2-Layer-DK-AlignMNiST.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-04-20T21:52:04.548207",
     "start_time": "2016-04-20T21:52:04.544771"
    }
   },
   "source": [
    "## 2 Layer MLP for Dark Knoweldge Experiments w/AlignMNIST\n",
    "\n",
    "75 epochs of RMSProp\n",
    "\n",
    "Eval extra data first, on same size network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:22.195643",
     "start_time": "2016-06-13T11:41:20.925794"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:22.202863",
     "start_time": "2016-06-13T11:41:22.197182"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "%run augmentmnist.py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:22.207613",
     "start_time": "2016-06-13T11:41:22.204990"
    },
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "num_epochs = 75\n",
    "batch_size = 125\n",
    "\n",
    "learning_rate = 0.001"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:22.215134",
     "start_time": "2016-06-13T11:41:22.209358"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Network Parameters\n",
    "n_hidden_1 = 800 # 1st layer num features\n",
    "n_hidden_2 = 800 # 2nd layer num features\n",
    "n_input = 784 # MNIST data input (img shape: 28*28)\n",
    "n_classes = 10 # MNIST total classes (0-9 digits)\n",
    "\n",
    "std_0  = np.sqrt(6.0/(n_input+n_hidden_1))\n",
    "std_h1 = np.sqrt(6.0/(n_hidden_1+n_hidden_2))\n",
    "std_h2 = np.sqrt(6.0/(n_hidden_2+n_classes))\n",
    "\n",
    "logfile = \"2-layer-dlk-alignmnist\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:22.223215",
     "start_time": "2016-06-13T11:41:22.216948"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# tf Graph input\n",
    "x = tf.placeholder(\"float\", [None, n_input])\n",
    "y = tf.placeholder(\"float\", [None, n_classes])\n",
    "\n",
    "xt = tf.placeholder(\"float\", [None, n_input])\n",
    "yt = tf.placeholder(\"float\", [None, n_classes])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:22.229905",
     "start_time": "2016-06-13T11:41:22.225062"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def multilayer_perceptron(_X, _weights, _biases):\n",
    "    layer_1 = tf.nn.relu(tf.add(tf.matmul(_X, _weights['h1']), _biases['b1'])) \n",
    "    layer_2 = tf.nn.relu(tf.add(tf.matmul(layer_1, _weights['h2']), _biases['b2'])) \n",
    "    return tf.matmul(layer_2, weights['out']) + biases['out']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:22.273322",
     "start_time": "2016-06-13T11:41:22.231294"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Store layers weight & bias\n",
    "weights = {\n",
    "    'h1': tf.Variable(tf.random_normal([n_input, n_hidden_1], stddev=std_0)),\n",
    "    'h2': tf.Variable(tf.random_normal([n_hidden_1, n_hidden_2], stddev=std_h1)),\n",
    "    'out': tf.Variable(tf.random_normal([n_hidden_2, n_classes], stddev=std_h2))\n",
    "}\n",
    "biases = {\n",
    "    'b1': tf.Variable(tf.random_normal([n_hidden_1], stddev=0.1)),\n",
    "    'b2': tf.Variable(tf.random_normal([n_hidden_2], stddev=0.01)),\n",
    "    'out': tf.Variable(tf.random_normal([n_classes], stddev=0.001))\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:22.285464",
     "start_time": "2016-06-13T11:41:22.274578"
    },
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "mlp = multilayer_perceptron(x, weights, biases )\n",
    "mlp_test = multilayer_perceptron(xt, weights, biases )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:22.300218",
     "start_time": "2016-06-13T11:41:22.287006"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "cost =  tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(mlp, y)) \n",
    "gs = tf.get_variable(\"global_step\",[],trainable=False,initializer=tf.constant_initializer(0))\n",
    "lr = tf.constant(learning_rate) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:36.705000",
     "start_time": "2016-06-13T11:41:36.565738"
    },
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "train_op = tf.contrib.layers.optimize_loss(cost, global_step=gs, learning_rate=lr,optimizer=\"RMSProp\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:36.726580",
