tf-tutorial.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "1da31d44-7f5f-48c8-82f0-32e85e2bd8a5"
    }
   },
   "source": [
    "## What is Tensorflow?\n",
    "\n",
    "*TF is not a magical machine learning robot*\n",
    "\n",
    "- It has had a wide coverage in the general news media, e.g. in NYT.\n",
    "- It is merely an extensive library for performing numerical computations that are optimized towards running neural networks on GPUs as well as CPUs.\n",
    "- TF is not that different from other numerical libraries based on computational graphs, e.g. Theano."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "ab65fc5a-c5be-445c-94f0-3d4962dc0674"
    }
   },
   "source": [
    "## Computational Graphs\n",
    "\n",
    "Computational graphs can be seen as a nice way of structuring mathematical expressions wrt order of operations, declarative programming. Let us say that we have following expression:\n",
    "\n",
    "$c=\\sqrt{a^2+b^2}\\ .$\n",
    "\n",
    "This equation needs to be performed in the correct order to give the correct result. Therefore we define 3 operations, in this example with fixed values (a=3, b=4):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true,
    "nbpresent": {
     "id": "8adb5b2b-d87f-4cda-9838-9048c47c63ee"
    }
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false,
    "nbpresent": {
     "id": "7254a2d5-81ee-4540-b311-acf855602662"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor(\"Const_21:0\", shape=(), dtype=float32) Tensor(\"Const_22:0\", shape=(), dtype=float32) Tensor(\"Sqrt:0\", shape=(), dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "a = tf.constant(3.0, tf.float32)\n",
    "b = tf.constant(4.0, tf.float32)\n",
    "c = tf.sqrt(tf.add(tf.square(a), tf.square(b)))\n",
    "\n",
    "print(a, b, c)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3.0 4.0 5.0\n"
     ]
    }
   ],
   "source": [
    "sess = tf.Session()\n",
    "print(*sess.run([a, b, c]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false,
    "nbpresent": {
     "id": "3aecea89-e6a4-45c2-806e-f72ee4630b21"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor(\"Placeholder_16:0\", dtype=float32) Tensor(\"Placeholder_17:0\", dtype=float32) Tensor(\"Sqrt_1:0\", dtype=float32)\n",
      "4.0 3.0 5.0\n"
     ]
    }
   ],
   "source": [
    "a = tf.placeholder(tf.float32)\n",
    "b = tf.placeholder(tf.float32)\n",
    "c = tf.sqrt(tf.add(tf.square(a), tf.square(b)))\n",
    "\n",
    "print(a, b, c)\n",
    "\n",
    "sess = tf.Session()\n",
    "print(*sess.run([a, b, c], feed_dict={a: 4., b: 3.}))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false,
    "nbpresent": {
     "id": "24c1f5c2-cc87-49d4-bd5c-7750803371ee"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4.0 4.0 5.65685\n"
     ]
    }
   ],
   "source": [
    "print(*sess.run([a, b, c], feed_dict={a: 4., b: 4.}))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "04034e25-6c6e-463c-8398-42e6ee82e8ca"
    }
   },
   "source": [
    "### Differentiation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "cfbfa0bd-c611-48ca-814c-40a06923e76e"
    }
   },
   "source": [
    "Backpropagation is one of the most essential building blocks in deep learning. An optimal way of performing backpropagation is by using the computational graph to take the partial derivatives between variabels. This way we sum over the derivatives of the nodes in the graph.\n",
    "\n",
    "So let's consider a linear regression problem."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true,
    "nbpresent": {
     "id": "b8223ccb-9617-4391-bd2a-86834d812fea"
    }
   },
   "outputs": [],
   "source": [
    "W = tf.Variable([.3], tf.float32)\n",
    "b = tf.Variable([-.3], tf.float32)\n",
    "x = tf.placeholder(tf.float32)\n",
    "linear_model = W * x + b"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "8b9f2e50-8f8f-4e8a-8956-26f7e6810438"
    }
   },
   "source": [
    "<img src=\"graph.png\">"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true,
    "nbpresent": {
     "id": "74e3c084-3cec-4ba3-81a9-8d1819ce27e4"
    }
   },
   "outputs": [],
   "source": [
    "init = tf.global_variables_initializer()\n",
    "sess.run(init)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false,
    "nbpresent": {
     "id": "613625df-0772-4c3c-9a36-7df55cf2147a"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 0.          0.30000001  0.60000002  0.90000004]\n"
     ]
    }
   ],
   "source": [
    "print(sess.run(linear_model, {x:[1,2,3,4]}))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false,
    "nbpresent": {
     "id": "3f691c8e-5ebe-41b9-bbaa-3c5555175656"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "23.66\n"
     ]
    }
   ],
   "source": [
    "y = tf.placeholder(tf.float32)\n",
    "squared_deltas = tf.square(linear_model - y)\n",
    "loss = tf.reduce_sum(squared_deltas)\n",
    "print(sess.run(loss, {x:[1,2,3,4], y:[0,-1,-2,-3]}))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false,
    "nbpresent": {
     "id": "3c5184c0-f5c1-4936-846d-5de2bb2bad55"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0\n"
     ]
    }
   ],
   "source": [
    "fixW = tf.assign(W, [-1.])\n",
    "fixb = tf.assign(b, [1.])\n",
    "sess.run([fixW, fixb])\n",
    "print(sess.run(loss, {x:[1,2,3,4], y:[0,-1,-2,-3]}))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true,
    "nbpresent": {
     "id": "a271c69d-d8da-4452-b3dc-95c83f1098d2"
    }
   },
   "outputs": [],
   "source": [
    "optimizer = tf.train.GradientDescentOptimizer(0.01)\n",
    "train = optimizer.minimize(loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false,
