examples/models-usages/mlp-classification-regression/mnist_multiclass.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Simple MNIST multiclass classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "tags": []
   },
   "source": [
    "## Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-11-14T19:05:21.610786900Z",
     "start_time": "2024-11-14T19:05:12.664757Z"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from tensorflow.keras.datasets import mnist\n",
    "\n",
    "from neuralnetlib.activations import Sigmoid, Softmax\n",
    "from neuralnetlib.layers import Input, Dense, Activation, Dropout\n",
    "from neuralnetlib.losses import CategoricalCrossentropy\n",
    "from neuralnetlib.models import Sequential\n",
    "from neuralnetlib.optimizers import SGD\n",
    "from neuralnetlib.preprocessing import one_hot_encode\n",
    "from neuralnetlib.metrics import accuracy_score, f1_score, recall_score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Loading a dataset (in this case, MNIST)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-11-14T19:05:21.766319Z",
     "start_time": "2024-11-14T19:05:21.613296100Z"
    }
   },
   "outputs": [],
   "source": [
    "(x_train, y_train), (x_test, y_test) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-11-14T19:05:21.859478500Z",
     "start_time": "2024-11-14T19:05:21.766319Z"
    }
   },
   "outputs": [],
   "source": [
    "x_train = x_train.reshape(-1, 28 * 28) / 255.0  # Normalization and flattening of the images\n",
    "x_test = x_test.reshape(-1, 28 * 28) / 255.0  # Normalization and flattening of the images\n",
    "y_train = one_hot_encode(y_train, num_classes=10)  # One-hot encoding of the labels\n",
    "y_test = one_hot_encode(y_test, num_classes=10)  # One-hot encoding of the labels"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Model definition"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-11-14T19:05:21.874505200Z",
     "start_time": "2024-11-14T19:05:21.861483900Z"
    }
   },
   "outputs": [],
   "source": [
    "input_neurons = x_train.shape[1:][0]  # MNIST images are 28x28\n",
    "num_hidden_layers = 2  # Number of hidden layers\n",
    "hidden_neurons = 30  # Number of neurons in each hidden layer\n",
    "output_neurons = 10  # Assuming 10 classes for MNIST\n",
    "\n",
    "model = Sequential()\n",
    "model.add(Input(input_neurons))\n",
    "model.add(Dense(hidden_neurons, random_state=42))  # First hidden layer\n",
    "model.add(Activation(Sigmoid()))  # ...and its function activation\n",
    "\n",
    "for _ in range(num_hidden_layers - 1):  # Add the rest of the hidden layers\n",
    "    model.add(Dense(hidden_neurons, random_state=42))  # Hidden layer must have the same number of neurons as the previous one\n",
    "    model.add(Activation(Sigmoid()))  # ...and its function activation\n",
    "\n",
    "model.add(Dropout(0.5, random_state=42))  # Dropout layer\n",
    "model.add(Dense(output_neurons, random_state=42))  # Output layer\n",
    "model.add(Activation(Softmax()))  # ...and its function activation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Model compilation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-11-14T19:05:21.920165800Z",
     "start_time": "2024-11-14T19:05:21.875506100Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sequential(gradient_clip_threshold=5.0, enable_padding=False, padding_size=32, random_state=1733515497021160100)\n",
      "-------------------------------------------------\n",
      "Layer 1: Input(input_shape=(784,))\n",
      "Layer 2: Dense(units=30)\n",
      "Layer 3: Activation(Sigmoid)\n",
      "Layer 4: Dense(units=30)\n",
      "Layer 5: Activation(Sigmoid)\n",
      "Layer 6: Dropout(rate=0.5)\n",
      "Layer 7: Dense(units=10)\n",
      "Layer 8: Activation(Softmax)\n",
      "-------------------------------------------------\n",
      "Loss function: CategoricalCrossentropy\n",
      "Optimizer: SGD(learning_rate=0.1)\n",
      "-------------------------------------------------\n",
      "\n"
     ]
    }
   ],
   "source": [
    "model.compile(loss_function=CategoricalCrossentropy(), optimizer=SGD(learning_rate=0.1))\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Model training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-11-14T19:06:24.402719700Z",
     "start_time": "2024-11-14T19:05:21.892523Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[==============================] 100% Epoch 1/10 - 6.62s - loss: 1.0245 - accuracy: 0.6601\n",
      "[==============================] 100% Epoch 2/10 - 6.38s - loss: 0.4829 - accuracy: 0.8642\n",
      "[==============================] 100% Epoch 3/10 - 6.37s - loss: 0.3815 - accuracy: 0.8938\n",
      "[==============================] 100% Epoch 4/10 - 6.27s - loss: 0.3226 - accuracy: 0.9108\n",
      "[==============================] 100% Epoch 5/10 - 6.27s - loss: 0.2833 - accuracy: 0.9219\n",
      "[==============================] 100% Epoch 6/10 - 6.30s - loss: 0.2557 - accuracy: 0.9297\n",
      "[==============================] 100% Epoch 7/10 - 6.25s - loss: 0.2350 - accuracy: 0.9358\n",
      "[==============================] 100% Epoch 8/10 - 6.22s - loss: 0.2187 - accuracy: 0.9403\n",
      "[==============================] 100% Epoch 9/10 - 6.33s - loss: 0.2055 - accuracy: 0.9439\n",
      "[==============================] 100% Epoch 10/10 - 6.23s - loss: 0.1946 - accuracy: 0.9471\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(x_train, y_train, epochs=10, batch_size=48, metrics=[accuracy_score], random_state=42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6. Model evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-11-14T19:06:24.513544300Z",
     "start_time": "2024-11-14T19:06:24.373637900Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 0.1708824620646567\n"
     ]
    }
   ],
   "source": [
    "loss, _ = model.evaluate(x_test, y_test)\n",
    "print(f'Test loss: {loss}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 7. Model prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-11-14T19:06:24.560136Z",
     "start_time": "2024-11-14T19:06:24.515542300Z"
    }
   },
   "outputs": [],
   "source": [
    "y_pred = model.predict(x_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 8. Printing some metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-11-14T19:06:24.564652900Z",
     "start_time": "2024-11-14T19:06:24.547612700Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy: 0.9524\n",
      "f1_score: 0.952080979200471\n",
      "recall_score 0.951536374068995\n"
     ]
    }
   ],
   "source": [
    "print(\"accuracy:\", accuracy_score(y_pred, y_test))\n",
    "print(\"f1_score:\", f1_score(y_pred, y_test))\n",
    "print(\"recall_score\", recall_score(y_pred, y_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 9. Plot the first 10 test images, their predicted labels, and the true labels."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-11-14T19:06:24.781079900Z",
     "start_time": "2024-11-14T19:06:24.564652900Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x1000 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(10, 10))\n",
    "for i in range(10):\n",
    "    ax = fig.add_subplot(5, 2, i + 1, xticks=[], yticks=[])\n",
    "    ax.imshow(x_test[i].reshape(28, 28), cmap='gray')\n",
    "    ax.set_title(f\"Predicted: {np.argmax(y_pred[i])}, Actual: {np.argmax(y_test[i])}\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 10. Save the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-11-14T19:06:24.857759900Z",
     "start_time": "2024-11-14T19:06:24.782589300Z"
    }
   },
   "outputs": [],
   "source": [
    "model.save(\"my_mnist_model.npz\")"
   ]
  }
 ],
 "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": "3.10.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}