assignment1/softmax.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Softmax exercise\n",
    "\n",
    "*Complete and hand in this completed worksheet (including its outputs and any supporting code outside of the worksheet) with your assignment submission. For more details see the [assignments page](http://vision.stanford.edu/teaching/cs231n/assignments.html) on the course website.*\n",
    "\n",
    "This exercise is analogous to the SVM exercise. You will:\n",
    "\n",
    "- implement a fully-vectorized **loss function** for the Softmax classifier\n",
    "- implement the fully-vectorized expression for its **analytic gradient**\n",
    "- **check your implementation** with numerical gradient\n",
    "- use a validation set to **tune the learning rate and regularization** strength\n",
    "- **optimize** the loss function with **SGD**\n",
    "- **visualize** the final learned weights\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import random\n",
    "import numpy as np\n",
    "from cs231n.data_utils import load_CIFAR10\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "plt.rcParams['figure.figsize'] = (10.0, 8.0) # set default size of plots\n",
    "plt.rcParams['image.interpolation'] = 'nearest'\n",
    "plt.rcParams['image.cmap'] = 'gray'\n",
    "\n",
    "# for auto-reloading extenrnal modules\n",
    "# see http://stackoverflow.com/questions/1907993/autoreload-of-modules-in-ipython\n",
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train data shape:  (49000, 3073)\n",
      "Train labels shape:  (49000,)\n",
      "Validation data shape:  (1000, 3073)\n",
      "Validation labels shape:  (1000,)\n",
      "Test data shape:  (1000, 3073)\n",
      "Test labels shape:  (1000,)\n",
      "dev data shape:  (500, 3073)\n",
      "dev labels shape:  (500,)\n"
     ]
    }
   ],
   "source": [
    "def get_CIFAR10_data(num_training=49000, num_validation=1000, num_test=1000, num_dev=500):\n",
    "  \"\"\"\n",
    "  Load the CIFAR-10 dataset from disk and perform preprocessing to prepare\n",
    "  it for the linear classifier. These are the same steps as we used for the\n",
    "  SVM, but condensed to a single function.  \n",
    "  \"\"\"\n",
    "  # Load the raw CIFAR-10 data\n",
    "  cifar10_dir = 'cs231n/datasets/cifar-10-batches-py'\n",
    "  X_train, y_train, X_test, y_test = load_CIFAR10(cifar10_dir)\n",
    "  \n",
    "  # subsample the data\n",
    "  mask = range(num_training, num_training + num_validation)\n",
    "  X_val = X_train[mask]\n",
    "  y_val = y_train[mask]\n",
    "  mask = range(num_training)\n",
    "  X_train = X_train[mask]\n",
    "  y_train = y_train[mask]\n",
    "  mask = range(num_test)\n",
    "  X_test = X_test[mask]\n",
    "  y_test = y_test[mask]\n",
    "  mask = np.random.choice(num_training, num_dev, replace=False)\n",
    "  X_dev = X_train[mask]\n",
    "  y_dev = y_train[mask]\n",
    "  \n",
    "  # Preprocessing: reshape the image data into rows\n",
    "  X_train = np.reshape(X_train, (X_train.shape[0], -1))\n",
    "  X_val = np.reshape(X_val, (X_val.shape[0], -1))\n",
    "  X_test = np.reshape(X_test, (X_test.shape[0], -1))\n",
    "  X_dev = np.reshape(X_dev, (X_dev.shape[0], -1))\n",
    "  \n",
    "  # Normalize the data: subtract the mean image\n",
    "  mean_image = np.mean(X_train, axis = 0)\n",
    "  X_train -= mean_image\n",
    "  X_val -= mean_image\n",
    "  X_test -= mean_image\n",
