(X_train, y_train), (X_test, y_test) = mnist.load_data()"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "`X_train` and `y_train` form the training set, the data that the model will\n",
+ "learn from. The model will then be tested on the test set, `X_test` and `y_test`. The images are encoded as NumPy arrays, and the labels are an array of digits, ranging from 0 to 9. The images and labels have a one-to-one correspondence.\n",
+ "Let’s look at the training data:"
+ ]
+ },
{
"cell_type": "code",
- "execution_count": 32,
+ "execution_count": 73,
"metadata": {},
"outputs": [
{
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "Requirement already satisfied: tensorflow-datasets in /opt/conda/lib/python3.7/site-packages (4.5.2)\n",
- "Requirement already satisfied: dill in /opt/conda/lib/python3.7/site-packages (from tensorflow-datasets) (0.3.4)\n",
- "Requirement already satisfied: absl-py in /opt/conda/lib/python3.7/site-packages (from tensorflow-datasets) (1.0.0)\n",
- "Requirement already satisfied: numpy in /opt/conda/lib/python3.7/site-packages (from tensorflow-datasets) (1.19.1)\n",
- "Requirement already satisfied: tensorflow-metadata in /opt/conda/lib/python3.7/site-packages (from tensorflow-datasets) (1.6.0)\n",
- "Requirement already satisfied: typing-extensions; python_version < \"3.8\" in /opt/conda/lib/python3.7/site-packages (from tensorflow-datasets) (3.7.4.3)\n",
- "Requirement already satisfied: promise in /opt/conda/lib/python3.7/site-packages (from tensorflow-datasets) (2.3)\n",
- "Requirement already satisfied: six in /opt/conda/lib/python3.7/site-packages (from tensorflow-datasets) (1.14.0)\n",
- "Requirement already satisfied: importlib-resources; python_version < \"3.9\" in /opt/conda/lib/python3.7/site-packages (from tensorflow-datasets) (5.4.0)\n",
- "Requirement already satisfied: tqdm in /opt/conda/lib/python3.7/site-packages (from tensorflow-datasets) (4.45.0)\n",
- "Requirement already satisfied: requests>=2.19.0 in /opt/conda/lib/python3.7/site-packages (from tensorflow-datasets) (2.23.0)\n",
- "Requirement already satisfied: protobuf>=3.12.2 in /opt/conda/lib/python3.7/site-packages (from tensorflow-datasets) (3.19.4)\n",
- "Requirement already satisfied: termcolor in /opt/conda/lib/python3.7/site-packages (from tensorflow-datasets) (1.1.0)\n",
- "Requirement already satisfied: googleapis-common-protos<2,>=1.52.0 in /opt/conda/lib/python3.7/site-packages (from tensorflow-metadata->tensorflow-datasets) (1.55.0)\n",
- "Requirement already satisfied: zipp>=3.1.0; python_version < \"3.10\" in /opt/conda/lib/python3.7/site-packages (from importlib-resources; python_version < \"3.9\"->tensorflow-datasets) (3.1.0)\n",
- "Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /opt/conda/lib/python3.7/site-packages (from requests>=2.19.0->tensorflow-datasets) (1.25.9)\n",
- "Requirement already satisfied: idna<3,>=2.5 in /opt/conda/lib/python3.7/site-packages (from requests>=2.19.0->tensorflow-datasets) (2.9)\n",
- "Requirement already satisfied: chardet<4,>=3.0.2 in /opt/conda/lib/python3.7/site-packages (from requests>=2.19.0->tensorflow-datasets) (3.0.4)\n",
- "Requirement already satisfied: certifi>=2017.4.17 in /opt/conda/lib/python3.7/site-packages (from requests>=2.19.0->tensorflow-datasets) (2020.6.20)\n",
- "\u001b[33mWARNING: You are using pip version 20.2.4; however, version 22.0.3 is available.\n",
- "You should consider upgrading via the '/opt/conda/bin/python3 -m pip install --upgrade pip' command.\u001b[0m\n",
- "Collecting tfds-nightly\n",
- " Downloading tfds_nightly-4.5.2.dev202202240044-py3-none-any.whl (4.2 MB)\n",
- "\u001b[K |████████████████████████████████| 4.2 MB 5.1 MB/s eta 0:00:01\n",
- "\u001b[?25hRequirement already satisfied: protobuf>=3.12.2 in /opt/conda/lib/python3.7/site-packages (from tfds-nightly) (3.19.4)\n",
- "Requirement already satisfied: numpy in /opt/conda/lib/python3.7/site-packages (from tfds-nightly) (1.19.1)\n",
- "Requirement already satisfied: termcolor in /opt/conda/lib/python3.7/site-packages (from tfds-nightly) (1.1.0)\n",
- "Requirement already satisfied: tensorflow-metadata in /opt/conda/lib/python3.7/site-packages (from tfds-nightly) (1.6.0)\n",
- "Collecting etils[epath-no-tf]\n",
- " Downloading etils-0.4.0-py3-none-any.whl (76 kB)\n",
- "\u001b[K |████████████████████████████████| 76 kB 2.2 MB/s eta 0:00:01\n",
- "\u001b[?25hRequirement already satisfied: dill in /opt/conda/lib/python3.7/site-packages (from tfds-nightly) (0.3.4)\n",
- "Requirement already satisfied: requests>=2.19.0 in /opt/conda/lib/python3.7/site-packages (from tfds-nightly) (2.23.0)\n",
- "Requirement already satisfied: toml in /opt/conda/lib/python3.7/site-packages (from tfds-nightly) (0.10.2)\n",
- "Requirement already satisfied: absl-py in /opt/conda/lib/python3.7/site-packages (from tfds-nightly) (1.0.0)\n",
- "Requirement already satisfied: six in /opt/conda/lib/python3.7/site-packages (from tfds-nightly) (1.14.0)\n",
- "Requirement already satisfied: importlib-resources; python_version < \"3.9\" in /opt/conda/lib/python3.7/site-packages (from tfds-nightly) (5.4.0)\n",
- "Requirement already satisfied: typing-extensions; python_version < \"3.8\" in /opt/conda/lib/python3.7/site-packages (from tfds-nightly) (3.7.4.3)\n",
- "Requirement already satisfied: tqdm in /opt/conda/lib/python3.7/site-packages (from tfds-nightly) (4.45.0)\n",
- "Requirement already satisfied: promise in /opt/conda/lib/python3.7/site-packages (from tfds-nightly) (2.3)\n",
- "Requirement already satisfied: googleapis-common-protos<2,>=1.52.0 in /opt/conda/lib/python3.7/site-packages (from tensorflow-metadata->tfds-nightly) (1.55.0)\n",
- "Requirement already satisfied: zipp; extra == \"epath-no-tf\" in /opt/conda/lib/python3.7/site-packages (from etils[epath-no-tf]->tfds-nightly) (3.1.0)\n",
- "Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /opt/conda/lib/python3.7/site-packages (from requests>=2.19.0->tfds-nightly) (1.25.9)\n",
- "Requirement already satisfied: certifi>=2017.4.17 in /opt/conda/lib/python3.7/site-packages (from requests>=2.19.0->tfds-nightly) (2020.6.20)\n",
- "Requirement already satisfied: chardet<4,>=3.0.2 in /opt/conda/lib/python3.7/site-packages (from requests>=2.19.0->tfds-nightly) (3.0.4)\n",
- "Requirement already satisfied: idna<3,>=2.5 in /opt/conda/lib/python3.7/site-packages (from requests>=2.19.0->tfds-nightly) (2.9)\n",
- "Installing collected packages: etils, tfds-nightly\n",
- "Successfully installed etils-0.4.0 tfds-nightly-4.5.2.dev202202240044\n",
- "\u001b[33mWARNING: You are using pip version 20.2.4; however, version 22.0.3 is available.\n",
- "You should consider upgrading via the '/opt/conda/bin/python3 -m pip install --upgrade pip' command.\u001b[0m\n"
- ]
+ "data": {
+ "text/plain": [
+ "(60000, 28, 28)"
+ ]
+ },
+ "execution_count": 73,
+ "metadata": {},
+ "output_type": "execute_result"
}
],
"source": [
- "!pip install tensorflow-datasets\n",
- "!pip install tfds-nightly"
+ "X_train.shape"
]
},
{
"cell_type": "code",
- "execution_count": 33,
+ "execution_count": 74,
"metadata": {},
"outputs": [
{
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "2.7.1\n"
- ]
+ "data": {
+ "text/plain": [
+ "60000"
+ ]
+ },
+ "execution_count": 74,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "len(y_train)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 75,
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "array([5, 0, 4, ..., 5, 6, 8], dtype=uint8)"
+ ]
+ },
+ "execution_count": 75,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "y_train"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "Let us display an the fourth digit:"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 76,
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ "<Figure size 432x288 with 1 Axes>"
+ ]
+ },
+ "metadata": {
+ "needs_background": "light"
+ },
+ "output_type": "display_data"
}
],
"source": [
- "from __future__ import absolute_import, division, print_function, unicode_literals\n",
