MNIST-nn-scipy.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# A neural network from first principles\n",
    "\n",
    "\n",
    "The code below was adpated from the code supplied in Andrew Ng's Coursera course on machine learning. The original code was written in Matlab/Octave and in order to further my understanding and enhance my Python skills, I ported it over to Python."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To start, various libraries are imported.\n",
    "\n",
    "Specifically, the *fmin_l_bfgs_b* function will be used to optimize the cost function."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Import libraries\n",
    "import numpy as np\n",
    "import scipy.io\n",
    "import matplotlib.pyplot as plt\n",
    "import math\n",
    "from scipy.optimize import fmin_l_bfgs_b\n",
    "from sklearn.metrics import accuracy_score\n",
    "import pickle"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First, we load the data. For details, please see the accompanying notebook *MNIST-loader.ipynb* for details."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Load data\n",
    "with open('./data/pickled/xtrain.pickle', 'rb') as f:\n",
    "    xtrain = pickle.load(f)\n",
    "\n",
    "with open('./data/pickled/ytrain.pickle', 'rb') as f:\n",
    "    ytrain = pickle.load(f)\n",
    "\n",
    "with open('./data/pickled/xtest.pickle', 'rb') as f:\n",
    "    xtest = pickle.load(f)\n",
    "\n",
    "with open('./data/pickled/ytest.pickle', 'rb') as f:\n",
    "    ytest = pickle.load(f)\n",
    "\n",
    "with open('./data/pickled/xval.pickle', 'rb') as f:\n",
    "    xval = pickle.load(f)\n",
    "\n",
    "with open('./data/pickled/yval.pickle', 'rb') as f:\n",
    "    yval = pickle.load(f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's define some useful functions for the neural network to use. First is the sigmoid activation function:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Sigmoid function\n",
    "def sigmoid(z):\n",
    "    return 1.0 / (1.0 + np.exp(-z))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### The neural network\n",
    "\n",
    "Below is the real meat of this notebook: the neural network funcion. The function, as defined, takes only one argument: theta. These are the weights as they are known in the moment when the function is called. At the start, these will be randomly initialized, however as the network learns these values will be changed and improved in order to minimize the cost function.\n",
    "\n",
    "Let's walk through the basics. Firstly, the weights are provided in a \"rolled\" format, in other words instead of two matrices of weights, we have a single long list of weights. The first job is to \"unroll\" the weights into two matrices that can be efficiently used by the numpy matrix multiplication methods. To do this, we take the first *(n+1)*hidden* (i.e. 401*20 = 820) values, and reshape them into a 20X401 matrix. Next, we take the remaining 210 values (i.e. *classes*(hidden+1)*) and reshape them into a 10X21 matrix. The *n+1* and *hidden+1* take into account the bias term, which I'll discuss below.\n",
    "#### Forward propagation\n",
    "\n",
    "Next, we perform forward propagation. This is relatively simple: we multiply the input values in the layer by the weights for that layer, sum the total and add the bias term, and then and apply the sigmoid function. Recall from the reading in week one:\n",
    "\n",
    "$$\n",
    "z = w.x+b\n",
    "$$\n",
    "\n",
    "$$\n",
    "a = \\sigma(z) = \\frac{1}{(1+e^{-z})}\n",
    "$$\n",
    "\n",
    "Since *w.x* is the dot product, it implies the sum. Basic matrix multiplication says that when multiplying two matrices that have the same internal dimension (ie number of columns in matrix one is the same as number of rows in matrix two), each element in row *i* of matrix one is multiplied by each element in column *i* of matrix two, and all of those products are summed. This value goes into the first row of the resulting matrix. Subsequently, the same is repeated for the remaining columns of the second matrix, and the first row of the output matrix is completed. This process goes on for all remaining rows in the first matrix.\n",
    "\n",
    "In our example, the first matrix contains MNIST images in a row-wise format. The second matrix contains the weights for each connection between the input layers and the hidden layers. So following the logic from above, the first row of the input matrix (i.e. the first image in the data set) is multiplied by each of 10 sets of weights (10 columns in the weights matrix), one for each hidden layer. Because it's matrix mulitplication, all of these products are automatically summed.\n",