    "  X_dev -= mean_image\n",
    "  \n",
    "  # add bias dimension and transform into columns\n",
    "  X_train = np.hstack([X_train, np.ones((X_train.shape[0], 1))])\n",
    "  X_val = np.hstack([X_val, np.ones((X_val.shape[0], 1))])\n",
    "  X_test = np.hstack([X_test, np.ones((X_test.shape[0], 1))])\n",
    "  X_dev = np.hstack([X_dev, np.ones((X_dev.shape[0], 1))])\n",
    "  \n",
    "  return X_train, y_train, X_val, y_val, X_test, y_test, X_dev, y_dev\n",
    "\n",
    "\n",
    "# Invoke the above function to get our data.\n",
    "X_train, y_train, X_val, y_val, X_test, y_test, X_dev, y_dev = get_CIFAR10_data()\n",
    "print 'Train data shape: ', X_train.shape\n",
    "print 'Train labels shape: ', y_train.shape\n",
    "print 'Validation data shape: ', X_val.shape\n",
    "print 'Validation labels shape: ', y_val.shape\n",
    "print 'Test data shape: ', X_test.shape\n",
    "print 'Test labels shape: ', y_test.shape\n",
    "print 'dev data shape: ', X_dev.shape\n",
    "print 'dev labels shape: ', y_dev.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Softmax Classifier\n",
    "\n",
    "Your code for this section will all be written inside **cs231n/classifiers/softmax.py**. \n",
    "\n",
    "The loss function now is a cross-entropy loss, for a single data point it forms as:\n",
    "\n",
    "$$\n",
    "L_i = -\\log\\left(\\frac{e^{f_{y_i}}}{ \\sum_j e^{f_j} }\\right) \\hspace{0.5in} \\text{or equivalently} \\hspace{0.5in} L_i = -f_{y_i} + \\log\\sum_j e^{f_j} \\hspace{0.5in} \\text{where} \\hspace{0.5in} f(x_i; W) =  W x_i\n",
    "$$\n",
    "\n",
    "Consider of the numeric stability issues, it's better to have a normalization term $\\log C = -\\max_j f_j$ in the division of the softmax function:\n",
    "\n",
    "$$\n",
    "\\frac{e^{f_{y_i}}}{\\sum_j e^{f_j}}\n",
    "= \\frac{Ce^{f_{y_i}}}{C\\sum_j e^{f_j}}\n",
    "= \\frac{e^{f_{y_i} + \\log C}}{\\sum_j e^{f_j + \\log C}}\n",
    "$$\n",
    "\n",
    "The loss of the entire data set with regularization term forms as:\n",
    "\n",
    "$$\n",
    "\\begin{align}\n",
    "L &= \\frac{1}{N}\\sum_iL_i + \\lambda R(W)\\\\\n",
    "&= \\frac{1}{N}\\sum_i\\left(-\\left(w_{y_i}^Tx_i+\\log C\\right)+\\log\\sum_j\\exp\\left(w_j^Tx_i+\\log C\\right)\\right) + \\frac{\\lambda}{2}\\sum_k\\sum_lW_{k,l}^2\\\\\n",
    "\\end{align}\n",
    "$$\n",
    "\n",
    "The gradient of the loss function for a single data point without regularization term is:\n",
    "\n",
    "$$\n",
    "\\begin{align}\n",
    "\\nabla_{w_{y_i}} L_i &= -x_i+\\frac{\\exp\\left(w_{y_i}^Tx_i+\\log C\\right)}{\\sum_j\\exp\\left(w_j^Tx_i+\\log C\\right)}x_i\\\\\n",
    "&=-\\frac{\\sum_{j\\neq y_i}\\exp\\left(w_j^Tx_i+\\log C\\right)}{\\sum_j\\exp\\left(w_j^Tx_i+\\log C\\right)}x_i\\\\\n",
    "\\nabla_{w_{j}} L_i &= \\frac{\\exp\\left(w_j^Tx_i+\\log C\\right)}{\\sum_j\\exp\\left(w_j^Tx_i+\\log C\\right)}x_i\n",
    "\\end{align}\n",
    "$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss: 2.380654\n",
      "sanity check: 2.302585\n"
     ]
    }
   ],
   "source": [
    "# First implement the naive softmax loss function with nested loops.\n",
    "# Open the file cs231n/classifiers/softmax.py and implement the\n",
    "# softmax_loss_naive function.\n",
    "\n",
    "from cs231n.classifiers.softmax import softmax_loss_naive\n",
    "import time\n",
    "\n",
    "# Generate a random softmax weight matrix and use it to compute the loss.\n",
