4.1.2_Neural_Network_RELU_vs_Sigmoid.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<p style=\"text-align:center\">\n",
    "    <a href=\"https://skills.network/?utm_medium=Exinfluencer&utm_source=Exinfluencer&utm_content=000026UJ&utm_term=10006555&utm_id=NA-SkillsNetwork-Channel-SkillsNetworkCoursesIBMDeveloperSkillsNetworkCV0101ENCoursera25797139-2022-01-01\" target=\"_blank\">\n",
    "    <img src=\"https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/assets/logos/SN_web_lightmode.png\" width=\"200\" alt=\"Skills Network Logo\"  />\n",
    "    </a>\n",
    "</p>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h1>Neural Network Rectified Linear Unit (ReLU) vs Sigmoid</h1>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h3>Objective for this Notebook<h3>    \n",
    "<h5> 1. Define Several Neural Network, Criterion function, Optimizer.</h5>\n",
    "<h5> 2. Test Sigmoid and Relu. </h5>\n",
    "<h5> 3. Analyze Results. </h5>     \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2>Table of Contents</h2>\n",
    "<p>In this lab, you will test Sigmoid and Relu activation functions on the MNIST dataset with two hidden Layers.</p>\n",
    "\n",
    "<ul>\n",
    "    <li><a href=\"https://#Model\">Neural Network Module and Training Function</a></li>\n",
    "    <li><a href=\"https://#Makeup_Data\">Make Some Data</a></li>\n",
    "    <li><a href=\"https://#Train\">Define Several Neural Network, Criterion function, Optimizer</a></li>\n",
    "    <li><a href=\"https://#Test\">Test Sigmoid and Relu </a></li>\n",
    "    <li><a href=\"https://#Result\">Analyze Results</a></li>\n",
    "</ul>\n",
    "<p>Estimated Time Needed: <strong>25 min</strong></p>\n",
    "\n",
    "<hr>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We'll need the following libraries\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<torch._C.Generator at 0x7f7d0c36e250>"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Import the libraries we need for this lab\n",
    "\n",
    "# Using the following line code to install the torchvision library\n",
    "# !conda install -y torchvision\n",
    "\n",
    "# PyTorch Library\n",
    "import torch \n",
    "# PyTorch Neural Network\n",
    "import torch.nn as nn\n",
    "# Allows us to transform tensors\n",
    "import torchvision.transforms as transforms\n",
    "# Allows us to download datasets\n",
    "import torchvision.datasets as dsets\n",
    "# Allows us to use activation functions\n",
    "import torch.nn.functional as F\n",
    "# Used to graph data and loss curves\n",
    "import matplotlib.pylab as plt\n",
    "# Allows us to use arrays to manipulate and store data\n",
    "import numpy as np\n",
    "# Setting the seed will allow us to control randomness and give us reproducibility\n",
    "torch.manual_seed(2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<!--Empty Space for separating topics-->\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2 id=\"Model\">Neural Network Module and Training Function</h2> \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define the neural network module or class, with two hidden Layers\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src=\"https://ibm.box.com/shared/static/5wtclahun0f70qlwkn2kxzh3amnbq4zg.png\" width=\"200\" alt=\"Neural Network Model\" />\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create the model class using Sigmoid as the activation function\n",
    "\n",
    "class Net(nn.Module):\n",
    "    \n",
    "    # Constructor\n",
    "    def __init__(self, D_in, H1, H2, D_out):\n",
    "        # D_in is the input size of the first layer (size of input layer)\n",
    "        # H1 is the output size of the first layer and input size of the second layer (size of first hidden layer)\n",
    "        # H2 is the outpout size of the second layer and the input size of the third layer (size of second hidden layer)\n",
    "        # D_out is the output size of the third layer (size of output layer)\n",
    "        super(Net, self).__init__()\n",
    "        self.linear1 = nn.Linear(D_in, H1)\n",
    "        self.linear2 = nn.Linear(H1, H2)\n",
    "        self.linear3 = nn.Linear(H2, D_out)\n",
    "    \n",
    "    # Prediction\n",
    "    def forward(self,x):\n",
    "        # Puts x through the first layers then the sigmoid function\n",
    "        x = torch.sigmoid(self.linear1(x)) \n",
    "        # Puts results of previous line through second layer then sigmoid function\n",
    "        x = torch.sigmoid(self.linear2(x))\n",
    "        # Puts result of previous line through third layer\n",
    "        x = self.linear3(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define the class for the Relu activation function\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create the model class using Relu as the activation function\n",
    "\n",
    "class NetRelu(nn.Module):\n",
    "    \n",
    "    # Constructor\n",
    "    def __init__(self, D_in, H1, H2, D_out):\n",
    "        # D_in is the input size of the first layer (size of input layer)\n",
    "        # H1 is the output size of the first layer and input size of the second layer (size of first hidden layer)\n",
    "        # H2 is the outpout size of the second layer and the input size of the third layer (size of second hidden layer)\n",
    "        # D_out is the output size of the third layer (size of output layer)\n",
    "        super(NetRelu, self).__init__()\n",