    "#### A quick note about bias\n",
    "\n",
    "If you look at the code below (and elsewhere in this notebook) you'll find a number of *n+1*'s and *hidden+1*'s, etc. These account for bias, the *b* term in the equation above. Every time forward propagation is run, and extra column of ones is appended onto the end of the matrix (these are not a part of the actual data). When the weights are randomly initialized, they too have an extra weight included for this bias term (i.e. the dimensions are *n*+1X*hidden*). These two values, bias in the input matrix and bias in the weights matrix, are multiplied during matrix multiplication and their product is added to the total sum for that neuron. Because the value for bias in the input matrix is always 1, the actual value of the bias is thus coded in the weights and can be learned just like a regular weight.\n",
    "\n",
    "So to sum it all up, for each connection between a node in the input (i.e. a feature, a pixel in the image) and a node in the hidden layer, the input value is multiplied by the weight of each connection and these products for all features are added. To incorporate bias, we include an extra input value of 1 and multiply is by it's own weight. The sigmoid function is applied to this sum, and generates the value of the hidden layer for this particular data point.\n",
    "\n",
    "#### Continuing on with forward propagation\n",
    "\n",
    "Now we have the values of the hidden layer, we repeat this process once again with the weights for the connections between the hidden layer and the output layer. Nothing changes here, except for the sizes of the matrices. Recall that we had *n* input nodes and, say, 20 hidden layer nodes. That means we had *n+1* weights (adding 1 for the bias term), so here we will have *hidden+1* weights.\n",
    "\n",
    "At the end of the second forward propagation, we will have a matrix with a row for each example in the data set and a column for each output class in the neural network (i.e. 10). The columns will contain the value the neural network determined for each class. If the network learned how to identify handwritten digits, the highest of these values will correspond with the correct output. At this point, however, our network has done no learning so we wouldn't expect anything better than a random guess (since the weights were randomly initialized!)\n",
    "\n",
    "#### The cost function\n",
    "\n",
    "Next comes the cost function. Here, we implement the cross entropy cost function with weight decay or L2 regularization. I implemented this in two lines for clarity's sake. First, the unregularized cost is determined and subsequently the regularization term is added.\n",
    "\n",
    "#### And finally, back propagation\n",
    "\n",
    "First, we find the difference between the output values in the 10 classes and the real value. In this case, the real value for each of the 10 possible digits is a vector or length 10, with a 1 in the position representing the number in question and 0's everywhere else. For example, the number 3 is represented by [0 0 0 1 0 0 0 0 0 0]. Since the outputs are from sigmoid neurons, the values will be between 0 and 1 with the highest value indicating the number the model predicted. Sticking with the above example, we might expect our model to output something like [0.1 0.2 0.1 0.6 0.1 0.2 0.2 0.1 0.2 0.3]. Subtracting these two will give a measure of the error. The larger the value, the more incorrect that class prediction was.\n",
    "\n",
    "To perform backpropagation, first we find the delta for the final layer (in this case, d3). This is simply the actual value (which is one-hot encoded) subtracted from the neural networks prediction for that value.\n",
    "\n",
    "Next, we multiple the error from layer 3 by the weights that produced layer three's activation values. This is a matrix multiplication which automatically sums the totals. In this case, the matrices have the dimensions [3750X10] and [10X11], for a dot product of [3750X11]. We can simply perform an elementwise multiplication with the derivative of the sigmoid function with respect to the activations in layer 2 to get the error at layer 2.\n",
    "\n",
    "Since we only have three layers here, we're done. There is no error on layer 1 since this was the input layer. We can't find the error on the raw values that are input to the network!\n",
    "\n",