     "start_time": "2016-06-13T11:41:36.706363"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tf.Tensor 'ScalarSummary_2:0' shape=() dtype=string>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#with tf.name_scope(\"training accuracy\"):\n",
    "pred = tf.equal(tf.argmax(mlp, 1), tf.argmax(y, 1)) # Count correct predictions\n",
    "train_acc_op = tf.reduce_mean(tf.cast(pred, \"float\"))  # Cast boolean to float to average\n",
    "tf.scalar_summary(\"training accuracy\", train_acc_op)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:36.867225",
     "start_time": "2016-06-13T11:41:36.852334"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tf.Tensor 'ScalarSummary_3:0' shape=() dtype=string>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_pred = tf.equal(tf.argmax(mlp_test, 1), tf.argmax(yt, 1)) # Count correct predictions\n",
    "test_acc_op = tf.reduce_mean(tf.cast(test_pred, \"float\"))  # Cast boolean to float to average\n",
    "tf.scalar_summary(\"test 0 accuracy\", test_acc_op)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:37.240735",
     "start_time": "2016-06-13T11:41:37.003741"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[34m2-layer-mlp-alignmnist\u001b[m\u001b[m \u001b[34m5-layer-mlp-alignmnist\u001b[m\u001b[m\r\n",
      "\u001b[34m2-layer-mlp-infimnist\u001b[m\u001b[m  \u001b[34m5-layer-mlp-infimnist\u001b[m\u001b[m\r\n"
     ]
    }
   ],
   "source": [
    "!rm -rf ./logs/{logfile}\n",
    "!ls logs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-05-04T21:57:37.534602",
     "start_time": "2016-05-04T21:57:37.532739"
    },
    "collapsed": true
   },
   "source": [
    "### Original MNIST Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:39.384250",
     "start_time": "2016-06-13T11:41:38.452026"
    },
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Extracting ./train-images-idx3-ubyte.gz\n",
      "Extracting ./train-labels-idx1-ubyte.gz\n",
      "Extracting ./t10k-images-idx3-ubyte.gz\n",
      "Extracting ./t10k-labels-idx1-ubyte.gz\n"
     ]
    }
   ],
   "source": [
    "from tensorflow.examples.tutorials.mnist import input_data\n",
    "mnist = input_data.read_data_sets(\".\", one_hot=True)\n",
    "trX_0, trY_0 = mnist.train.images, mnist.train.labels\n",
    "teX_0, teY_0 = mnist.test.images, mnist.test.labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:41:39.517621",
     "start_time": "2016-06-13T11:41:39.514257"
    },
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "alignmnist = AlignMNIST()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:47:34.431224",
     "start_time": "2016-06-13T11:41:46.812423"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(0, 0.97108334, 0.96319997)\n",
      "(1, 0.96214998, 0.97640002)\n",
      "(2, 0.97236669, 0.98009998)\n",
      "(3, 0.97483331, 0.9831)\n",
      "(4, 0.97855002, 0.98299998)\n",
      "(5, 0.98044997, 0.98479998)\n",
      "(6, 0.98294997, 0.98629999)\n",
      "(7, 0.98183334, 0.98699999)\n",
      "(8, 0.97925001, 0.98409998)\n",
      "(9, 0.98268336, 0.98629999)\n",
      "(10, 0.98040003, 0.98500001)\n",
      "(11, 0.98363334, 0.98549998)\n",
      "(12, 0.98479998, 0.98699999)\n",
      "(13, 0.98316669, 0.98619998)\n",
      "(14, 0.97815001, 0.9817)\n",
      "(15, 0.98456669, 0.98799998)\n",
      "(16, 0.98154998, 0.9842)\n",
      "(17, 0.97898334, 0.98449999)\n",
      "(18, 0.98654997, 0.98940003)\n",
      "(19, 0.98245001, 0.98439997)\n",
      "(20, 0.98091668, 0.98460001)\n",
      "(21, 0.98548335, 0.98860002)\n",
      "(22, 0.98708332, 0.98769999)\n",
      "(23, 0.98656666, 0.9892)\n",
      "(24, 0.98521668, 0.98519999)\n",
      "(25, 0.98616666, 0.98860002)\n",
      "(26, 0.97931665, 0.98390001)\n",
      "(27, 0.98733336, 0.9892)\n",
      "(28, 0.98273331, 0.98750001)\n",
      "(29, 0.97896665, 0.98439997)\n",
      "(30, 0.98436666, 0.98720002)\n",
      "(31, 0.98438334, 0.98809999)\n",
      "(32, 0.98693335, 0.98940003)\n",
