12. AutoEncoders.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "12. AutoEncoders.ipynb",
      "version": "0.3.2",
      "views": {},
      "default_view": {},
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "metadata": {
        "id": "0z2ORbKFgsLJ",
        "colab_type": "code",
        "colab": {
          "autoexec": {
            "startup": false,
            "wait_interval": 0
          }
        }
      },
      "cell_type": "code",
      "source": [
        "import tensorflow as tf\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "\n",
        "from tensorflow.examples.tutorials.mnist import input_data"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "oI2r1aeagsLw",
        "colab_type": "code",
        "colab": {
          "autoexec": {
            "startup": false,
            "wait_interval": 0
          },
          "output_extras": [
            {
              "item_id": 2
            }
          ],
          "base_uri": "https://localhost:8080/",
          "height": 156
        },
        "outputId": "3474f908-a79f-468b-9f88-e64d7e94c17b",
        "executionInfo": {
          "status": "ok",
          "timestamp": 1521518709491,
          "user_tz": -330,
          "elapsed": 1310,
          "user": {
            "displayName": "Mayur Kulkarni",
            "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s128",
            "userId": "116301151646646358552"
          }
        }
      },
      "cell_type": "code",
      "source": [
        "mnist = input_data.read_data_sets(\".\", one_hot=True)"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Successfully downloaded train-images-idx3-ubyte.gz 9912422 bytes.\n",
            "Extracting ./train-images-idx3-ubyte.gz\n",
            "Successfully downloaded train-labels-idx1-ubyte.gz 28881 bytes.\n",
            "Extracting ./train-labels-idx1-ubyte.gz\n",
            "Successfully downloaded t10k-images-idx3-ubyte.gz 1648877 bytes.\n",
            "Extracting ./t10k-images-idx3-ubyte.gz\n",
            "Successfully downloaded t10k-labels-idx1-ubyte.gz 4542 bytes.\n",
            "Extracting ./t10k-labels-idx1-ubyte.gz\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "tpQ_Mao9gsMY",
        "colab_type": "code",
        "colab": {
          "autoexec": {
            "startup": false,
            "wait_interval": 0
          },
          "output_extras": [
            {
              "item_id": 1
            },
            {
              "item_id": 2
            }
          ],
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "outputId": "6ef2f1d6-b9c9-40dd-d2a1-388e10015bf1",
        "executionInfo": {
          "status": "ok",
          "timestamp": 1521518710390,
          "user_tz": -330,
          "elapsed": 730,
          "user": {
            "displayName": "Mayur Kulkarni",
            "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s128",
            "userId": "116301151646646358552"
          }
        }
      },
      "cell_type": "code",
      "source": [
        "# show a random image from MNIST dataset\n",
        "plt.imshow(mnist.train.next_batch(1)[0].reshape(28, 28), cmap=\"gist_gray\")"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f1577268e80>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPoAAAD4CAYAAADFJPs2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f157771c4a8>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "YWHM2wJYgsM4",
        "colab_type": "code",
        "colab": {
          "autoexec": {
            "startup": false,
            "wait_interval": 0
          }
        }
      },
      "cell_type": "code",
      "source": [
        "class AutoEncoder:\n",
        "    \"\"\"Creates an AutoEncoder for compressing MNIST\n",
        "    dataset\n",
        "    \n",
        "    Architecture\n",
        "    ------------\n",
        "    Image (28x28) \n",
        "       -- Input Layer(1x784) -- Dense Layer(784x100) -- Output Layer(100x784)\n",
        "    Loss Function: Mean Squared Error\n",
        "    Optimizer: Adam Optimizer\n",
        "    \"\"\"\n",
        "    def __init__(self):\n",
        "        self.num_input = 28*28\n",
        "        self.num_hidden_1 = 100\n",
        "        self.num_output = self.num_input\n",
        "        self.X = tf.placeholder(tf.float32, \n",
        "                                shape=[None, self.num_input])\n",
        "    \n",
        "    def forward_pass(self):\n",