    "W = np.random.randn(3073, 10) * 0.0001\n",
    "loss, grad = softmax_loss_naive(W, X_dev, y_dev, 0.0)\n",
    "\n",
    "# As a rough sanity check, our loss should be something close to -log(0.1).\n",
    "print 'loss: %f' % loss\n",
    "print 'sanity check: %f' % (-np.log(0.1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Inline Question 1:\n",
    "Why do we expect our loss to be close to -log(0.1)? Explain briefly.**\n",
    "\n",
    "**Your answer:** *Because the accuracy of an random weight matrix which classifies the images into 10 classes is 10%, then the log softmax function is close to $\\displaystyle-\\log\\frac{1}{10}.$*\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "numerical: -0.057818 analytic: -0.057818, relative error: 7.044087e-07\n",
      "numerical: 0.016043 analytic: 0.016043, relative error: 2.495744e-06\n",
      "numerical: 3.835465 analytic: 3.835465, relative error: 1.234337e-08\n",
      "numerical: -0.601061 analytic: -0.601062, relative error: 8.216985e-08\n",
      "numerical: -0.330275 analytic: -0.330275, relative error: 7.685813e-09\n",
      "numerical: 2.721561 analytic: 2.721561, relative error: 1.660158e-09\n",
      "numerical: -3.546576 analytic: -3.546576, relative error: 6.107912e-09\n",
      "numerical: -1.811729 analytic: -1.811730, relative error: 3.784843e-08\n",
      "numerical: -0.459017 analytic: -0.459017, relative error: 7.141842e-08\n",
      "numerical: 3.864395 analytic: 3.864395, relative error: 1.819635e-08\n",
      "numerical: 0.360173 analytic: 0.360173, relative error: 9.379433e-08\n",
      "numerical: 2.403055 analytic: 2.403055, relative error: 3.081273e-08\n",
      "numerical: 2.806226 analytic: 2.806226, relative error: 2.915318e-09\n",
      "numerical: -1.107608 analytic: -1.107608, relative error: 6.558224e-09\n",
      "numerical: -2.084383 analytic: -2.084383, relative error: 1.761838e-08\n",
      "numerical: -0.512677 analytic: -0.512677, relative error: 4.629438e-08\n",
      "numerical: -1.605896 analytic: -1.605896, relative error: 2.217550e-08\n",
      "numerical: 0.419437 analytic: 0.419437, relative error: 1.057662e-07\n",
      "numerical: 0.317961 analytic: 0.317961, relative error: 3.571370e-08\n",
      "numerical: 0.721530 analytic: 0.721530, relative error: 1.388511e-08\n"
     ]
    }
   ],
   "source": [
    "# Complete the implementation of softmax_loss_naive and implement a (naive)\n",
    "# version of the gradient that uses nested loops.\n",
    "loss, grad = softmax_loss_naive(W, X_dev, y_dev, 0.0)\n",
    "\n",
    "# As we did for the SVM, use numeric gradient checking as a debugging tool.\n",
    "# The numeric gradient should be close to the analytic gradient.\n",
    "from cs231n.gradient_check import grad_check_sparse\n",
    "f = lambda w: softmax_loss_naive(w, X_dev, y_dev, 0.0)[0]\n",
    "grad_numerical = grad_check_sparse(f, W, grad, 10)\n",
    "\n",
    "# similar to SVM case, do another gradient check with regularization\n",
    "loss, grad = softmax_loss_naive(W, X_dev, y_dev, 1e2)\n",
    "f = lambda w: softmax_loss_naive(w, X_dev, y_dev, 1e2)[0]\n",
    "grad_numerical = grad_check_sparse(f, W, grad, 10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "From another perspective, softmax function compute the probability of each sample in all class:\n",
    "\n",
    "$$\n",
    "p\\left(y_i=k\\mid x_i;w\\right)=\\frac{\\exp\\left(w_k^Tx_i+\\log C\\right)}{\\sum_j\\exp\\left(w_j^Tx_i+\\log C\\right)}\n",
