Bad but fast conv example

bad_but_fast_conv_example.ipynb

{
  "cells": [
    {
      "metadata": {},
      "cell_type": "markdown",
      "source": "Bad but fast conv example\n=====\nbased on [this example](https://github.com/aymericdamien/TensorFlow-Examples/blob/master/examples/3_NeuralNetworks/convolutional_network.py)"
    },
    {
      "metadata": {
        "scrolled": true,
        "collapsed": false,
        "trusted": true
      },
      "cell_type": "code",
      "source": "from __future__ import print_function\n\nimport tensorflow as tf\nfrom tensorflow.python.framework import graph_util\nfrom tensorflow.python.platform import gfile\n\n# Import MNIST data\nfrom tensorflow.examples.tutorials.mnist import input_data\nmnist = input_data.read_data_sets(\"/tmp/data/\", one_hot=True)\n\n# Parameters\nlearning_rate = 0.005\ntraining_iters = 5000\nbatch_size = 128\ndisplay_step = 10\n\n# Network Parameters\nn_input = 784 # MNIST data input (img shape: 28*28)\nn_classes = 10 # MNIST total classes (0-9 digits)\ndropout = 0.75 # Dropout, probability to keep units\n\n# tf Graph input\nx = tf.placeholder(tf.float32, [None, n_input], name='input')\ny = tf.placeholder(tf.float32, [None, n_classes])\nkeep_prob = tf.placeholder(tf.float32, name='keep_prob') #dropout (keep probability)\n\n# Create some wrappers for simplicity\ndef conv2d(x, W, b, strides=1):\n    # Conv2D wrapper, with bias and relu activation\n    x = tf.nn.conv2d(x, W, strides=[1, strides, strides, 1], padding='SAME')\n    x = tf.nn.bias_add(x, b)\n    return tf.nn.relu(x)\n\n\ndef maxpool2d(x, k=2):\n    # MaxPool2D wrapper\n    return tf.nn.max_pool(x, ksize=[1, k, k, 1], strides=[1, k, k, 1],\n                          padding='SAME')\n\n\n# Create model\ndef conv_net(x, weights, biases, dropout):\n    # Reshape input picture\n    x = tf.reshape(x, shape=[-1, 28, 28, 1])\n\n    # Convolution Layer\n    conv1 = conv2d(x, weights['wc1'], biases['bc1'])\n    # Max Pooling (down-sampling)\n    conv1 = maxpool2d(conv1, k=2)\n\n    # Convolution Layer\n    conv2 = conv2d(conv1, weights['wc2'], biases['bc2'])\n    # Max Pooling (down-sampling)\n    conv2 = maxpool2d(conv2, k=2)\n\n    # Fully connected layer\n    # Reshape conv2 output to fit fully connected layer input\n    fc1 = tf.reshape(conv2, [-1, weights['wd1'].get_shape().as_list()[0]])\n    fc1 = tf.add(tf.matmul(fc1, weights['wd1']), biases['bd1'])\n    fc1 = tf.nn.relu(fc1)\n    # Apply Dropout\n    fc1 = tf.nn.dropout(fc1, dropout)\n\n    # Output, class prediction\n    out = tf.add(tf.matmul(fc1, weights['out']), biases['out'])\n    return out\n\n# Store layers weight & bias\nweights = {\n    # 5x5 conv, 1 input, 16 outputs\n    'wc1': tf.Variable(tf.random_normal([5, 5, 1, 16])),\n    # 5x5 conv, 16 inputs, 32 outputs\n    'wc2': tf.Variable(tf.random_normal([5, 5, 16, 32])),\n    # fully connected, 7*7*32 inputs, 256 outputs\n    'wd1': tf.Variable(tf.random_normal([7*7*32, 256])),\n    # 256 inputs, 10 outputs (class prediction)\n    'out': tf.Variable(tf.random_normal([256, n_classes]))\n}\n\nbiases = {\n    'bc1': tf.Variable(tf.random_normal([16])),\n    'bc2': tf.Variable(tf.random_normal([32])),\n    'bd1': tf.Variable(tf.random_normal([256])),\n    'out': tf.Variable(tf.random_normal([n_classes]))\n}\n\n# Construct model\npred = conv_net(x, weights, biases, keep_prob)\nfinal_tensor = tf.nn.softmax(pred, name='final_result')\n\n# Define loss and optimizer\ncost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=pred, labels=y))\noptimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost)\n\n# Evaluate model\ncorrect_pred = tf.equal(tf.argmax(pred, 1), tf.argmax(y, 1), name='correct_pred')\naccuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))\n\n# Initializing the variables\ninit = tf.global_variables_initializer()\n\n# Launch the graph\nwith tf.Session() as sess:\n    sess.run(init)\n    step = 1\n    # Keep training until reach max iterations\n    while step * batch_size < training_iters:\n        batch_x, batch_y = mnist.train.next_batch(batch_size)\n        # Run optimization op (backprop)\n        sess.run(optimizer, feed_dict={x: batch_x, y: batch_y,\n                                       keep_prob: dropout})\n        if step % display_step == 0:\n            # Calculate batch loss and accuracy\n            loss, acc = sess.run([cost, accuracy], feed_dict={x: batch_x,\n                                                              y: batch_y,\n                                                              keep_prob: 1.})\n            print(\"Iter \" + str(step*batch_size) + \", Minibatch Loss= \" + \\\n                  \"{:.6f}\".format(loss) + \", Training Accuracy= \" + \\\n                  \"{:.5f}\".format(acc))\n        step += 1\n    print(\"Optimization Finished!\")\n\n    # Calculate accuracy for 256 mnist test images\n    print(\"Testing Accuracy:\", \\\n        sess.run(accuracy, feed_dict={x: mnist.test.images[:256],\n                                      y: mnist.test.labels[:256],\n                                      keep_prob: 1.}))\n    \n    output_graph_def = graph_util.convert_variables_to_constants(\n        sess, sess.graph.as_graph_def(), ['final_result'])\n\n    with gfile.FastGFile('./frozen_model.pb', 'wb') as f:\n        f.write(output_graph_def.SerializeToString())",
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": "Extracting /tmp/data/train-images-idx3-ubyte.gz\nExtracting /tmp/data/train-labels-idx1-ubyte.gz\nExtracting /tmp/data/t10k-images-idx3-ubyte.gz\nExtracting /tmp/data/t10k-labels-idx1-ubyte.gz\nIter 1280, Minibatch Loss= 3652.552490, Training Accuracy= 0.36719\nIter 2560, Minibatch Loss= 1037.778564, Training Accuracy= 0.67188\nIter 3840, Minibatch Loss= 839.923706, Training Accuracy= 0.73438\nOptimization Finished!\nTesting Accuracy: 0.808594\nINFO:tensorflow:Froze 28 variables.\nConverted 8 variables to const ops.\n"
        }
      ]
    },
    {
      "metadata": {
        "collapsed": true,
        "trusted": true
      },
      "cell_type": "code",
      "source": "",
      "execution_count": null,
      "outputs": []
    }
  ],
  "metadata": {
    "kernelspec": {
      "name": "python2",
      "display_name": "Python 2",
      "language": "python"
    },
    "language_info": {
      "mimetype": "text/x-python",
      "nbconvert_exporter": "python",
      "name": "python",
      "pygments_lexer": "ipython2",
      "version": "2.7.12",
      "file_extension": ".py",
      "codemirror_mode": {
        "version": 2,
        "name": "ipython"
      }
    },
    "gist": {
      "id": "",
      "data": {
        "description": "Bad but fast conv example",
        "public": true
      }
    }
  },
  "nbformat": 4,
  "nbformat_minor": 2
}