    "        self.linear1 = nn.Linear(D_in, H1)\n",
    "        self.linear2 = nn.Linear(H1, H2)\n",
    "        self.linear3 = nn.Linear(H2, D_out)\n",
    "    \n",
    "    # Prediction\n",
    "    def forward(self, x):\n",
    "        # Puts x through the first layers then the relu function\n",
    "        x = torch.relu(self.linear1(x))  \n",
    "        # Puts results of previous line through second layer then relu function\n",
    "        x = torch.relu(self.linear2(x))\n",
    "        # Puts result of previous line through third layer\n",
    "        x = self.linear3(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define a function to  train the model, in this case, the function returns a Python dictionary to store the training loss and accuracy on the validation data\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Model Training Function\n",
    "\n",
    "def train(model, criterion, train_loader, validation_loader, optimizer, epochs=100):\n",
    "    i = 0\n",
    "    useful_stuff = {'training_loss': [], 'validation_accuracy': []}  \n",
    "    # Number of times we train on the entire training dataset\n",
    "    for epoch in range(epochs):\n",
    "        # For each batch in the train loader\n",
    "        for i, (x, y) in enumerate(train_loader):\n",
    "            # Resets the calculated gradient value, this must be done each time as it accumulates if we do not reset\n",
    "            optimizer.zero_grad()\n",
    "            # Makes a prediction on the image tensor by flattening it to a 1 by 28*28 tensor\n",
    "            z = model(x.view(-1, 28 * 28))\n",
    "            # Calculate the loss between the prediction and actual class\n",
    "            loss = criterion(z, y)\n",
    "            # Calculates the gradient value with respect to each weight and bias\n",
    "            loss.backward()\n",
    "            # Updates the weight and bias according to calculated gradient value\n",
    "            optimizer.step()\n",
    "            # Saves the loss\n",
    "            useful_stuff['training_loss'].append(loss.data.item())\n",
    "        \n",
    "        # Counter to keep track of correct predictions\n",
    "        correct = 0\n",
    "        # For each batch in the validation dataset\n",
    "        for x, y in validation_loader:\n",
    "            # Make a prediction\n",
    "            z = model(x.view(-1, 28 * 28))\n",
    "            # Get the class that has the maximum value\n",
    "            _, label = torch.max(z, 1)\n",
    "            # Check if our prediction matches the actual class\n",
    "            correct += (label == y).sum().item()\n",
    "    \n",
    "        # Saves the percent accuracy\n",
    "        accuracy = 100 * (correct / len(validation_dataset))\n",
    "        useful_stuff['validation_accuracy'].append(accuracy)\n",
    "    \n",
    "    return useful_stuff"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<!--Empty Space for separating topics-->\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2 id=\"Makeup_Data\">Make Some Data</h2> \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load the training dataset by setting the parameters <code>train</code> to <code>True</code> and convert it to a tensor  by placing a transform object int the argument <code>transform</code>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create the training dataset\n",
    "\n",
    "train_dataset = dsets.MNIST(root='./data', train=True, download=True, transform=transforms.ToTensor())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load the testing dataset by setting the parameters <code>train</code> to <code>False</code> and convert it to a tensor  by placing a transform object int the argument <code>transform</code>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create the validating dataset\n",
    "\n",
    "validation_dataset = dsets.MNIST(root='./data', train=False, download=True, transform=transforms.ToTensor())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create the criterion function\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create the criterion function\n",
    "\n",
    "criterion = nn.CrossEntropyLoss()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create the training-data loader and the validation-data loader object\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create the training data loader and validation data loader object\n",
    "\n",
    "# Batch size is 2000 and shuffle=True means the data will be shuffled at every epoch\n",
    "train_loader = torch.utils.data.DataLoader(dataset=train_dataset, batch_size=2000, shuffle=True)\n",
    "# Batch size is 5000 and the data will not be shuffled at every epoch\n",
    "validation_loader = torch.utils.data.DataLoader(dataset=validation_dataset, batch_size=5000, shuffle=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<!--Empty Space for separating topics-->\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2 id=\"Train\">Define Neural Network, Criterion function, Optimizer and Train the Model</h2> \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Create  the model with 100 hidden neurons\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set the parameters to create the model\n",
    "\n",
    "input_dim = 28 * 28 # Diemension of an image\n",
    "hidden_dim1 = 50\n",
    "hidden_dim2 = 50\n",