- "\n",
- "\n",
- "# Import TensorFlow and TensorFlow Datasets\n",
- "import tensorflow as tf\n",
- "import tensorflow_datasets as tfds\n",
- "tfds.disable_progress_bar()\n",
- "\n",
- "# Helper libraries\n",
- "import math\n",
- "import numpy as np\n",
"import matplotlib.pyplot as plt\n",
- "\n",
- "\n",
- "print(tf.__version__)\n"
+ "digit = X_train[4]\n",
+ "plt.imshow(digit, cmap=plt.cm.binary)\n",
+ "plt.show()"
]
},
{
"cell_type": "code",
- "execution_count": 34,
+ "execution_count": 77,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "9"
+ ]
+ },
+ "execution_count": 77,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
"source": [
- "import logging\n",
- "logger = tf.get_logger()\n",
- "logger.setLevel(logging.ERROR)"
+ "y_train[4]"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "And the test data:"
]
},
{
"cell_type": "code",
- "execution_count": 35,
+ "execution_count": 78,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "(10000, 28, 28)"
+ ]
+ },
+ "execution_count": 78,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
"source": [
- "dataset, metadata = tfds.load('mnist', as_supervised=True, with_info=True)\n",
- "train_dataset, test_dataset = dataset['train'], dataset['test']"
+ "X_test.shape"
]
},
{
"cell_type": "code",
- "execution_count": 36,
+ "execution_count": 79,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "10000"
+ ]
+ },
+ "execution_count": 79,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "len(y_test)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 80,
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "array([7, 2, 1, ..., 4, 5, 6], dtype=uint8)"
+ ]
+ },
+ "execution_count": 80,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "y_test"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "Before training, we’ll preprocess the data by reshaping it into the shape the model\n",
+ "expects and scaling it so that all values are in the $[0, 1]$ interval. Previously, our training images were stored in an array of shape $(60000, 28, 28)$ of type `uint8` with values in the $[0, 255]$ interval. We’ll transform it into a `float32` array of shape $(60000, 28 * 28)$ with values between $0$ and $1$."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 81,
+ "metadata": {},
+ "outputs": [
+ {
+ "name": "stdout",
+ "output_type": "stream",
+ "text": [
+ "(60000, 784)\n",
+ "(10000, 784)\n"
+ ]
+ }
+ ],
"source": [
- "class_names = ['zero', 'one', 'two', 'three', 'four',\n",
- " 'five', 'six', 'seven', 'eight', 'nine']"
+ "X_train = X_train.reshape((60000, 28 * 28))\n",
+ "print(X_train.shape)\n",
+ "X_train = X_train.astype(\"float32\") / 255\n",
+ "X_test = X_test.reshape((10000, 28 * 28))\n",
+ "X_test = X_test.astype(\"float32\") / 255\n",
+ "print(X_test.shape)"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "In addition, we need to _one-hot-encode_ the labels: "
]
},
{
"cell_type": "code",
- "execution_count": 37,
+ "execution_count": 82,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
- "Number of training examples: 60000\n",
- "Number of test examples: 10000\n"
+ "5\n",
+ "[0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n"
]
}
],
"source": [
- "num_train_examples = metadata.splits['train'].num_examples\n",
- "num_test_examples = metadata.splits['test'].num_examples\n",
- "print(\"Number of training examples: {}\".format(num_train_examples))\n",
- "print(\"Number of test examples: {}\".format(num_test_examples))"
+ "from tensorflow.keras.utils import to_categorical\n",
+ "y_train_cat = to_categorical(y_train)\n",
+ "print(y_train[0])\n",
+ "print(y_train_cat[0])"
]
},
{
"cell_type": "code",
- "execution_count": 38,
+ "execution_count": 83,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
- "Shape of image training data : (3000, 784)\n",
- "Shape of training data labels : (3000,)\n"
+ "7\n",
+ "[0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n"
]
}
],
"source": [
- "i=0\n",
- "for (image, label) in train_dataset.take(3000):\n",
- " if i==0:\n",
- " X_train = image.numpy().reshape((1,28*28))\n",
- " y_train = np.array([label])\n",
- " else:\n",
- " X_train = np.concatenate([X_train, image.numpy().reshape((1,28*28))], axis=0)\n",
- " y_train = np.concatenate([y_train, np.array([label])], axis=0)\n",
- " i+=1\n",
- "print(\"Shape of image training data : \", X_train.shape)\n",
- "print(\"Shape of training data labels : \", y_train.shape)"
+ "y_test_cat = to_categorical(y_test)\n",
+ "print(y_test[0])\n",
+ "print(y_test_cat[0])"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "The workflow will be as follows: First, we’ll feed the neural network the training data, `X_train` and `y_train`. The network will then learn to associate images and labels. Finally, we’ll ask the network to produce predictions for test_images, and we’ll verify whether these predictions match the labels from `test_labels`."
]
},
{
@@ -3519,17 +3649,18 @@
},
{
"cell_type": "code",
- "execution_count": 39,
+ "execution_count": 84,
"metadata": {},
"outputs": [],
"source": [
- "model = tf.keras.Sequential()\n",
- "# From Input to first hidden layer\n",
- "model.add(tf.keras.layers.Dense(500, activation= tf.nn.sigmoid, \n",
- " batch_input_shape=(None,784)))\n",
- "model.add(tf.keras.layers.Dense(50, activation=tf.nn.sigmoid))\n",
- "model.add(tf.keras.layers.Dense(10, activation=tf.nn.softmax)) \n",
- "\n"
+ "from tensorflow import keras\n",
+ "from tensorflow.keras import layers\n",
+ "\n",
+ "model = keras.Sequential([\n",
+ "layers.Dense(500, activation=\"relu\", input_shape=(784,)),\n",
+ "layers.Dense(50, activation=\"relu\"), \n",
+ "layers.Dense(10, activation=\"softmax\")\n",
+ "])"
]
},
{
@@ -3540,11 +3671,11 @@
"\n",
"\n",
"\n",
- "* **\"hidden\"** `tf.keras.layers.Dense`— A densely connected layer of 500 neurons. Each neuron (or node) takes input from all 784 nodes in the previous layer, weighting that input according to hidden parameters which will be learned during training, and outputs a single value to the next layer.\n",
+ "* **\"hidden\"** `layers.Dense`— A densely connected layer of 500 neurons. Each neuron (or node) takes input from all 784 nodes in the previous layer - by specifying an `input_shape` to the first layer in the Sequential model - , weighting that input according to hidden parameters which will be learned during training, and outputs a single value to the next layer. \n",
"\n",
- "* **\"hidden\"** `tf.keras.layers.Dense`— A densely connected layer of 50 neurons. Each neuron (or node) takes input from all 500 nodes in the previous layer, weighting that input according to hidden parameters which will be learned during training, and outputs a single value to the next layer.\n",
+ "* **\"hidden\"** `layers.Dense`— A densely connected layer of 50 neurons. Each neuron (or node) takes input from all 500 nodes in the previous layer, weighting that input according to hidden parameters which will be learned during training, and outputs a single value to the next layer.\n",
"\n",
- "* **output** `tf.keras.layers.Dense` — A 10-node *softmax* layer, with each node representing a class of clothing. As in the previous layer, each node takes input from the 50 nodes in the layer before it. Each node weights the input according to learned parameters, and then outputs a value in the range `[0, 1]`, representing the probability that the image belongs to that class. The sum of all 10 node values is 1.\n"
+ "* **output** `layers.Dense` — A 10-node *softmax* layer, with each node representing a class of clothing. As in the previous layer, each node takes input from the 50 nodes in the layer before it. Each node weights the input according to learned parameters, and then outputs a value in the range `[0, 1]`, representing the probability that the image belongs to that class. The sum of all 10 node values is 1.\n"
]
},
{
@@ -3564,33 +3695,33 @@
},
{
"cell_type": "code",
- "execution_count": 40,
+ "execution_count": 85,
"metadata": {},
"outputs": [],
"source": [
