ML_Lab06_02_FancySoftmaxClassifier.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Fancy Softmax Classifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "with __cross_entropy, one_hot, reshape__ functions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Users\\iknee\\Anaconda3\\lib\\site-packages\\tensorflow\\python\\framework\\op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n",
      "WARNING:tensorflow:From <ipython-input-1-00f3571df478>:27: softmax_cross_entropy_with_logits (from tensorflow.python.ops.nn_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "\n",
      "Future major versions of TensorFlow will allow gradients to flow\n",
      "into the labels input on backprop by default.\n",
      "\n",
      "See `tf.nn.softmax_cross_entropy_with_logits_v2`.\n",
      "\n",
      "Step:     0\tLoss: 5.271\tAcc: 14.85%\n",
      "Step:   100\tLoss: 0.703\tAcc: 85.15%\n",
      "Step:   200\tLoss: 0.438\tAcc: 88.12%\n",
      "Step:   300\tLoss: 0.322\tAcc: 90.10%\n",
      "Step:   400\tLoss: 0.251\tAcc: 93.07%\n",
      "Step:   500\tLoss: 0.205\tAcc: 96.04%\n",
      "Step:   600\tLoss: 0.172\tAcc: 97.03%\n",
      "Step:   700\tLoss: 0.148\tAcc: 99.01%\n",
      "Step:   800\tLoss: 0.129\tAcc: 99.01%\n",
      "Step:   900\tLoss: 0.115\tAcc: 99.01%\n",
      "Step:  1000\tLoss: 0.104\tAcc: 99.01%\n",
      "Step:  1100\tLoss: 0.094\tAcc: 99.01%\n",
      "Step:  1200\tLoss: 0.087\tAcc: 99.01%\n",
      "Step:  1300\tLoss: 0.080\tAcc: 100.00%\n",
      "Step:  1400\tLoss: 0.074\tAcc: 100.00%\n",
      "Step:  1500\tLoss: 0.069\tAcc: 100.00%\n",
      "Step:  1600\tLoss: 0.065\tAcc: 100.00%\n",
      "Step:  1700\tLoss: 0.061\tAcc: 100.00%\n",
      "Step:  1800\tLoss: 0.058\tAcc: 100.00%\n",
      "Step:  1900\tLoss: 0.055\tAcc: 100.00%\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 3 True Y: 3\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 3 True Y: 3\n",
      "[True] Prediction: 3 True Y: 3\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 3 True Y: 3\n",
      "[True] Prediction: 6 True Y: 6\n",
      "[True] Prediction: 6 True Y: 6\n",
      "[True] Prediction: 6 True Y: 6\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 3 True Y: 3\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 5 True Y: 5\n",
      "[True] Prediction: 4 True Y: 4\n",
      "[True] Prediction: 4 True Y: 4\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 5 True Y: 5\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 3 True Y: 3\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 3 True Y: 3\n",
      "[True] Prediction: 5 True Y: 5\n",
      "[True] Prediction: 5 True Y: 5\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 5 True Y: 5\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 6 True Y: 6\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 5 True Y: 5\n",
      "[True] Prediction: 4 True Y: 4\n",
      "[True] Prediction: 6 True Y: 6\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 3 True Y: 3\n",
      "[True] Prediction: 3 True Y: 3\n",
      "[True] Prediction: 2 True Y: 2\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 6 True Y: 6\n",
      "[True] Prediction: 3 True Y: 3\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 2 True Y: 2\n",
      "[True] Prediction: 6 True Y: 6\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 2 True Y: 2\n",
      "[True] Prediction: 6 True Y: 6\n",
      "[True] Prediction: 3 True Y: 3\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 6 True Y: 6\n",
      "[True] Prediction: 3 True Y: 3\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 5 True Y: 5\n",
      "[True] Prediction: 4 True Y: 4\n",
      "[True] Prediction: 2 True Y: 2\n",
      "[True] Prediction: 2 True Y: 2\n",
      "[True] Prediction: 3 True Y: 3\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 1 True Y: 1\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 5 True Y: 5\n",
      "[True] Prediction: 0 True Y: 0\n",
      "[True] Prediction: 6 True Y: 6\n",
      "[True] Prediction: 1 True Y: 1\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "\n",
    "# Predicting animal type based on various features\n",
    "xy = np.loadtxt('data-04-zoo.csv', delimiter=',', dtype=np.float32)\n",
    "x_data = xy[:, 0:-1]\n",
    "y_data = xy[:, [-1]]\n",
    "\n",
    "nb_classes = 7  # 0~6\n",
    "\n",
    "X = tf.placeholder(tf.float32, [None, 16])\n",
    "Y = tf.placeholder(tf.int32, [None, 1])  # 0~6\n",
    "\n",
    "Y_one_hot = tf.one_hot(Y, nb_classes)\n",
    "Y_one_hot = tf.reshape(Y_one_hot, [-1, nb_classes])\n",
    "# rank가 더해짐 (차원이 더해짐). 그래서 줄인다. (-1은 '모든'의 의미)\n",
    "\n",
    "W = tf.Variable(tf.random_normal([16, nb_classes]), name='weight')\n",
    "b = tf.Variable(tf.random_normal([nb_classes]), name='bias')\n",
    "\n",
    "# softmax = exp(logits) / reduce_sum(exp(logits), dim)\n",
    "logits = tf.matmul(X, W) + b\n",
    "hypothesis = tf.nn.softmax(logits)\n",
    "\n",
    "# Cross entropy cost/loss\n",
    "cost_i = tf.nn.softmax_cross_entropy_with_logits(logits=logits,\n",
    "                                                labels=Y_one_hot)\n",
    "cost = tf.reduce_mean(cost_i)\n",
    "optimizer = tf.train.GradientDescentOptimizer(learning_rate=0.1).minimize(cost)\n",
    "\n",
    "prediction = tf.argmax(hypothesis, 1)\n",
    "correct_prediction = tf.equal(prediction, tf.argmax(Y_one_hot, 1))\n",
    "accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))\n",
    "\n",
    "# Launch graph\n",
    "with tf.Session() as sess:\n",
    "    sess.run(tf.global_variables_initializer())\n",
    "    \n",
    "    for step in range(2000):\n",
    "        sess.run(optimizer, feed_dict={X: x_data, Y: y_data})\n",
    "        if step % 100 == 0:\n",
    "            loss, acc = sess.run([cost, accuracy], feed_dict={\n",
    "                X: x_data, Y: y_data})\n",
    "            print(\"Step: {:5}\\tLoss: {:.3f}\\tAcc: {:.2%}\".format(step, loss, acc))\n",
    "            \n",
    "    pred = sess.run(prediction, feed_dict={X: x_data})\n",
    "    # y_data: (N, 1) = flatten => (N, ) matches pred.shape\n",
    "    for p, y in zip(pred, y_data.flatten()):\n",
    "        print(\"[{}] Prediction: {} True Y: {}\".format(p == int(y), p, int(y)))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}