    "$$\n",
    "\n",
    "The loss function can be written as:\n",
    "\n",
    "$$\n",
    "\\begin{align}\n",
    "L&=-\\frac{1}{N}\\left[\\sum_{i=1}^N\\sum_{j=1}^C1\\{y_i=j\\}\\log\\left(\\frac{\\exp\\left(w_k^Tx_i+\\log C\\right)}{\\sum_j\\exp\\left(w_j^Tx_i+\\log C\\right)}\\right)\\right]+\\frac{\\lambda}{2}\\sum_{k=1}^D\\sum_{l=1}^Cw_{k,l}^2\\\\\n",
    "&=-\\frac{1}{N}\\left[\\sum_{i=1}^N\\sum_{j=1}^C1\\{y_i=j\\}\\log\\left(p\\left(y_i=k\\mid x_i;w\\right)\\right)\\right]+\\frac{\\lambda}{2}\\sum_{k=1}^D\\sum_{l=1}^Cw_{k,l}^2\\\\\n",
    "\\end{align}\n",
    "$$\n",
    "\n",
    "Then the gradient is:\n",
    "\n",
    "$$\n",
    "\\nabla_{w_j}L=-\\frac{1}{N}\\sum_{i=1}^N[x_i(1\\{y_i=j\\}-p\\left(y_i=k\\mid x_i;w\\right))]+\\lambda w_j\n",
    "$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "naive loss: 2.380654e+00 computed in 0.141194s\n",
      "vectorized loss: 2.380654e+00 computed in 0.004839s\n",
      "Loss difference: 0.000000\n",
      "Gradient difference: 0.000000\n"
     ]
    }
   ],
   "source": [
    "# Now that we have a naive implementation of the softmax loss function and its gradient,\n",
    "# implement a vectorized version in softmax_loss_vectorized.\n",
    "# The two versions should compute the same results, but the vectorized version should be\n",
    "# much faster.\n",
    "tic = time.time()\n",
    "loss_naive, grad_naive = softmax_loss_naive(W, X_dev, y_dev, 0.00001)\n",
    "toc = time.time()\n",
    "print 'naive loss: %e computed in %fs' % (loss_naive, toc - tic)\n",
    "\n",
    "from cs231n.classifiers.softmax import softmax_loss_vectorized\n",
    "tic = time.time()\n",
    "loss_vectorized, grad_vectorized = softmax_loss_vectorized(W, X_dev, y_dev, 0.00001)\n",
    "toc = time.time()\n",
    "print 'vectorized loss: %e computed in %fs' % (loss_vectorized, toc - tic)\n",
    "\n",
    "# As we did for the SVM, we use the Frobenius norm to compare the two versions\n",
    "# of the gradient.\n",
    "grad_difference = np.linalg.norm(grad_naive - grad_vectorized, ord='fro')\n",
    "print 'Loss difference: %f' % np.abs(loss_naive - loss_vectorized)\n",
    "print 'Gradient difference: %f' % grad_difference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "iteration 0 / 1500: loss 778.940626\n",
      "iteration 100 / 1500: loss 285.381676\n",
      "iteration 200 / 1500: loss 105.751914\n",
      "iteration 300 / 1500: loss 40.018756\n",
      "iteration 400 / 1500: loss 15.913623\n",
      "iteration 500 / 1500: loss 7.169015\n",
      "iteration 600 / 1500: loss 3.920216\n",
      "iteration 700 / 1500: loss 2.739906\n",
      "iteration 800 / 1500: loss 2.331901\n",
      "iteration 900 / 1500: loss 2.144304\n",
      "iteration 1000 / 1500: loss 2.102148\n",
      "iteration 1100 / 1500: loss 2.113951\n",
      "iteration 1200 / 1500: loss 2.039363\n",
      "iteration 1300 / 1500: loss 2.080034\n",
      "iteration 1400 / 1500: loss 2.106774\n",
      "iteration 0 / 1500: loss 1529242.425188\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "cs231n/classifiers/softmax.py:86: RuntimeWarning: divide by zero encountered in log\n",
      "  Margin = np.log(Probability)\n",
      "cs231n/classifiers/softmax.py:89: RuntimeWarning: invalid value encountered in multiply\n",