- "model.compile(optimizer='adadelta',\n",
+ "model.compile(optimizer='sgd',\n",
" loss='categorical_crossentropy',\n",
" metrics=['accuracy'])"
]
},
{
"cell_type": "code",
- "execution_count": 41,
+ "execution_count": 86,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
- "Model: \"sequential_3\"\n",
+ "Model: \"sequential_4\"\n",
"_________________________________________________________________\n",
" Layer (type) Output Shape Param # \n",
"=================================================================\n",
- " dense_6 (Dense) (None, 500) 392500 \n",
+ " dense_12 (Dense) (None, 500) 392500 \n",
" \n",
- " dense_7 (Dense) (None, 50) 25050 \n",
+ " dense_13 (Dense) (None, 50) 25050 \n",
" \n",
- " dense_8 (Dense) (None, 10) 510 \n",
+ " dense_14 (Dense) (None, 10) 510 \n",
" \n",
"=================================================================\n",
"Total params: 418,060\n",
@@ -3615,724 +3746,200 @@
"cell_type": "markdown",
"metadata": {},
"source": [
- "Training is performed by calling the `model.fit` method:\n"
+ "We’re now ready to train the model, which in Keras is done via a call to the model’s\n",
+ "`model.fit` method — we fit the model to its training data.\n"
]
},
{
"cell_type": "code",
- "execution_count": 42,
+ "execution_count": 94,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
- "(3000, 784)\n",
- "(3000, 10)\n",
- "63/63 [==============================] - 1s 3ms/step - loss: 2.5060 - accuracy: 0.1035\n",
- "Accuracy : 0.10350000113248825\n",
- "Loss : 2.5059869289398193\n"
+ "Epoch 1/5\n",
+ "469/469 [==============================] - 7s 15ms/step - loss: 0.2764 - accuracy: 0.9218\n",
+ "Epoch 2/5\n",
+ "469/469 [==============================] - 7s 15ms/step - loss: 0.2598 - accuracy: 0.9262 0s - los\n",
+ "Epoch 3/5\n",
+ "469/469 [==============================] - 6s 14ms/step - loss: 0.2455 - accuracy: 0.9304\n",
+ "Epoch 4/5\n",
+ "469/469 [==============================] - 7s 15ms/step - loss: 0.2331 - accuracy: 0.9341\n",
+ "Epoch 5/5\n",
+ "469/469 [==============================] - 7s 15ms/step - loss: 0.2216 - accuracy: 0.9379\n"
]
}
],
"source": [
- "from tensorflow.keras import utils\n",
- "y_cat = utils.to_categorical(y_train)\n",
- "print(X_train.shape)\n",
- "print(y_cat.shape)\n",
+ "history = model.fit(X_train, y_train_cat, epochs=5, batch_size=128)"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "Two quantities are displayed during training: the __loss__ of the model over the training\n",
+ "data, and the __accuracy__ of the model over the training data. We quickly reach an accuracy of $0.92$ on the training data."
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "## 4. Evaluate Network\n",
"\n",
- "# Model evaluation of untrained model\n",
- "loss, accuracy = model.evaluate(X_train[0:2000], y_cat[0:2000])\n",
- "print('Accuracy :', accuracy)\n",
- "print('Loss :', loss)\n",
- "\n"
+ "On average, how good is our model at classifying never-before-seen digits? Let’s\n",
+ "check by computing average accuracy over the entire test set."
]
},
{
"cell_type": "code",
- "execution_count": 43,
+ "execution_count": 101,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
- "Epoch 1/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.5072 - accuracy: 0.1054 - val_loss: 2.4629 - val_accuracy: 0.1000\n",
- "Epoch 2/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.4977 - accuracy: 0.1050 - val_loss: 2.4551 - val_accuracy: 0.1033\n",
- "Epoch 3/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.4888 - accuracy: 0.1063 - val_loss: 2.4477 - val_accuracy: 0.1033\n",
- "Epoch 4/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.4803 - accuracy: 0.1079 - val_loss: 2.4405 - val_accuracy: 0.1050\n",
- "Epoch 5/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.4720 - accuracy: 0.1100 - val_loss: 2.4335 - val_accuracy: 0.1083\n",
- "Epoch 6/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.4639 - accuracy: 0.1100 - val_loss: 2.4266 - val_accuracy: 0.1100\n",
- "Epoch 7/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 2.4561 - accuracy: 0.1108 - val_loss: 2.4200 - val_accuracy: 0.1100\n",
- "Epoch 8/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.4484 - accuracy: 0.1112 - val_loss: 2.4135 - val_accuracy: 0.1133\n",
- "Epoch 9/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.4408 - accuracy: 0.1129 - val_loss: 2.4071 - val_accuracy: 0.1133\n",
- "Epoch 10/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.4334 - accuracy: 0.1146 - val_loss: 2.4008 - val_accuracy: 0.1167\n",
- "Epoch 11/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.4262 - accuracy: 0.1150 - val_loss: 2.3946 - val_accuracy: 0.1183\n",
- "Epoch 12/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.4191 - accuracy: 0.1167 - val_loss: 2.3886 - val_accuracy: 0.1183\n",
- "Epoch 13/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.4122 - accuracy: 0.1179 - val_loss: 2.3828 - val_accuracy: 0.1167\n",
- "Epoch 14/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.4054 - accuracy: 0.1183 - val_loss: 2.3771 - val_accuracy: 0.1183\n",
- "Epoch 15/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.3988 - accuracy: 0.1196 - val_loss: 2.3714 - val_accuracy: 0.1183\n",
- "Epoch 16/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.3923 - accuracy: 0.1200 - val_loss: 2.3659 - val_accuracy: 0.1233\n",
- "Epoch 17/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.3859 - accuracy: 0.1208 - val_loss: 2.3605 - val_accuracy: 0.1267\n",
- "Epoch 18/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.3797 - accuracy: 0.1225 - val_loss: 2.3552 - val_accuracy: 0.1267\n",
- "Epoch 19/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.3735 - accuracy: 0.1238 - val_loss: 2.3499 - val_accuracy: 0.1283\n",
- "Epoch 20/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 2.3674 - accuracy: 0.1254 - val_loss: 2.3448 - val_accuracy: 0.1300\n",
- "Epoch 21/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 2.3615 - accuracy: 0.1271 - val_loss: 2.3397 - val_accuracy: 0.1300\n",
- "Epoch 22/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.3556 - accuracy: 0.1283 - val_loss: 2.3348 - val_accuracy: 0.1300\n",
- "Epoch 23/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.3498 - accuracy: 0.1304 - val_loss: 2.3299 - val_accuracy: 0.1333\n",
- "Epoch 24/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 2.3442 - accuracy: 0.1308 - val_loss: 2.3252 - val_accuracy: 0.1333\n",
- "Epoch 25/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.3386 - accuracy: 0.1308 - val_loss: 2.3205 - val_accuracy: 0.1333\n",
- "Epoch 26/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.3331 - accuracy: 0.1325 - val_loss: 2.3160 - val_accuracy: 0.1367\n",
- "Epoch 27/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 2.3278 - accuracy: 0.1338 - val_loss: 2.3115 - val_accuracy: 0.1367\n",
- "Epoch 28/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 2.3225 - accuracy: 0.1342 - val_loss: 2.3071 - val_accuracy: 0.1400\n",
- "Epoch 29/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.3173 - accuracy: 0.1350 - val_loss: 2.3028 - val_accuracy: 0.1400\n",
- "Epoch 30/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.3122 - accuracy: 0.1379 - val_loss: 2.2985 - val_accuracy: 0.1433\n",
- "Epoch 31/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.3072 - accuracy: 0.1388 - val_loss: 2.2943 - val_accuracy: 0.1417\n",
- "Epoch 32/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.3023 - accuracy: 0.1412 - val_loss: 2.2902 - val_accuracy: 0.1417\n",
- "Epoch 33/300\n",
- "75/75 [==============================] - 1s 10ms/step - loss: 2.2974 - accuracy: 0.1429 - val_loss: 2.2862 - val_accuracy: 0.1400\n",
- "Epoch 34/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2926 - accuracy: 0.1446 - val_loss: 2.2822 - val_accuracy: 0.1417\n",
- "Epoch 35/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 2.2879 - accuracy: 0.1471 - val_loss: 2.2783 - val_accuracy: 0.1383\n",