    "Typically, now we would use these two delta values to update the weights and biases and then run the network again. Rinse and repeat until the cost function is appreciably minimized. But instead of going that route, I decided to implement an optimizer from the scipy package, specifically fmin_l_bfgs_b, which requires *gradients* of the objective function and a set of initial weights. As such, instead of actually updating the weights, we'll simply have the function return rate of change of the cost with respect to the biases and weights. The optimizer will handle the process of actually updating the weights (more on this later).\n",
    "\n",
    "#### Wrapping this function up\n",
    "\n",
    "As you can see, the nn function takes in a set of weights, performs forward propagation to predict the output, calculates the regularized cost using cross entropy and L2 regularization, and then performs backpropagation to determine the rate of change of the cost function with respect to the biases and weights. The gradients are rolled into a vector which can be passed directly to the optimizer. We skipped past this at the beginning of the function, but you can see that the first thing done in the function is to unroll the weights and reshape them into matrices for efficient matrix multiplication during forward and backpropagation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def nn(weights,x,y):\n",
    "    \n",
    "    n = x.shape[0]\n",
    "    bias = np.ones((n,1))\n",
    "\n",
    "    # Unroll weights from a single vector\n",
    "    w2 = weights[:(m+1)*hidden].reshape((hidden,(m+1))) # [Hiddden X m + 1]\n",
    "    w3 = weights[(m+1)*hidden:].reshape((classes,(hidden+1))) # [classes X Hidden + 1]\n",
    "\n",
    "    ### Forward propagation ###\n",
    "\n",
    "    a1 = np.concatenate((bias,x),axis=1) # [nXm+1]\n",
    "\n",
    "    a2 = sigmoid(np.dot(a1,w2.T)) # [nXm+1] . [m+1XHidden] = [nXHidden]\n",
    "    a2 = np.concatenate((bias,a2),axis=1) # [nXHidden+1]\n",
    "\n",
    "    a3 = sigmoid(np.dot(a2,w3.T)) # [nXHidden+1] . [Hidden+1Xclasses] = [nXclasses]\n",
    "\n",
    "    # Cost function: regularized cross entropy\n",
    "    C = np.sum(np.nan_to_num(-y*np.log(a3) - (1-y)*(np.log(1-a3))))/n\n",
    "    C += ((Lambda/(2*n)) * (np.sum(w2[:,1:]**2) + np.sum(w3[:,1:]**2))) # Add regularization to the cost function\n",
    "\n",
    "    ### Backpropagation ###\n",
    "\n",
    "    d3 = (a3 - y) # [nXclasses]\n",
    "    d2 = np.dot(d3,w3) * (a2*(1-a2)) # [nXclasses] . [classesXHidden+1] = [nXHidden+1]\n",
    "    \n",
    "    # We don't need to do this for the input layer, obviously.\n",
    "    # There were no calculations that produced the input layer, that is raw data\n",
    "\n",
    "    # Find the gradients for the weights\n",
    "    grad2 = np.dot(d2[:,1:].T,a1)/n # [HiddenXn] . [nXm+1] = [HiddenXm+1]\n",
    "    grad3 = np.dot(d3.T,a2)/n # [classesXn] . [nXHidden+1] = [classesXHidden+1]\n",
    "\n",
    "    # Regularize the weights only\n",
    "    if Lambda !=0:\n",
    "        grad2[:,1:] -= (Lambda/n * w2[:,1:])\n",
    "        grad3[:,1:] -= (Lambda/n * w3[:,1:])\n",
    "\n",
    "    # Roll weights and biases back into a single vector\n",
    "    grads = np.concatenate((grad2.reshape(((m+1)*hidden)),grad3.reshape(((hidden+1)*classes))))\n",
    "    \n",
    "    return C, grads"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next is the predict function. This function takes the learned weights and performs forward propagation through the netwwork using the x values supplied in the arguments. The effect of this is essentially to predict the output class of the given data using the weights that have been learned. This function will only be called by the accuracy tools at the end, and thus does not perform backpropagation to actually learn the weights."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Predict function\n",
    "def predict(weights,x):\n",
    "    \n",
    "    # Establish some useful variables\n",
    "    n = np.int(x.shape[0])\n",
    "    bias = np.ones((n,1))\n",
    "    \n",
    "    # Unroll weights\n",
    "    w2 = weights[:(m+1)*hidden].reshape((hidden,(m+1)))\n",
    "    w3 = weights[(m+1)*hidden:].reshape((classes,(hidden+1)))\n",
    "    \n",
    "    ### Forward propagation ###\n",
    "    a1 = np.concatenate((bias,x),axis=1)\n",
    "\n",
    "    a2 = sigmoid(np.dot(a1,w2.T))\n",
    "    a2 = np.concatenate((bias,a2),axis=1)\n",
    "\n",
    "    a3 = sigmoid(np.dot(a2,w3.T))\n",
    "    \n",
    "    return np.argmax(a3, axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We initialize theta with a set of random weights with a standard deviation of $ 1/\\sqrt{n} $"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def weight_init(L_in,L_out):\n",