    "nbpresent": {
     "id": "35d4d19b-1bb7-48c0-b19c-f3fb20caa76b"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[array([-0.9999969], dtype=float32), array([ 0.99999082], dtype=float32)]\n"
     ]
    }
   ],
   "source": [
    "sess.run(init) # reset values to incorrect defaults.\n",
    "for i in range(1000):\n",
    "  sess.run(train, {x:[1,2,3,4], y:[0,-1,-2,-3]})\n",
    "\n",
    "print(sess.run([W, b]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "nbpresent": {
     "id": "4e1e5585-ab39-4a0d-9615-cdd5b93f1391"
    }
   },
   "source": [
    "### More Complex Models?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true,
    "nbpresent": {
     "id": "9c7d4ecc-b190-4e72-bf0f-98ba2dae9598"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": false,
    "nbpresent": {
     "id": "a8804b83-55b0-4854-b377-a0bff00323b8"
    }
   },
   "outputs": [],
   "source": [
    "def onehot(t, num_classes):\n",
    "    out = np.zeros((t.shape[0], num_classes))\n",
    "    for row, col in enumerate(t):\n",
    "        out[row, col] = 1\n",
    "    return out\n",
    "\n",
    "data = np.load('mnist.npz')\n",
    "num_classes = 10\n",
    "x_train = data['X_train'][:2000].astype('float32')\n",
    "t_train = data['y_train'][:2000]\n",
    "t_train = onehot(t_train, num_classes)\n",
    "\n",
    "x_valid = data['X_valid'][:500].astype('float32')\n",
    "t_valid = data['y_valid'][:500]\n",
    "t_valid = onehot(t_valid, num_classes)\n",
    "\n",
    "x_test = data['X_test'][:500].astype('float32')\n",
    "t_test = data['y_test'][:500]\n",
    "t_test = onehot(t_test, num_classes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": false,
    "nbpresent": {
     "id": "1c0eb187-ea18-45fc-ba69-26c3a287f6f6"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1211b0160>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWoAAAFdCAYAAAAqpyf2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1211b0278>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(dpi=300)\n",
    "r = 10\n",
    "c = 10\n",
    "i = 0\n",
    "img_out = np.zeros((28 * r, 28 * c))\n",
    "for x in range(r):\n",
    "    for y in range(c):\n",
    "        im = np.reshape(x_train[i], (28, 28))\n",
    "        i += 1\n",
    "        xa, xb = x * 28, (x + 1) * 28\n",
    "        ya, yb = y * 28, (y + 1) * 28\n",
    "        img_out[xa:xb, ya:yb] = im\n",
    "\n",
    "plt.matshow(img_out, cmap=\"gray\")\n",
    "plt.axis('off')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": true,
    "nbpresent": {
     "id": "a80b611f-5992-4565-a692-2de737b10eb5"
    }
   },
   "outputs": [],
   "source": [
    "x = tf.placeholder(tf.float32, shape=[None, 784])\n",
    "y_ = tf.placeholder(tf.float32, shape=[None, 10])\n",
    "\n",
    "W = tf.Variable(tf.zeros([784, 10]))\n",
    "b = tf.Variable(tf.zeros([10]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false,
    "nbpresent": {
     "id": "518e79ed-46f7-4b86-ac26-ddecd1ee5178"
    }
   },
   "outputs": [],
   "source": [
    "sess = sess = tf.InteractiveSession()\n",
    "\n",
    "sess.run(tf.global_variables_initializer())\n",
    "\n",
    "y = tf.nn.softmax(tf.matmul(x, W) + b)\n",
    "cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y), reduction_indices=[1]))\n",
    "\n",
    "train_step = tf.train.GradientDescentOptimizer(0.001).minimize(cross_entropy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false,
    "nbpresent": {
     "id": "c92924a6-68f6-4b2a-9a9d-b935d36da40e"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test.acc=0.3340\n",
      "test.acc=0.7140\n",
      "test.acc=0.7400\n",
      "test.acc=0.7660\n",
      "test.acc=0.7900\n",
      "test.acc=0.8040\n",
      "test.acc=0.8120\n",
      "test.acc=0.8200\n",
      "test.acc=0.8280\n",
      "test.acc=0.8280\n",
      "test.acc=0.8340\n",
      "test.acc=0.8360\n",
      "test.acc=0.8400\n",
      "test.acc=0.8440\n",
      "test.acc=0.8440\n",
      "test.acc=0.8480\n",
      "test.acc=0.8500\n",
      "test.acc=0.8500\n",
      "test.acc=0.8520\n",
      "test.acc=0.8520\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "train_accuracies = []\n",
    "test_accuracies = []\n",
    "\n",
    "for i in range(20000):\n",
    "    random_indices = np.random.choice(x_train.shape[0], 100, replace=False)\n",
    "    batch = x_train[random_indices], t_train[random_indices]\n",
    "    train_step.run(feed_dict={x: batch[0], y_: batch[1]})\n",
    "    if i % 1000 == 0:\n",
    "        correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))\n",
    "        accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))\n",
    "        train_accuracies += [accuracy.eval(feed_dict={x: x_train, y_: t_train})]\n",
    "        test_accuracies += [accuracy.eval(feed_dict={x: x_test, y_: t_test})]\n",
    "        print(\"test.acc=%.4f\" % test_accuracies[-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false,
    "nbpresent": {
     "id": "a1e6cfc6-e873-40ef-8dae-2bac46c0d90c"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgkAAAFkCAYAAACq4KjhAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11e995208>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure()\n",
    "plt.plot(train_accuracies, label=\"Train\")\n",
    "plt.plot(test_accuracies, label=\"Test\")\n",
    "plt.legend(loc=0)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Higher abstractions?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [conda env:python3]",
   "language": "python",
   "name": "conda-env-python3-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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}