- "Epoch 36/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 2.2833 - accuracy: 0.1488 - val_loss: 2.2745 - val_accuracy: 0.1400\n",
- "Epoch 37/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2787 - accuracy: 0.1521 - val_loss: 2.2707 - val_accuracy: 0.1400\n",
- "Epoch 38/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2742 - accuracy: 0.1554 - val_loss: 2.2670 - val_accuracy: 0.1400\n",
- "Epoch 39/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2698 - accuracy: 0.1554 - val_loss: 2.2633 - val_accuracy: 0.1400\n",
- "Epoch 40/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 2.2654 - accuracy: 0.1562 - val_loss: 2.2597 - val_accuracy: 0.1417\n",
- "Epoch 41/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2611 - accuracy: 0.1579 - val_loss: 2.2562 - val_accuracy: 0.1450\n",
- "Epoch 42/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2569 - accuracy: 0.1587 - val_loss: 2.2527 - val_accuracy: 0.1450\n",
- "Epoch 43/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2527 - accuracy: 0.1612 - val_loss: 2.2492 - val_accuracy: 0.1450\n",
- "Epoch 44/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2486 - accuracy: 0.1637 - val_loss: 2.2458 - val_accuracy: 0.1483\n",
- "Epoch 45/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2445 - accuracy: 0.1663 - val_loss: 2.2424 - val_accuracy: 0.1500\n",
- "Epoch 46/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 2.2404 - accuracy: 0.1671 - val_loss: 2.2391 - val_accuracy: 0.1550\n",
- "Epoch 47/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2364 - accuracy: 0.1696 - val_loss: 2.2358 - val_accuracy: 0.1567\n",
- "Epoch 48/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2325 - accuracy: 0.1733 - val_loss: 2.2326 - val_accuracy: 0.1600\n",
- "Epoch 49/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2286 - accuracy: 0.1750 - val_loss: 2.2294 - val_accuracy: 0.1600\n",
- "Epoch 50/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2247 - accuracy: 0.1758 - val_loss: 2.2263 - val_accuracy: 0.1600\n",
- "Epoch 51/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.2209 - accuracy: 0.1783 - val_loss: 2.2232 - val_accuracy: 0.1600\n",
- "Epoch 52/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.2172 - accuracy: 0.1804 - val_loss: 2.2201 - val_accuracy: 0.1600\n",
- "Epoch 53/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.2134 - accuracy: 0.1821 - val_loss: 2.2171 - val_accuracy: 0.1600\n",
- "Epoch 54/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 2.2097 - accuracy: 0.1846 - val_loss: 2.2141 - val_accuracy: 0.1600\n",
- "Epoch 55/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2061 - accuracy: 0.1875 - val_loss: 2.2112 - val_accuracy: 0.1617\n",
- "Epoch 56/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.2025 - accuracy: 0.1908 - val_loss: 2.2083 - val_accuracy: 0.1617\n",
- "Epoch 57/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1989 - accuracy: 0.1958 - val_loss: 2.2054 - val_accuracy: 0.1633\n",
- "Epoch 58/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1954 - accuracy: 0.1992 - val_loss: 2.2025 - val_accuracy: 0.1633\n",
- "Epoch 59/300\n",
- "75/75 [==============================] - 1s 10ms/step - loss: 2.1918 - accuracy: 0.2013 - val_loss: 2.1997 - val_accuracy: 0.1650\n",
- "Epoch 60/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1883 - accuracy: 0.2050 - val_loss: 2.1969 - val_accuracy: 0.1650\n",
- "Epoch 61/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1848 - accuracy: 0.2067 - val_loss: 2.1941 - val_accuracy: 0.1683\n",
- "Epoch 62/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 2.1814 - accuracy: 0.2083 - val_loss: 2.1914 - val_accuracy: 0.1750\n",
- "Epoch 63/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1780 - accuracy: 0.2104 - val_loss: 2.1887 - val_accuracy: 0.1767\n",
- "Epoch 64/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 2.1747 - accuracy: 0.2129 - val_loss: 2.1860 - val_accuracy: 0.1783\n",
- "Epoch 65/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 2.1713 - accuracy: 0.2154 - val_loss: 2.1833 - val_accuracy: 0.1800\n",
- "Epoch 66/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.1680 - accuracy: 0.2179 - val_loss: 2.1807 - val_accuracy: 0.1817\n",
- "Epoch 67/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1647 - accuracy: 0.2254 - val_loss: 2.1781 - val_accuracy: 0.1833\n",
- "Epoch 68/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1615 - accuracy: 0.2296 - val_loss: 2.1755 - val_accuracy: 0.1883\n",
- "Epoch 69/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1583 - accuracy: 0.2304 - val_loss: 2.1729 - val_accuracy: 0.1900\n",
- "Epoch 70/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.1551 - accuracy: 0.2346 - val_loss: 2.1704 - val_accuracy: 0.1883\n",
- "Epoch 71/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 2.1519 - accuracy: 0.2404 - val_loss: 2.1678 - val_accuracy: 0.1933\n",
- "Epoch 72/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1488 - accuracy: 0.2454 - val_loss: 2.1653 - val_accuracy: 0.2017\n",
- "Epoch 73/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 2.1456 - accuracy: 0.2492 - val_loss: 2.1628 - val_accuracy: 0.2017\n",
- "Epoch 74/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 2.1425 - accuracy: 0.2562 - val_loss: 2.1603 - val_accuracy: 0.2050\n",
- "Epoch 75/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1395 - accuracy: 0.2604 - val_loss: 2.1578 - val_accuracy: 0.2117\n",
- "Epoch 76/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 2.1364 - accuracy: 0.2667 - val_loss: 2.1553 - val_accuracy: 0.2150\n",
- "Epoch 77/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1334 - accuracy: 0.2708 - val_loss: 2.1529 - val_accuracy: 0.2217\n",
- "Epoch 78/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1304 - accuracy: 0.2779 - val_loss: 2.1504 - val_accuracy: 0.2250\n",
- "Epoch 79/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 2.1274 - accuracy: 0.2854 - val_loss: 2.1480 - val_accuracy: 0.2317\n",
- "Epoch 80/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 2.1244 - accuracy: 0.2900 - val_loss: 2.1456 - val_accuracy: 0.2367\n",
- "Epoch 81/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1215 - accuracy: 0.2950 - val_loss: 2.1432 - val_accuracy: 0.2433\n",
- "Epoch 82/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.1186 - accuracy: 0.2971 - val_loss: 2.1408 - val_accuracy: 0.2467\n",
- "Epoch 83/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1157 - accuracy: 0.3033 - val_loss: 2.1384 - val_accuracy: 0.2450\n",
- "Epoch 84/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1128 - accuracy: 0.3129 - val_loss: 2.1361 - val_accuracy: 0.2533\n",
- "Epoch 85/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 2.1099 - accuracy: 0.3171 - val_loss: 2.1338 - val_accuracy: 0.2583\n",
- "Epoch 86/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.1071 - accuracy: 0.3225 - val_loss: 2.1314 - val_accuracy: 0.2617\n",
- "Epoch 87/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1042 - accuracy: 0.3254 - val_loss: 2.1291 - val_accuracy: 0.2683\n",
- "Epoch 88/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.1014 - accuracy: 0.3300 - val_loss: 2.1268 - val_accuracy: 0.2750\n",
- "Epoch 89/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 2.0986 - accuracy: 0.3338 - val_loss: 2.1245 - val_accuracy: 0.2850\n",
- "Epoch 90/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0958 - accuracy: 0.3408 - val_loss: 2.1222 - val_accuracy: 0.2850\n",
- "Epoch 91/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.0930 - accuracy: 0.3471 - val_loss: 2.1199 - val_accuracy: 0.2950\n",
- "Epoch 92/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.0902 - accuracy: 0.3504 - val_loss: 2.1176 - val_accuracy: 0.2983\n",
- "Epoch 93/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 2.0875 - accuracy: 0.3562 - val_loss: 2.1154 - val_accuracy: 0.3083\n",
- "Epoch 94/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0847 - accuracy: 0.3617 - val_loss: 2.1131 - val_accuracy: 0.3083\n",
