RNN-Classification.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['data/names\\\\Arabic.txt', 'data/names\\\\Chinese.txt', 'data/names\\\\Czech.txt', 'data/names\\\\Dutch.txt', 'data/names\\\\English.txt', 'data/names\\\\French.txt', 'data/names\\\\German.txt', 'data/names\\\\Greek.txt', 'data/names\\\\Irish.txt', 'data/names\\\\Italian.txt', 'data/names\\\\Japanese.txt', 'data/names\\\\Korean.txt', 'data/names\\\\Polish.txt', 'data/names\\\\Portuguese.txt', 'data/names\\\\Russian.txt', 'data/names\\\\Scottish.txt', 'data/names\\\\Spanish.txt', 'data/names\\\\Turkish.txt', 'data/names\\\\Vietnamese.txt']\n"
     ]
    }
   ],
   "source": [
    "import glob\n",
    "\n",
    "all_filenames = glob.glob('data/names/*.txt')\n",
    "print(all_filenames)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Slusarski\n"
     ]
    }
   ],
   "source": [
    "import unicodedata\n",
    "import string\n",
    "\n",
    "all_letters = string.ascii_letters + \" .,;'\"\n",
    "n_letters = len(all_letters)\n",
    "\n",
    "# Turn a Unicode string to plain ASCII, thanks to http://stackoverflow.com/a/518232/2809427\n",
    "def unicode_to_ascii(s):\n",
    "    return ''.join(\n",
    "        c for c in unicodedata.normalize('NFD', s)\n",
    "        if unicodedata.category(c) != 'Mn'\n",
    "        and c in all_letters\n",
    "    )\n",
    "\n",
    "print(unicode_to_ascii('Ślusàrski'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "n_categories = 19\n"
     ]
    }
   ],
   "source": [
    "category_lines = {}\n",
    "all_categories = []\n",
    "\n",
    "# Read a file and split into lines\n",
    "def readLines(filename):\n",
    "    lines = open(filename, encoding=\"ISO-8859-1\").read().strip().split('\\n')\n",
    "    return [unicode_to_ascii(line) for line in lines]\n",
    "\n",
    "for filename in all_filenames:\n",
    "    category = filename.split('/')[-1].split('.')[0]\n",
    "    all_categories.append(category)\n",
    "    lines = readLines(filename)\n",
    "    category_lines[category] = lines\n",
    "\n",
    "n_categories = len(all_categories)\n",
    "print('n_categories =', n_categories) \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "\n",
    "def letter_to_tensor(letter):\n",
    "    tensor = torch.zeros(1, n_letters)\n",
    "    letter_index = all_letters.find(letter)\n",
    "    tensor[0][letter_index] = 1\n",
    "    return tensor\n",
    "\n",
    "def line_to_tensor(line):\n",
    "    tensor = torch.zeros(len(line), 1, n_letters)\n",
    "    for li, letter in enumerate(line):\n",
    "        letter_index = all_letters.find(letter)\n",
    "        tensor[li][0][letter_index] = 1\n",
    "    return  tensor    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "Columns 0 to 12 \n",
      "    0     0     0     0     0     0     0     0     0     0     0     0     0\n",
      "\n",
      "Columns 13 to 25 \n",
      "    0     0     0     0     0     0     0     0     0     0     0     0     0\n",
      "\n",
      "Columns 26 to 38 \n",
      "    0     0     0     0     0     0     0     0     0     1     0     0     0\n",
      "\n",
      "Columns 39 to 51 \n",
      "    0     0     0     0     0     0     0     0     0     0     0     0     0\n",
      "\n",
      "Columns 52 to 56 \n",
      "    0     0     0     0     0\n",
      "[torch.FloatTensor of size 1x57]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(letter_to_tensor('J'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([6, 1, 57])\n"
     ]
    }
   ],
   "source": [
    "print(line_to_tensor('Furkan').size())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import torch.nn as nn\n",
    "from torch.autograd import Variable\n",
    "\n",
    "class RNN(nn.Module):\n",
    "    def __init__(self, input_size, hidden_size, output_size):\n",
    "        super(RNN, self).__init__()\n",
    "        \n",