      "(33, 0.98796666, 0.98940003)\n",
      "(34, 0.98711669, 0.98930001)\n",
      "(35, 0.98346668, 0.98439997)\n",
      "(36, 0.98594999, 0.98640001)\n",
      "(37, 0.98141664, 0.98409998)\n",
      "(38, 0.98643333, 0.9878)\n",
      "(39, 0.98633331, 0.9885)\n",
      "(40, 0.98839998, 0.98790002)\n",
      "(41, 0.98385, 0.98400003)\n",
      "(42, 0.98755002, 0.9892)\n",
      "(43, 0.98618335, 0.98769999)\n",
      "(44, 0.98753333, 0.9885)\n",
      "(45, 0.98326665, 0.98589998)\n",
      "(46, 0.98491669, 0.98640001)\n",
      "(47, 0.98301667, 0.98500001)\n",
      "(48, 0.98571664, 0.98699999)\n",
      "(49, 0.98563331, 0.9885)\n",
      "(50, 0.98408335, 0.98570001)\n",
      "(51, 0.98473334, 0.98860002)\n",
      "(52, 0.98675001, 0.98750001)\n",
      "(53, 0.98734999, 0.98799998)\n",
      "(54, 0.98014998, 0.9831)\n",
      "(55, 0.98483336, 0.98610002)\n",
      "(56, 0.98828334, 0.98900002)\n",
      "(57, 0.98536664, 0.9878)\n",
      "(58, 0.98659998, 0.98890001)\n",
      "(59, 0.98795003, 0.9896)\n",
      "(60, 0.98463333, 0.98680001)\n",
      "(61, 0.98760003, 0.98720002)\n",
      "(62, 0.97866666, 0.98549998)\n",
      "(63, 0.98540002, 0.98839998)\n",
      "(64, 0.98608333, 0.98830003)\n",
      "(65, 0.98636669, 0.9874)\n",
      "(66, 0.9885, 0.98890001)\n",
      "(67, 0.98696667, 0.98820001)\n",
      "(68, 0.98673332, 0.98970002)\n",
      "(69, 0.98570001, 0.98860002)\n",
      "(70, 0.98316669, 0.98659998)\n",
      "(71, 0.98798335, 0.98879999)\n",
      "(72, 0.98633331, 0.98689997)\n",
      "(73, 0.98764998, 0.9892)\n",
      "(74, 0.98833334, 0.98940003)\n"
     ]
    }
   ],
   "source": [
    "test_accuracies = []\n",
    "train_accuracies = []\n",
    "with tf.Session() as sess:\n",
    "    # create a log writer. run 'tensorboard --logdir=./logs/{logfile}'\n",
    "    writer = tf.train.SummaryWriter(\"./logs/{0}\".format(logfile), sess.graph) # for 0.8\n",
    "    merged = tf.merge_all_summaries()\n",
    " \n",
    "    tf.initialize_all_variables().run()\n",
    "\n",
    "    for epoch in range(num_epochs):\n",
    "        trX, trY = alignmnist.next_epoch()\n",
    "      \n",
    "        for start, end in zip(range(0, len(trX), batch_size), range(batch_size, len(trX), batch_size)):\n",
    "            sess.run(train_op, feed_dict={x: trX[start:end], y: trY[start:end]})\n",
    "                 \n",
    "        summary, trn_acc, tst_acc = sess.run([merged, train_acc_op, test_acc_op], feed_dict={x: trX, y: trY, xt: teX_0, yt: teY_0})\n",
    "        writer.add_summary(summary, epoch)  \n",
    "        \n",
    "        print(epoch, trn_acc, tst_acc)\n",
    "        train_accuracies.append(trn_acc)\n",
    "        test_accuracies.append(tst_acc)\n",
    "\n",
    "        writer.flush()\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:47:34.432223",
     "start_time": "2016-06-13T18:41:47.546Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import matplotlib\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:47:34.432647",
     "start_time": "2016-06-13T18:41:47.697Z"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x11ccd2c90>]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYQAAAEACAYAAACznAEdAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11ccd2cd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(train_accuracies)\n",
    "plt.plot(test_accuracies)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:47:34.433373",
     "start_time": "2016-06-13T18:41:48.203Z"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.98970002"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.max(test_accuracies)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Number test errors is"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2016-06-13T11:47:34.433755",
     "start_time": "2016-06-13T18:42:01.002Z"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "102.99980640411377"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "10000*(1.0-np.max(test_accuracies))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}