- "Epoch 95/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0820 - accuracy: 0.3654 - val_loss: 2.1108 - val_accuracy: 0.3200\n",
- "Epoch 96/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0792 - accuracy: 0.3692 - val_loss: 2.1086 - val_accuracy: 0.3233\n",
- "Epoch 97/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 2.0765 - accuracy: 0.3742 - val_loss: 2.1064 - val_accuracy: 0.3283\n",
- "Epoch 98/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.0738 - accuracy: 0.3783 - val_loss: 2.1041 - val_accuracy: 0.3300\n",
- "Epoch 99/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0711 - accuracy: 0.3833 - val_loss: 2.1019 - val_accuracy: 0.3333\n",
- "Epoch 100/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 2.0684 - accuracy: 0.3887 - val_loss: 2.0998 - val_accuracy: 0.3433\n",
- "Epoch 101/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0657 - accuracy: 0.3946 - val_loss: 2.0976 - val_accuracy: 0.3483\n",
- "Epoch 102/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0630 - accuracy: 0.4013 - val_loss: 2.0954 - val_accuracy: 0.3517\n",
- "Epoch 103/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.0604 - accuracy: 0.4054 - val_loss: 2.0932 - val_accuracy: 0.3567\n",
- "Epoch 104/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0577 - accuracy: 0.4133 - val_loss: 2.0910 - val_accuracy: 0.3633\n",
- "Epoch 105/300\n",
- "75/75 [==============================] - 1s 10ms/step - loss: 2.0551 - accuracy: 0.4179 - val_loss: 2.0888 - val_accuracy: 0.3617\n",
- "Epoch 106/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0524 - accuracy: 0.4212 - val_loss: 2.0866 - val_accuracy: 0.3667\n",
- "Epoch 107/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 2.0497 - accuracy: 0.4283 - val_loss: 2.0844 - val_accuracy: 0.3700\n",
- "Epoch 108/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 2.0471 - accuracy: 0.4346 - val_loss: 2.0822 - val_accuracy: 0.3717\n",
- "Epoch 109/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0445 - accuracy: 0.4379 - val_loss: 2.0801 - val_accuracy: 0.3767\n",
- "Epoch 110/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0418 - accuracy: 0.4446 - val_loss: 2.0779 - val_accuracy: 0.3817\n",
- "Epoch 111/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0392 - accuracy: 0.4504 - val_loss: 2.0757 - val_accuracy: 0.3833\n",
- "Epoch 112/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.0366 - accuracy: 0.4592 - val_loss: 2.0735 - val_accuracy: 0.3933\n",
- "Epoch 113/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0340 - accuracy: 0.4663 - val_loss: 2.0714 - val_accuracy: 0.3983\n",
- "Epoch 114/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0314 - accuracy: 0.4717 - val_loss: 2.0692 - val_accuracy: 0.4050\n",
- "Epoch 115/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0289 - accuracy: 0.4758 - val_loss: 2.0670 - val_accuracy: 0.4050\n",
- "Epoch 116/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 2.0263 - accuracy: 0.4812 - val_loss: 2.0649 - val_accuracy: 0.4067\n",
- "Epoch 117/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0237 - accuracy: 0.4879 - val_loss: 2.0627 - val_accuracy: 0.4100\n",
- "Epoch 118/300\n",
- "75/75 [==============================] - 1s 10ms/step - loss: 2.0212 - accuracy: 0.4946 - val_loss: 2.0605 - val_accuracy: 0.4183\n",
- "Epoch 119/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.0186 - accuracy: 0.5021 - val_loss: 2.0584 - val_accuracy: 0.4217\n",
- "Epoch 120/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0161 - accuracy: 0.5050 - val_loss: 2.0562 - val_accuracy: 0.4250\n",
- "Epoch 121/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0135 - accuracy: 0.5092 - val_loss: 2.0541 - val_accuracy: 0.4283\n",
- "Epoch 122/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0110 - accuracy: 0.5117 - val_loss: 2.0519 - val_accuracy: 0.4317\n",
- "Epoch 123/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0085 - accuracy: 0.5167 - val_loss: 2.0498 - val_accuracy: 0.4367\n",
- "Epoch 124/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 2.0060 - accuracy: 0.5179 - val_loss: 2.0477 - val_accuracy: 0.4383\n",
- "Epoch 125/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 2.0035 - accuracy: 0.5225 - val_loss: 2.0456 - val_accuracy: 0.4433\n",
- "Epoch 126/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 2.0010 - accuracy: 0.5254 - val_loss: 2.0434 - val_accuracy: 0.4467\n",
- "Epoch 127/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.9985 - accuracy: 0.5279 - val_loss: 2.0413 - val_accuracy: 0.4533\n",
- "Epoch 128/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9960 - accuracy: 0.5312 - val_loss: 2.0392 - val_accuracy: 0.4583\n",
- "Epoch 129/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.9935 - accuracy: 0.5375 - val_loss: 2.0371 - val_accuracy: 0.4617\n",
- "Epoch 130/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9911 - accuracy: 0.5400 - val_loss: 2.0351 - val_accuracy: 0.4617\n",
- "Epoch 131/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.9886 - accuracy: 0.5450 - val_loss: 2.0330 - val_accuracy: 0.4617\n",
- "Epoch 132/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.9862 - accuracy: 0.5483 - val_loss: 2.0309 - val_accuracy: 0.4650\n",
- "Epoch 133/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9837 - accuracy: 0.5487 - val_loss: 2.0288 - val_accuracy: 0.4683\n",
- "Epoch 134/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.9813 - accuracy: 0.5558 - val_loss: 2.0268 - val_accuracy: 0.4750\n",
- "Epoch 135/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.9789 - accuracy: 0.5596 - val_loss: 2.0247 - val_accuracy: 0.4783\n",
- "Epoch 136/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9765 - accuracy: 0.5629 - val_loss: 2.0227 - val_accuracy: 0.4800\n",
- "Epoch 137/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9740 - accuracy: 0.5679 - val_loss: 2.0207 - val_accuracy: 0.4833\n",
- "Epoch 138/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.9716 - accuracy: 0.5700 - val_loss: 2.0186 - val_accuracy: 0.4850\n",
- "Epoch 139/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9692 - accuracy: 0.5742 - val_loss: 2.0166 - val_accuracy: 0.4867\n",
- "Epoch 140/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9668 - accuracy: 0.5775 - val_loss: 2.0146 - val_accuracy: 0.4850\n",
- "Epoch 141/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9644 - accuracy: 0.5796 - val_loss: 2.0126 - val_accuracy: 0.4850\n",
- "Epoch 142/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 1.9620 - accuracy: 0.5825 - val_loss: 2.0106 - val_accuracy: 0.4883\n",
- "Epoch 143/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9596 - accuracy: 0.5854 - val_loss: 2.0086 - val_accuracy: 0.4917\n",
- "Epoch 144/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.9572 - accuracy: 0.5883 - val_loss: 2.0066 - val_accuracy: 0.4950\n",
- "Epoch 145/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.9548 - accuracy: 0.5929 - val_loss: 2.0046 - val_accuracy: 0.4967\n",
- "Epoch 146/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9524 - accuracy: 0.5979 - val_loss: 2.0026 - val_accuracy: 0.4983\n",
- "Epoch 147/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.9500 - accuracy: 0.5996 - val_loss: 2.0006 - val_accuracy: 0.4983\n",
- "Epoch 148/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.9477 - accuracy: 0.6012 - val_loss: 1.9986 - val_accuracy: 0.5000\n",
- "Epoch 149/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9453 - accuracy: 0.6037 - val_loss: 1.9966 - val_accuracy: 0.5033\n",
- "Epoch 150/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9429 - accuracy: 0.6087 - val_loss: 1.9946 - val_accuracy: 0.5067\n",
- "Epoch 151/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.9406 - accuracy: 0.6108 - val_loss: 1.9926 - val_accuracy: 0.5100\n",
- "Epoch 152/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.9382 - accuracy: 0.6117 - val_loss: 1.9906 - val_accuracy: 0.5133\n",