    "        self.input_size = input_size\n",
    "        self.hidden_size = hidden_size\n",
    "        self.output_size = output_size\n",
    "        \n",
    "        self.i2h = nn.Linear(input_size + hidden_size, hidden_size)\n",
    "        self.i2o = nn.Linear(input_size + hidden_size, output_size)\n",
    "        self.softmax = nn.LogSoftmax()\n",
    "        \n",
    "    def forward(self, input, hidden):\n",
    "        combined = torch.cat((input, hidden), 1)\n",
    "        hidden = self.i2h(combined)\n",
    "        output = self.i2o(combined)\n",
    "        output = self.softmax(output)\n",
    "        return output, hidden\n",
    "    \n",
    "    def init_hidden(self):\n",
    "        return Variable(torch.zeros(1, self.hidden_size))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "n_hidden = 128\n",
    "rnn = RNN(n_letters, n_hidden, n_categories)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "output.size = torch.Size([1, 19])\n"
     ]
    }
   ],
   "source": [
    "input = Variable(letter_to_tensor('A'))\n",
    "hidden = rnn.init_hidden()\n",
    "\n",
    "output, next_hidden = rnn(input, hidden)\n",
    "print('output.size =', output.size())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Variable containing:\n",
      "\n",
      "Columns 0 to 9 \n",
      "-2.8793 -2.9766 -2.9214 -2.8627 -3.0481 -2.9694 -3.0046 -2.8606 -3.0026 -2.9232\n",
      "\n",
      "Columns 10 to 18 \n",
      "-2.9396 -2.9808 -3.0050 -2.8690 -3.0083 -2.9856 -2.8857 -2.9522 -2.8996\n",
      "[torch.FloatTensor of size 1x19]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "input =  Variable(line_to_tensor('Furkan'))\n",
    "hidden = Variable(torch.zeros(1, n_hidden))\n",
    "\n",
    "output, next_hidden = rnn(input[0], hidden)\n",
    "print(output)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('names\\\\Greek', 7)\n"
     ]
    }
   ],
   "source": [
    "def category_from_output(output):\n",
    "    top_n, top_i = output.data.topk(1)\n",
    "    category_i = top_i[0][0]\n",
    "    return all_categories[category_i], category_i\n",
    "\n",
    "print(category_from_output(output))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "category = names\\Vietnamese / line = Tieu\n",
      "category = names\\Russian / line = Billert\n",
      "category = names\\Chinese / line = Sze \n",
      "category = names\\Dutch / line = Vann\n",
      "category = names\\Scottish / line = Mclean\n",
      "category = names\\Scottish / line = Martin\n",
      "category = names\\Dutch / line = Asch\n",
      "category = names\\Japanese / line = Ishiyama\n",
      "category = names\\Polish / line = Walentowicz\n",
      "category = names\\French / line = SAverin\n"
     ]
    }
   ],
   "source": [
    "import random\n",
    "\n",
    "def random_training_pair():\n",
    "    category = random.choice(all_categories)\n",
    "    line = random.choice(category_lines[category])\n",
    "    category_tensor = Variable(torch.LongTensor([all_categories.index(category)]))\n",
    "    line_tensor = Variable(line_to_tensor(line))\n",
    "    return category, line, category_tensor, line_tensor\n",
    "\n",
    "for i in range(10):\n",
    "    category, line, category_tensor, line_tensor = random_training_pair()\n",
    "    print('category =', category, '/ line =', line)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "criterion = nn.NLLLoss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "learning_rate = 0.005\n",
    "optimizer = torch.optim.SGD(rnn.parameters(), lr=learning_rate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def train(category_tensor, line_tensor):\n",
    "    rnn.zero_grad()\n",
    "    hidden = rnn.init_hidden()\n",
    "    \n",
    "    for i in range(line_tensor.size()[0]):\n",
    "        output, hidden = rnn(line_tensor[i], hidden)\n",
    "        \n",