        "        \"\"\" Make a forward pass in the Neural Network\"\"\"\n",
        "        W1 = tf.Variable(initial_value=\n",
        "                         tf.truncated_normal([self.num_input, \n",
        "                                              self.num_hidden_1]))\n",
        "        W2 = tf.transpose(W1)\n",
        "        b1 = tf.Variable(initial_value=tf.random_normal([self.num_hidden_1]))\n",
        "        b2 = tf.Variable(initial_value=tf.random_normal([self.num_output]))\n",
        "        \n",
        "        hidden_output = tf.matmul(self.X, W1) + b1\n",
        "        final_output = tf.matmul(hidden_output, W2) + b2\n",
        "        return final_output\n",
        "    \n",
        "    def train(self):\n",
        "        \"\"\"Trains the Network\"\"\"\n",
        "        y_pred = self.forward_pass()\n",
        "        error = tf.losses.mean_squared_error(labels=self.X, \n",
        "                                             predictions=y_pred)\n",
        "        optimizer = tf.train.AdamOptimizer(learning_rate=0.01)\n",
        "        train = optimizer.minimize(error)\n",
        "        init = tf.global_variables_initializer()\n",
        "        \n",
        "        with tf.Session() as sess:\n",
        "            sess.run(init)\n",
        "            for i in range(20000):\n",
        "                X_batch, y_batch = mnist.train.next_batch(50)\n",
        "                sess.run(train, feed_dict={self.X: X_batch})\n",
        "                loss = sess.run(error, feed_dict={self.X: X_batch})\n",
        "                if i % 2000 == 0:\n",
        "                    print(\"Iteration: {} Loss {}\".format(i, loss))\n",
        "            X_batch = mnist.test.next_batch(1)[0]\n",
        "            plt.imshow(X_batch.reshape(28, 28), cmap=\"gist_gray\")\n",
        "            plt.show()\n",
        "            plt.imshow(sess.run(y_pred,\n",
        "                                feed_dict={self.X: X_batch}).reshape(28, 28),\n",
        "                                cmap=\"gist_gray\")\n",
        "            plt.show()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "NC4UQl78lTAY",
        "colab_type": "code",
        "colab": {
          "autoexec": {
            "startup": false,
            "wait_interval": 0
          },
          "output_extras": [
            {
              "item_id": 10
            },
            {
              "item_id": 11
            },
            {
              "item_id": 12
            }
          ],
          "base_uri": "https://localhost:8080/",
          "height": 687
        },
        "outputId": "f35c5be9-3419-4f54-a643-a492a53af1db",
        "executionInfo": {
          "status": "ok",
          "timestamp": 1521519087402,
          "user_tz": -330,
          "elapsed": 49407,
          "user": {
            "displayName": "Mayur Kulkarni",
            "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s128",
            "userId": "116301151646646358552"
          }
        }
      },
      "cell_type": "code",
      "source": [
        "AutoEncoder().train()"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Iteration: 0 Loss 5440.36572265625\n",
            "Iteration: 2000 Loss 4.119136810302734\n",
            "Iteration: 4000 Loss 0.5287615656852722\n",
            "Iteration: 6000 Loss 0.1382046639919281\n",
            "Iteration: 8000 Loss 0.06375974416732788\n",
            "Iteration: 10000 Loss 0.021493952721357346\n",
            "Iteration: 12000 Loss 0.010436270385980606\n",
            "Iteration: 14000 Loss 0.006641633342951536\n",
            "Iteration: 16000 Loss 0.006758757401257753\n",
            "Iteration: 18000 Loss 0.0076367505826056\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPoAAAD4CAYAAADFJPs2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAU940UyAtYdIAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f1570432f98>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPoAAAD4CAYAAADFJPs2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAElFTkSuQmCC\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f156e1a2e80>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "frJCLqcjlVHs",
        "colab_type": "code",
        "colab": {
          "autoexec": {
            "startup": false,
            "wait_interval": 0
          }
        }
      },
      "cell_type": "code",
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}