- "Epoch 153/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9358 - accuracy: 0.6167 - val_loss: 1.9886 - val_accuracy: 0.5150\n",
- "Epoch 154/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.9335 - accuracy: 0.6167 - val_loss: 1.9867 - val_accuracy: 0.5183\n",
- "Epoch 155/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.9311 - accuracy: 0.6196 - val_loss: 1.9847 - val_accuracy: 0.5200\n",
- "Epoch 156/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9288 - accuracy: 0.6242 - val_loss: 1.9827 - val_accuracy: 0.5217\n",
- "Epoch 157/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.9264 - accuracy: 0.6250 - val_loss: 1.9807 - val_accuracy: 0.5250\n",
- "Epoch 158/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.9241 - accuracy: 0.6275 - val_loss: 1.9787 - val_accuracy: 0.5250\n",
- "Epoch 159/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9217 - accuracy: 0.6304 - val_loss: 1.9767 - val_accuracy: 0.5300\n",
- "Epoch 160/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9194 - accuracy: 0.6338 - val_loss: 1.9748 - val_accuracy: 0.5300\n",
- "Epoch 161/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9171 - accuracy: 0.6363 - val_loss: 1.9728 - val_accuracy: 0.5317\n",
- "Epoch 162/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.9147 - accuracy: 0.6371 - val_loss: 1.9708 - val_accuracy: 0.5350\n",
- "Epoch 163/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9124 - accuracy: 0.6400 - val_loss: 1.9688 - val_accuracy: 0.5350\n",
- "Epoch 164/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.9101 - accuracy: 0.6388 - val_loss: 1.9668 - val_accuracy: 0.5350\n",
- "Epoch 165/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.9077 - accuracy: 0.6413 - val_loss: 1.9648 - val_accuracy: 0.5367\n",
- "Epoch 166/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9054 - accuracy: 0.6446 - val_loss: 1.9629 - val_accuracy: 0.5367\n",
- "Epoch 167/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.9031 - accuracy: 0.6450 - val_loss: 1.9609 - val_accuracy: 0.5367\n",
- "Epoch 168/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.9008 - accuracy: 0.6467 - val_loss: 1.9589 - val_accuracy: 0.5383\n",
- "Epoch 169/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8985 - accuracy: 0.6488 - val_loss: 1.9570 - val_accuracy: 0.5467\n",
- "Epoch 170/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8962 - accuracy: 0.6525 - val_loss: 1.9550 - val_accuracy: 0.5483\n",
- "Epoch 171/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.8939 - accuracy: 0.6538 - val_loss: 1.9531 - val_accuracy: 0.5533\n",
- "Epoch 172/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8916 - accuracy: 0.6554 - val_loss: 1.9511 - val_accuracy: 0.5567\n",
- "Epoch 173/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.8893 - accuracy: 0.6562 - val_loss: 1.9492 - val_accuracy: 0.5583\n",
- "Epoch 174/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8870 - accuracy: 0.6600 - val_loss: 1.9473 - val_accuracy: 0.5600\n",
- "Epoch 175/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8847 - accuracy: 0.6621 - val_loss: 1.9453 - val_accuracy: 0.5617\n",
- "Epoch 176/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8824 - accuracy: 0.6658 - val_loss: 1.9434 - val_accuracy: 0.5633\n",
- "Epoch 177/300\n",
- "75/75 [==============================] - 1s 10ms/step - loss: 1.8802 - accuracy: 0.6679 - val_loss: 1.9415 - val_accuracy: 0.5667\n",
- "Epoch 178/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.8779 - accuracy: 0.6692 - val_loss: 1.9396 - val_accuracy: 0.5683\n",
- "Epoch 179/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8756 - accuracy: 0.6733 - val_loss: 1.9377 - val_accuracy: 0.5700\n",
- "Epoch 180/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 1.8734 - accuracy: 0.6750 - val_loss: 1.9358 - val_accuracy: 0.5700\n",
- "Epoch 181/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8711 - accuracy: 0.6775 - val_loss: 1.9339 - val_accuracy: 0.5733\n",
- "Epoch 182/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.8688 - accuracy: 0.6792 - val_loss: 1.9320 - val_accuracy: 0.5733\n",
- "Epoch 183/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.8666 - accuracy: 0.6812 - val_loss: 1.9301 - val_accuracy: 0.5750\n",
- "Epoch 184/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.8643 - accuracy: 0.6817 - val_loss: 1.9282 - val_accuracy: 0.5750\n",
- "Epoch 185/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8620 - accuracy: 0.6825 - val_loss: 1.9263 - val_accuracy: 0.5750\n",
- "Epoch 186/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8598 - accuracy: 0.6833 - val_loss: 1.9245 - val_accuracy: 0.5733\n",
- "Epoch 187/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8575 - accuracy: 0.6850 - val_loss: 1.9226 - val_accuracy: 0.5733\n",
- "Epoch 188/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 1.8553 - accuracy: 0.6867 - val_loss: 1.9207 - val_accuracy: 0.5750\n",
- "Epoch 189/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8531 - accuracy: 0.6858 - val_loss: 1.9188 - val_accuracy: 0.5767\n",
- "Epoch 190/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.8509 - accuracy: 0.6896 - val_loss: 1.9170 - val_accuracy: 0.5783\n",
- "Epoch 191/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8486 - accuracy: 0.6917 - val_loss: 1.9151 - val_accuracy: 0.5783\n",
- "Epoch 192/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8464 - accuracy: 0.6938 - val_loss: 1.9132 - val_accuracy: 0.5783\n",
- "Epoch 193/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8442 - accuracy: 0.6958 - val_loss: 1.9114 - val_accuracy: 0.5800\n",
- "Epoch 194/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.8420 - accuracy: 0.6971 - val_loss: 1.9095 - val_accuracy: 0.5833\n",
- "Epoch 195/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8398 - accuracy: 0.6988 - val_loss: 1.9077 - val_accuracy: 0.5850\n",
- "Epoch 196/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.8376 - accuracy: 0.6996 - val_loss: 1.9058 - val_accuracy: 0.5867\n",
- "Epoch 197/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8354 - accuracy: 0.7021 - val_loss: 1.9040 - val_accuracy: 0.5867\n",
- "Epoch 198/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.8332 - accuracy: 0.7025 - val_loss: 1.9021 - val_accuracy: 0.5883\n",
- "Epoch 199/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.8310 - accuracy: 0.7042 - val_loss: 1.9003 - val_accuracy: 0.5883\n",
- "Epoch 200/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8288 - accuracy: 0.7042 - val_loss: 1.8984 - val_accuracy: 0.5883\n",
- "Epoch 201/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8266 - accuracy: 0.7063 - val_loss: 1.8966 - val_accuracy: 0.5900\n",
- "Epoch 202/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.8244 - accuracy: 0.7079 - val_loss: 1.8947 - val_accuracy: 0.5917\n",
- "Epoch 203/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.8222 - accuracy: 0.7075 - val_loss: 1.8929 - val_accuracy: 0.5917\n",
- "Epoch 204/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 1.8200 - accuracy: 0.7088 - val_loss: 1.8910 - val_accuracy: 0.5917\n",
- "Epoch 205/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8179 - accuracy: 0.7100 - val_loss: 1.8892 - val_accuracy: 0.5933\n",
- "Epoch 206/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8157 - accuracy: 0.7100 - val_loss: 1.8873 - val_accuracy: 0.5933\n",
- "Epoch 207/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8135 - accuracy: 0.7117 - val_loss: 1.8855 - val_accuracy: 0.5933\n",
- "Epoch 208/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8113 - accuracy: 0.7117 - val_loss: 1.8836 - val_accuracy: 0.5967\n",
- "Epoch 209/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.8092 - accuracy: 0.7129 - val_loss: 1.8818 - val_accuracy: 0.5967\n",
- "Epoch 210/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.8070 - accuracy: 0.7146 - val_loss: 1.8800 - val_accuracy: 0.5967\n",