      "  Mask_Margin = Mask_y * Margin\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "iteration 100 / 1500: loss nan\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "cs231n/classifiers/softmax.py:90: RuntimeWarning: overflow encountered in double_scalars\n",
      "  loss = -np.sum(Mask_Margin) / num_train + 0.5*reg*np.sum(W*W)\n",
      "cs231n/classifiers/softmax.py:90: RuntimeWarning: overflow encountered in multiply\n",
      "  loss = -np.sum(Mask_Margin) / num_train + 0.5*reg*np.sum(W*W)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "iteration 200 / 1500: loss nan\n",
      "iteration 300 / 1500: loss nan\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "cs231n/classifiers/softmax.py:94: RuntimeWarning: overflow encountered in multiply\n",
      "  dW = -X.T.dot(dProbability) / num_train + reg*W\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "iteration 400 / 1500: loss nan\n",
      "iteration 500 / 1500: loss nan\n",
      "iteration 600 / 1500: loss nan\n",
      "iteration 700 / 1500: loss nan\n",
      "iteration 800 / 1500: loss nan\n",
      "iteration 900 / 1500: loss nan\n",
      "iteration 1000 / 1500: loss nan\n",
      "iteration 1100 / 1500: loss nan\n",
      "iteration 1200 / 1500: loss nan\n",
      "iteration 1300 / 1500: loss nan\n",
      "iteration 1400 / 1500: loss nan\n",
      "iteration 0 / 1500: loss 771.497124\n",
      "iteration 100 / 1500: loss 6.859397\n",
      "iteration 200 / 1500: loss 2.158726\n",
      "iteration 300 / 1500: loss 2.063420\n",
      "iteration 400 / 1500: loss 2.058667\n",
      "iteration 500 / 1500: loss 2.090075\n",
      "iteration 600 / 1500: loss 2.102466\n",
      "iteration 700 / 1500: loss 2.093505\n",
      "iteration 800 / 1500: loss 2.078734\n",
      "iteration 900 / 1500: loss 2.089915\n",
      "iteration 1000 / 1500: loss 2.003685\n",
      "iteration 1100 / 1500: loss 2.155332\n",
      "iteration 1200 / 1500: loss 2.130461\n",
      "iteration 1300 / 1500: loss 2.097626\n",
      "iteration 1400 / 1500: loss 2.119495\n",
      "iteration 0 / 1500: loss 1551567.186520\n",
      "iteration 100 / 1500: loss nan\n",
      "iteration 200 / 1500: loss nan\n",
      "iteration 300 / 1500: loss nan\n",
      "iteration 400 / 1500: loss nan\n",
      "iteration 500 / 1500: loss nan\n",
      "iteration 600 / 1500: loss nan\n",
      "iteration 700 / 1500: loss nan\n",
      "iteration 800 / 1500: loss nan\n",
      "iteration 900 / 1500: loss nan\n",
      "iteration 1000 / 1500: loss nan\n",
      "iteration 1100 / 1500: loss nan\n",
      "iteration 1200 / 1500: loss nan\n",
      "iteration 1300 / 1500: loss nan\n",
      "iteration 1400 / 1500: loss nan\n",
      "lr 1.000000e-07 reg 5.000000e+04 train accuracy: 0.329163 val accuracy: 0.344000\n",
      "lr 1.000000e-07 reg 1.000000e+08 train accuracy: 0.100265 val accuracy: 0.087000\n",
      "lr 5.000000e-07 reg 5.000000e+04 train accuracy: 0.323837 val accuracy: 0.345000\n",
      "lr 5.000000e-07 reg 1.000000e+08 train accuracy: 0.100265 val accuracy: 0.087000\n",
      "best validation accuracy achieved during cross-validation: 0.345000\n"
     ]
    }
   ],
   "source": [
    "# Use the validation set to tune hyperparameters (regularization strength and\n",
    "# learning rate). You should experiment with different ranges for the learning\n",
    "# rates and regularization strengths; if you are careful you should be able to\n",
    "# get a classification accuracy of over 0.35 on the validation set.\n",
    "from cs231n.classifiers import Softmax\n",
    "results = {}\n",