    "output_dim = 10 # Number of classes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The epoch number in the video is 35. You can try 10 for now. If you try 35, it may take a long time.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set the number of iterations\n",
    "\n",
    "cust_epochs = 10"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<!--Empty Space for separating topics-->\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2 id=\"Test\">Test Sigmoid and Relu</h2> \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train the network using the Sigmoid activation function\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Train the model with sigmoid function\n",
    "\n",
    "learning_rate = 0.01\n",
    "# Create an instance of the Net model\n",
    "model = Net(input_dim, hidden_dim1, hidden_dim2, output_dim)\n",
    "# Create an optimizer that updates model parameters using the learning rate and gradient\n",
    "optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate)\n",
    "# Train the model\n",
    "training_results = train(model, criterion, train_loader, validation_loader, optimizer, epochs=cust_epochs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train the network using the Relu activation function\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Train the model with relu function\n",
    "\n",
    "learning_rate = 0.01\n",
    "# Create an instance of the NetRelu model\n",
    "modelRelu = NetRelu(input_dim, hidden_dim1, hidden_dim2, output_dim)\n",
    "# Create an optimizer that updates model parameters using the learning rate and gradient\n",
    "optimizer = torch.optim.SGD(modelRelu.parameters(), lr=learning_rate)\n",
    "# Train the model\n",
    "training_results_relu = train(modelRelu, criterion, train_loader, validation_loader, optimizer, epochs=cust_epochs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<!--Empty Space for separating topics-->\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2 id=\"Result\">Analyze Results</h2> \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compare the training loss for each activation\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f7c895a8650>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Compare the training loss\n",
    "\n",
    "plt.plot(training_results['training_loss'], label='sigmoid')\n",
    "plt.plot(training_results_relu['training_loss'], label='relu')\n",
    "plt.ylabel('loss')\n",
    "plt.title('training loss iterations')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Compare the validation loss for each model\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Compare the validation loss\n",
    "\n",
    "plt.plot(training_results['validation_accuracy'], label = 'sigmoid')\n",
    "plt.plot(training_results_relu['validation_accuracy'], label = 'relu') \n",
    "plt.ylabel('validation accuracy')\n",
    "plt.xlabel('Iteration')   \n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a href=\"https://dataplatform.cloud.ibm.com/registration/stepone?utm_medium=Exinfluencer&utm_source=Exinfluencer&utm_content=000026UJ&utm_term=10006555&utm_id=NA-SkillsNetwork-Channel-SkillsNetworkCoursesIBMDeveloperSkillsNetworkCV0101ENCoursera25797139-2022-01-01&context=cpdaas&apps=data_science_experience%2Cwatson_machine_learning\"><img src=\"https://cf-courses-data.s3.us.cloud-object-storage.appdomain.cloud/IBMDeveloperSkillsNetwork-DL0110EN-SkillsNetwork/Template/module%201/images/Watson_Studio.png\"/></a>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<!--Empty Space for separating topics-->\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h2>About the Authors:</h2> \n",
    "\n",
    "<a href=\"https://www.linkedin.com/in/joseph-s-50398b136/?utm_medium=Exinfluencer&utm_source=Exinfluencer&utm_content=000026UJ&utm_term=10006555&utm_id=NA-SkillsNetwork-Channel-SkillsNetworkCoursesIBMDeveloperSkillsNetworkCV0101ENCoursera25797139-2022-01-01\">Joseph Santarcangelo</a> has a PhD in Electrical Engineering, his research focused on using machine learning, signal processing, and computer vision to determine how videos impact human cognition. Joseph has been working for IBM since he completed his PhD.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Other contributors: <a href=\"https://www.linkedin.com/in/michelleccarey/?utm_medium=Exinfluencer&utm_source=Exinfluencer&utm_content=000026UJ&utm_term=10006555&utm_id=NA-SkillsNetwork-Channel-SkillsNetworkCoursesIBMDeveloperSkillsNetworkCV0101ENCoursera25797139-2022-01-01\">Michelle Carey</a>, <a href=\"https://www.linkedin.com/in/jiahui-mavis-zhou-a4537814a?utm_medium=Exinfluencer&utm_source=Exinfluencer&utm_content=000026UJ&utm_term=10006555&utm_id=NA-SkillsNetwork-Channel-SkillsNetworkCoursesIBMDeveloperSkillsNetworkCV0101ENCoursera25797139-2022-01-01\">Mavis Zhou</a>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Change Log\n",
    "\n",
    "| Date (YYYY-MM-DD) | Version | Changed By | Change Description                                          |\n",
    "| ----------------- | ------- | ---------- | ----------------------------------------------------------- |\n",
    "| 2020-09-23        | 2.0     | Srishti    | Migrated Lab to Markdown and added to course repo in GitLab |\n",
    "\n",
    "<hr>\n",
    "\n",
    "## <h3 align=\"center\"> © IBM Corporation 2020. All rights reserved. <h3/>\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python",
   "language": "python",
   "name": "conda-env-python-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}