- "Epoch 211/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.8048 - accuracy: 0.7154 - val_loss: 1.8781 - val_accuracy: 0.5967\n",
- "Epoch 212/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.8027 - accuracy: 0.7175 - val_loss: 1.8763 - val_accuracy: 0.6017\n",
- "Epoch 213/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.8006 - accuracy: 0.7171 - val_loss: 1.8745 - val_accuracy: 0.6017\n",
- "Epoch 214/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.7984 - accuracy: 0.7183 - val_loss: 1.8727 - val_accuracy: 0.6033\n",
- "Epoch 215/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.7963 - accuracy: 0.7188 - val_loss: 1.8708 - val_accuracy: 0.6017\n",
- "Epoch 216/300\n",
- "75/75 [==============================] - 1s 10ms/step - loss: 1.7942 - accuracy: 0.7200 - val_loss: 1.8690 - val_accuracy: 0.6017\n",
- "Epoch 217/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7920 - accuracy: 0.7204 - val_loss: 1.8672 - val_accuracy: 0.6050\n",
- "Epoch 218/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7899 - accuracy: 0.7208 - val_loss: 1.8654 - val_accuracy: 0.6050\n",
- "Epoch 219/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.7878 - accuracy: 0.7212 - val_loss: 1.8636 - val_accuracy: 0.6083\n",
- "Epoch 220/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7857 - accuracy: 0.7229 - val_loss: 1.8618 - val_accuracy: 0.6100\n",
- "Epoch 221/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.7835 - accuracy: 0.7246 - val_loss: 1.8600 - val_accuracy: 0.6100\n",
- "Epoch 222/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.7814 - accuracy: 0.7254 - val_loss: 1.8581 - val_accuracy: 0.6100\n",
- "Epoch 223/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.7793 - accuracy: 0.7267 - val_loss: 1.8563 - val_accuracy: 0.6117\n",
- "Epoch 224/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.7772 - accuracy: 0.7271 - val_loss: 1.8545 - val_accuracy: 0.6150\n",
- "Epoch 225/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7751 - accuracy: 0.7287 - val_loss: 1.8527 - val_accuracy: 0.6167\n",
- "Epoch 226/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.7730 - accuracy: 0.7300 - val_loss: 1.8509 - val_accuracy: 0.6183\n",
- "Epoch 227/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7709 - accuracy: 0.7308 - val_loss: 1.8491 - val_accuracy: 0.6200\n",
- "Epoch 228/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7688 - accuracy: 0.7308 - val_loss: 1.8473 - val_accuracy: 0.6200\n",
- "Epoch 229/300\n",
- "75/75 [==============================] - 1s 10ms/step - loss: 1.7667 - accuracy: 0.7300 - val_loss: 1.8455 - val_accuracy: 0.6200\n",
- "Epoch 230/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7646 - accuracy: 0.7308 - val_loss: 1.8437 - val_accuracy: 0.6200\n",
- "Epoch 231/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7625 - accuracy: 0.7308 - val_loss: 1.8419 - val_accuracy: 0.6200\n",
- "Epoch 232/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7604 - accuracy: 0.7321 - val_loss: 1.8401 - val_accuracy: 0.6233\n",
- "Epoch 233/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 1.7583 - accuracy: 0.7325 - val_loss: 1.8383 - val_accuracy: 0.6250\n",
- "Epoch 234/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7562 - accuracy: 0.7329 - val_loss: 1.8365 - val_accuracy: 0.6250\n",
- "Epoch 235/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.7542 - accuracy: 0.7321 - val_loss: 1.8347 - val_accuracy: 0.6250\n",
- "Epoch 236/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.7521 - accuracy: 0.7337 - val_loss: 1.8329 - val_accuracy: 0.6250\n",
- "Epoch 237/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.7500 - accuracy: 0.7342 - val_loss: 1.8311 - val_accuracy: 0.6250\n",
- "Epoch 238/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7479 - accuracy: 0.7346 - val_loss: 1.8294 - val_accuracy: 0.6267\n",
- "Epoch 239/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.7459 - accuracy: 0.7346 - val_loss: 1.8276 - val_accuracy: 0.6267\n",
- "Epoch 240/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7438 - accuracy: 0.7342 - val_loss: 1.8259 - val_accuracy: 0.6267\n",
- "Epoch 241/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7418 - accuracy: 0.7346 - val_loss: 1.8241 - val_accuracy: 0.6250\n",
- "Epoch 242/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.7397 - accuracy: 0.7346 - val_loss: 1.8224 - val_accuracy: 0.6250\n",
- "Epoch 243/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7376 - accuracy: 0.7354 - val_loss: 1.8206 - val_accuracy: 0.6233\n",
- "Epoch 244/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7356 - accuracy: 0.7362 - val_loss: 1.8188 - val_accuracy: 0.6250\n",
- "Epoch 245/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.7335 - accuracy: 0.7354 - val_loss: 1.8171 - val_accuracy: 0.6267\n",
- "Epoch 246/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7315 - accuracy: 0.7354 - val_loss: 1.8153 - val_accuracy: 0.6283\n",
- "Epoch 247/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7294 - accuracy: 0.7371 - val_loss: 1.8136 - val_accuracy: 0.6283\n",
- "Epoch 248/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.7274 - accuracy: 0.7367 - val_loss: 1.8118 - val_accuracy: 0.6283\n",
- "Epoch 249/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.7254 - accuracy: 0.7383 - val_loss: 1.8101 - val_accuracy: 0.6283\n",
- "Epoch 250/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7233 - accuracy: 0.7392 - val_loss: 1.8083 - val_accuracy: 0.6300\n",
- "Epoch 251/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7213 - accuracy: 0.7396 - val_loss: 1.8066 - val_accuracy: 0.6300\n",
- "Epoch 252/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7192 - accuracy: 0.7404 - val_loss: 1.8048 - val_accuracy: 0.6317\n",
- "Epoch 253/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7172 - accuracy: 0.7408 - val_loss: 1.8031 - val_accuracy: 0.6317\n",
- "Epoch 254/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.7152 - accuracy: 0.7425 - val_loss: 1.8013 - val_accuracy: 0.6317\n",
- "Epoch 255/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.7131 - accuracy: 0.7425 - val_loss: 1.7995 - val_accuracy: 0.6333\n",
- "Epoch 256/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.7111 - accuracy: 0.7421 - val_loss: 1.7978 - val_accuracy: 0.6317\n",
- "Epoch 257/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7091 - accuracy: 0.7421 - val_loss: 1.7961 - val_accuracy: 0.6317\n",
- "Epoch 258/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7071 - accuracy: 0.7433 - val_loss: 1.7943 - val_accuracy: 0.6333\n",
- "Epoch 259/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7051 - accuracy: 0.7425 - val_loss: 1.7926 - val_accuracy: 0.6333\n",
- "Epoch 260/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.7031 - accuracy: 0.7433 - val_loss: 1.7909 - val_accuracy: 0.6333\n",
- "Epoch 261/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.7011 - accuracy: 0.7442 - val_loss: 1.7891 - val_accuracy: 0.6333\n",
- "Epoch 262/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.6991 - accuracy: 0.7450 - val_loss: 1.7874 - val_accuracy: 0.6350\n",
- "Epoch 263/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6971 - accuracy: 0.7442 - val_loss: 1.7857 - val_accuracy: 0.6350\n",
- "Epoch 264/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.6952 - accuracy: 0.7462 - val_loss: 1.7839 - val_accuracy: 0.6367\n",
- "Epoch 265/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.6932 - accuracy: 0.7462 - val_loss: 1.7822 - val_accuracy: 0.6367\n",
- "Epoch 266/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6912 - accuracy: 0.7471 - val_loss: 1.7805 - val_accuracy: 0.6367\n",
- "Epoch 267/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.6892 - accuracy: 0.7475 - val_loss: 1.7788 - val_accuracy: 0.6367\n",
- "Epoch 268/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.6872 - accuracy: 0.7475 - val_loss: 1.7770 - val_accuracy: 0.6367\n",
- "Epoch 269/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6853 - accuracy: 0.7471 - val_loss: 1.7753 - val_accuracy: 0.6367\n",