    "    return np.random.normal(scale=1/np.sqrt(L_in), size=(L_out,L_in+1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, we specify the model parameters. Lambda is the regularization parameter, which protects against overfitting.\n",
    "\n",
    "The variable *hidden* specifies the number of nodes in the hidden layer. The input layer is, of course, defined by the number of features. The variable *classes* specifies the number of nodes in the output layer. Finally, *maxiter* is the maximum number of iterations the optimizer will perform. Keeping this small simply lets the model run more quickly during the testing phase."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Model parameters\n",
    "hidden = int(50) # Number of hidden layers\n",
    "Lambda = 3 # Lambda regularization paramter\n",
    "classes = int(10) # Number of output classes\n",
    "maxiter = int(800) # Maximum number of iterations the optimizer is allowed to perform\n",
    "m = np.int(xtrain.shape[1]) # Number of features in each example"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Finally, we train the model\n",
    "\n",
    "Now that all of the various elements have been coded, and the parameters have been set, we're ready to train the model using the training set. Below, the weights are initialized and the model is passed to the *fmin_l_bfgs_b* optimizer. Finally, the learned weights (stored in *model*) are passed to the predict function to see how well the model is predicting the correct digits, and *accuracy_score* from *sklearn* is used to quickly assess the accuracy of the model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The accuracy is on the training set is 0.963060\n",
      "The accuracy is on the test set is 0.952700\n"
     ]
    }
   ],
   "source": [
    "# Initialize weights\n",
    "weights = np.concatenate((weight_init(m,hidden).reshape(((m+1)*hidden)),weight_init(hidden,classes).reshape(((hidden+1)*classes))))\n",
    "\n",
    "# Train the model\n",
    "model = fmin_l_bfgs_b(nn, x0=weights, args=(xtrain,ytrain),maxiter=maxiter)\n",
    "\n",
    "print(\"The accuracy is on the training set is %f\" %(accuracy_score(np.argmax(ytrain,axis=1), predict(model[0],xtrain))))\n",
    "print(\"The accuracy is on the test set is %f\" %(accuracy_score(np.argmax(ytest,axis=1), predict(model[0],xtest))))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing the data\n",
    "\n",
    "Here are two quick functions to visualize the data. First, we randomly select 100 data points and plot them. The second function grabs a single random data point, plots the image and uses the model above to predict the output."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXEAAAEACAYAAABF+UbAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAyQm4gCJPQAIewaSomdhRQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2accf391ddd8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize the data\n",
    "def drawplot(draw,x,y):\n",
    "    if draw:\n",
    "        n = x.shape[0]\n",
    "        idx = np.random.randint(0,n,size=100) # Make an array of random integers between 0 and n\n",
    "        fig, ax = plt.subplots(10, 10) # make the plots\n",
    "        img_size = math.sqrt(m) # Specify the image size (in these case sqrt(m) = 28)\n",
    "        for i in range(10):\n",
    "            for j in range(10):\n",
    "                Xi = x[idx[i*10+j],:].reshape(int(img_size), int(img_size)) # get each example and resize\n",
    "                ax[i,j].set_axis_off() # Turns off the axes for all the subplots for clarity\n",
    "                ax[i,j].imshow(Xi, aspect='auto',cmap='gray') # plots the current image in the correct position\n",
    "        plt.show()\n",
    "        \n",
    "drawplot(True,xtrain,ytrain)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Interactive printer function\n",
    "def printer(x,y,weights):\n",
    "    idx = np.random.randint(len(x),size=1)\n",
    "    img_size = int(math.sqrt(m))\n",
    "    xi = x[idx,:].reshape(img_size,img_size)\n",
    "    yi = predict(weights,x[idx,:])\n",
    "    plt.title('The predicted value is %i\\n The true value is %i' %(yi,np.argmax(y[idx,:],axis=1)))\n",
    "    plt.imshow(xi, aspect='auto',cmap='gray')\n",
    "    plt.axis('off')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAW0AAAEZCAYAAABcsnT1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2aee171d5a58>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "printer(xtrain,ytrain,model[0])"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}