    "best_val = -1\n",
    "best_softmax = None\n",
    "learning_rates = [1e-7, 5e-7]\n",
    "regularization_strengths = [5e4, 1e8]\n",
    "\n",
    "################################################################################\n",
    "# TODO:                                                                        #\n",
    "# Use the validation set to set the learning rate and regularization strength. #\n",
    "# This should be identical to the validation that you did for the SVM; save    #\n",
    "# the best trained softmax classifer in best_softmax.                          #\n",
    "################################################################################\n",
    "from itertools import product\n",
    "for lr, rs in product(learning_rates, regularization_strengths):\n",
    "    sm = Softmax()\n",
    "    loss_hist = sm.train(X_train, y_train, learning_rate=lr, reg=rs,\n",
    "                         num_iters=1500, verbose=True)\n",
    "    y_train_pred = sm.predict(X_train)\n",
    "    training_accuracy = np.mean(y_train == y_train_pred)\n",
    "    y_val_pred = sm.predict(X_val)\n",
    "    validation_accuracy = np.mean(y_val == y_val_pred)\n",
    "    results[(lr, rs)] = (training_accuracy, validation_accuracy)\n",
    "    if validation_accuracy > best_val:\n",
    "        best_val = validation_accuracy\n",
    "        best_softmax = sm\n",
    "################################################################################\n",
    "#                              END OF YOUR CODE                                #\n",
    "################################################################################\n",
    "    \n",
    "# Print out results.\n",
    "for lr, reg in sorted(results):\n",
    "    train_accuracy, val_accuracy = results[(lr, reg)]\n",
    "    print 'lr %e reg %e train accuracy: %f val accuracy: %f' % (\n",
    "                lr, reg, train_accuracy, val_accuracy)\n",
    "    \n",
    "print 'best validation accuracy achieved during cross-validation: %f' % best_val"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "softmax on raw pixels final test set accuracy: 0.333000\n"
     ]
    }
   ],
   "source": [
    "# evaluate on test set\n",
    "# Evaluate the best softmax on test set\n",
    "y_test_pred = best_softmax.predict(X_test)\n",
    "test_accuracy = np.mean(y_test == y_test_pred)\n",
    "print 'softmax on raw pixels final test set accuracy: %f' % (test_accuracy, )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAzQAAAIUCAYAAADMqWn1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10e120ad0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize the learned weights for each class\n",
    "w = best_softmax.W[:-1,:] # strip out the bias\n",
    "w = w.reshape(32, 32, 3, 10)\n",
    "\n",
    "w_min, w_max = np.min(w), np.max(w)\n",
    "\n",
    "classes = ['plane', 'car', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck']\n",
    "for i in xrange(10):\n",
    "  plt.subplot(2, 5, i + 1)\n",
    "  \n",
    "  # Rescale the weights to be between 0 and 255\n",
    "  wimg = 255.0 * (w[:, :, :, i].squeeze() - w_min) / (w_max - w_min)\n",
    "  plt.imshow(wimg.astype('uint8'))\n",
    "  plt.axis('off')\n",
    "  plt.title(classes[i])"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:python27]",
   "language": "python",
   "name": "conda-env-python27-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}