- "Epoch 270/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6833 - accuracy: 0.7496 - val_loss: 1.7736 - val_accuracy: 0.6367\n",
- "Epoch 271/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6813 - accuracy: 0.7504 - val_loss: 1.7718 - val_accuracy: 0.6350\n",
- "Epoch 272/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 1.6794 - accuracy: 0.7521 - val_loss: 1.7701 - val_accuracy: 0.6367\n",
- "Epoch 273/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6774 - accuracy: 0.7529 - val_loss: 1.7683 - val_accuracy: 0.6367\n",
- "Epoch 274/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.6754 - accuracy: 0.7529 - val_loss: 1.7666 - val_accuracy: 0.6367\n",
- "Epoch 275/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6735 - accuracy: 0.7542 - val_loss: 1.7649 - val_accuracy: 0.6367\n",
- "Epoch 276/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6715 - accuracy: 0.7546 - val_loss: 1.7632 - val_accuracy: 0.6350\n",
- "Epoch 277/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6696 - accuracy: 0.7571 - val_loss: 1.7615 - val_accuracy: 0.6350\n",
- "Epoch 278/300\n",
- "75/75 [==============================] - 0s 7ms/step - loss: 1.6677 - accuracy: 0.7583 - val_loss: 1.7598 - val_accuracy: 0.6367\n",
- "Epoch 279/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6657 - accuracy: 0.7579 - val_loss: 1.7581 - val_accuracy: 0.6400\n",
- "Epoch 280/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.6638 - accuracy: 0.7588 - val_loss: 1.7564 - val_accuracy: 0.6417\n",
- "Epoch 281/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.6619 - accuracy: 0.7583 - val_loss: 1.7547 - val_accuracy: 0.6433\n",
- "Epoch 282/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6600 - accuracy: 0.7579 - val_loss: 1.7530 - val_accuracy: 0.6433\n",
- "Epoch 283/300\n",
- "75/75 [==============================] - 0s 6ms/step - loss: 1.6581 - accuracy: 0.7583 - val_loss: 1.7513 - val_accuracy: 0.6433\n",
- "Epoch 284/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6562 - accuracy: 0.7588 - val_loss: 1.7496 - val_accuracy: 0.6450\n",
- "Epoch 285/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6542 - accuracy: 0.7596 - val_loss: 1.7479 - val_accuracy: 0.6450\n",
- "Epoch 286/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6523 - accuracy: 0.7592 - val_loss: 1.7462 - val_accuracy: 0.6450\n",
- "Epoch 287/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.6504 - accuracy: 0.7596 - val_loss: 1.7446 - val_accuracy: 0.6450\n",
- "Epoch 288/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.6485 - accuracy: 0.7608 - val_loss: 1.7429 - val_accuracy: 0.6450\n",
- "Epoch 289/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6467 - accuracy: 0.7625 - val_loss: 1.7412 - val_accuracy: 0.6450\n",
- "Epoch 290/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6448 - accuracy: 0.7625 - val_loss: 1.7396 - val_accuracy: 0.6450\n",
- "Epoch 291/300\n",
- "75/75 [==============================] - 1s 11ms/step - loss: 1.6429 - accuracy: 0.7633 - val_loss: 1.7379 - val_accuracy: 0.6467\n",
- "Epoch 292/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.6410 - accuracy: 0.7638 - val_loss: 1.7362 - val_accuracy: 0.6467\n",
- "Epoch 293/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.6391 - accuracy: 0.7650 - val_loss: 1.7346 - val_accuracy: 0.6450\n",
- "Epoch 294/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.6373 - accuracy: 0.7646 - val_loss: 1.7329 - val_accuracy: 0.6433\n",
- "Epoch 295/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6354 - accuracy: 0.7658 - val_loss: 1.7313 - val_accuracy: 0.6433\n",
- "Epoch 296/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.6335 - accuracy: 0.7654 - val_loss: 1.7296 - val_accuracy: 0.6433\n",
- "Epoch 297/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.6317 - accuracy: 0.7658 - val_loss: 1.7280 - val_accuracy: 0.6433\n",
- "Epoch 298/300\n",
- "75/75 [==============================] - 1s 7ms/step - loss: 1.6298 - accuracy: 0.7671 - val_loss: 1.7264 - val_accuracy: 0.6433\n",
- "Epoch 299/300\n",
- "75/75 [==============================] - 1s 9ms/step - loss: 1.6280 - accuracy: 0.7675 - val_loss: 1.7247 - val_accuracy: 0.6433\n",
- "Epoch 300/300\n",
- "75/75 [==============================] - 1s 8ms/step - loss: 1.6261 - accuracy: 0.7675 - val_loss: 1.7231 - val_accuracy: 0.6433\n"
+ "313/313 [==============================] - 1s 4ms/step - loss: 0.2117 - accuracy: 0.9394\n",
+ "Accuracy : 0.9394000172615051\n"
]
}
],
"source": [
- "history = model.fit(X_train[0:2400], y_cat[0:2400], epochs=300, batch_size=32, validation_data=[X_train[2400:3000], y_cat[2400:3000]])"
+ "loss, accuracy = model.evaluate(X_test, y_test_cat)\n",
+ "print('Accuracy :', accuracy)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
- "## 4. Evaluate Network"
+ "## 5. Make Predictions\n",
+ "\n",
+ "Now that we have a trained model, we can use it to predict class probabilities for\n",
+ "new digits—images that weren’t part of the training data, like those from the test set."
]
},
{
"cell_type": "code",
- "execution_count": 46,
+ "execution_count": 91,
"metadata": {},
"outputs": [
{
- "name": "stdout",
- "output_type": "stream",
- "text": [
- "19/19 [==============================] - 0s 4ms/step - loss: 1.7231 - accuracy: 0.6433\n",
- "Accuracy : 0.6433333158493042\n"
- ]
+ "data": {
+ "text/plain": [
+ "array([7, 2, 1, 0, 4, 1, 4, 9, 6, 9])"
+ ]
+ },
+ "execution_count": 91,
+ "metadata": {},
+ "output_type": "execute_result"
}
],
"source": [
- "loss, accuracy = model.evaluate(X_train[2400:3000], y_cat[2400:3000])\n",
- "print('Accuracy :', accuracy)"
+ "predictions=model.predict(X_test[0:10])\n",
+ "np.argmax(predictions, axis=1)"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 92,
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "array([7, 2, 1, 0, 4, 1, 4, 9, 5, 9], dtype=uint8)"
+ ]
+ },
+ "execution_count": 92,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "y_test[0:10]"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
- "## 5. Make Predictions"
+ "Each number of index $i$ in that array corresponds to the probability that digit image\n",
+ "`X_test[0]` belongs to class $i$. This first test digit has the highest probability score (0.9956499, almost 1) at\n",
+ "index 7, so according to our model, it must be a 7:"
]
},
{
"cell_type": "code",
- "execution_count": 51,
+ "execution_count": 95,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
- "array([9, 1, 0, 7, 8, 1, 2, 7, 1, 6])"
+ "array([1.69774721e-04, 1.92234711e-06, 3.67960223e-04, 1.45162235e-03,\n",
+ " 1.42459385e-05, 1.13506248e-04, 2.51079655e-07, 9.95649993e-01,\n",
+ " 1.35322000e-04, 2.09532795e-03], dtype=float32)"
]
},
- "execution_count": 51,
+ "execution_count": 95,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
- "np.argmax(model.predict(X_train[0:10]), axis=1)"
+ "predictions[0]"
]
},
{
"cell_type": "code",
- "execution_count": null,
+ "execution_count": 96,
"metadata": {},
- "outputs": [],
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "7"
+ ]
+ },
+ "execution_count": 96,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
"source": [
- "import numpy as np\n",
- "import matplotlib.pylab as plt\n",
- "%matplotlib inline \n",
- "plt.plot(history.history['accuracy'])\n",
- "plt.plot(history.history['val_accuracy'])\n",
- "plt.title('Model Accuracy')\n",
- "plt.xlabel('epoch')\n",
- "plt.ylabel('accuracy')\n",
- "plt.legend(['train_acc', 'val_acc'], loc='lower right')\n",
- "plt.show()"
+ "predictions[0].argmax()"
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {},
+ "source": [
+ "We can check that the test label agrees:"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "execution_count": 100,
+ "metadata": {},
+ "outputs": [
+ {
+ "data": {
+ "text/plain": [
+ "7"
+ ]
+ },
+ "execution_count": 100,
+ "metadata": {},
+ "output_type": "execute_result"
+ }
+ ],
+ "source": [
+ "y_test[0]"
]
}
],
--
GitLab