    "    loss = criterion(output, category_tensor)\n",
    "    loss.backward()\n",
    "    \n",
    "    optimizer.step()\n",
    "    \n",
    "    return output, loss.data[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5000 5% (0m 15s) 2.7477 Zoucha / names\\Japanese ✗ (names\\Czech)\n",
      "10000 10% (0m 30s) 2.1466 Meral / names\\Turkish ✓\n",
      "15000 15% (0m 46s) 0.5113 Malihoudis / names\\Greek ✓\n",
      "20000 20% (1m 1s) 3.1735 Jans / names\\Vietnamese ✗ (names\\German)\n",
      "25000 25% (1m 16s) 2.4844 Hutchings / names\\Greek ✗ (names\\English)\n",
      "30000 30% (1m 31s) 0.8191 Ruvelas / names\\Greek ✓\n",
      "35000 35% (1m 47s) 1.2930 Matjeka / names\\Polish ✗ (names\\Czech)\n",
      "40000 40% (2m 2s) 1.8819 YAksel / names\\Polish ✗ (names\\Turkish)\n",
      "45000 45% (2m 18s) 2.5236 Olmos / names\\Greek ✗ (names\\Spanish)\n",
      "50000 50% (2m 33s) 3.9071 Jimenez / names\\German ✗ (names\\Spanish)\n",
      "55000 55% (2m 48s) 3.4688 Hiorvst / names\\German ✗ (names\\Czech)\n",
      "60000 60% (3m 4s) 0.8183 Zellweger / names\\German ✓\n",
      "65000 65% (3m 19s) 1.7109 Adriyanoff / names\\Italian ✗ (names\\Russian)\n",
      "70000 70% (3m 35s) 0.3265 Mustafa / names\\Arabic ✓\n",
      "75000 75% (3m 50s) 2.2958 Hassel / names\\Arabic ✗ (names\\Dutch)\n",
      "80000 80% (4m 6s) 0.1797 Mesut / names\\Turkish ✓\n",
      "85000 85% (4m 21s) 0.7290 Traverse / names\\French ✓\n",
      "90000 90% (4m 38s) 0.9691 Zavala / names\\Spanish ✓\n",
      "95000 95% (4m 55s) 0.3484 Chikanatsu / names\\Japanese ✓\n",
      "100000 100% (5m 10s) 3.2205 Rosenberg / names\\Scottish ✗ (names\\German)\n"
     ]
    }
   ],
   "source": [
    "import time\n",
    "import math\n",
    "\n",
    "n_epochs = 100000\n",
    "print_every = 5000\n",
    "plot_every = 1000\n",
    "\n",
    "current_loss = 0\n",
    "all_losses = []\n",
    "\n",
    "def time_since(since):\n",
    "    now = time.time()\n",
    "    s = now - since\n",
    "    m = math.floor(s / 60)\n",
    "    s -= m * 60\n",
    "    return '%dm %ds' % (m, s)\n",
    "\n",
    "start = time.time()\n",
    "\n",
    "for epoch in range(1, n_epochs + 1):\n",
    "    category, line, category_tensor, line_tensor = random_training_pair()\n",
    "    output, loss = train(category_tensor, line_tensor)\n",
    "    current_loss += loss\n",
    "    \n",
    "    if epoch % print_every == 0:\n",
    "        guess, guess_i = category_from_output(output)\n",
    "        correct =  '✓' if guess == category else '✗ (%s)' % category\n",
    "        print('%d %d%% (%s) %.4f %s / %s %s' % (epoch, epoch / n_epochs * 100, time_since(start), loss, line, guess, correct))\n",
    "        \n",
    "    if epoch % plot_every == 0:\n",
    "        all_losses.append(current_loss / plot_every)\n",
    "        current_loss = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2807123b198>]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD8CAYAAACMwORRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\npVyQhrtSSrkgDXellHJBGu5KKeWC/j/vilba5oLp/AAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x28070e07b38>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.ticker as ticker\n",
    "%matplotlib inline\n",
    "\n",
    "plt.figure()\n",
    "plt.plot(all_losses)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYEAAAFVCAYAAAAJ9lMdAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ncBzHGcN4J+A4jjOG8U7AcRxnDPP/8lE7V/sbZO4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x280711e0588>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "confusion = torch.zeros(n_categories, n_categories)\n",
    "n_confusion = 10000\n",
    "\n",
    "def evaluate(line_tensor):\n",
    "    hidden = rnn.init_hidden()\n",
    "    \n",
    "    for i in range(line_tensor.size()[0]):\n",
    "        output, hidden = rnn(line_tensor[i], hidden)\n",
    "    \n",
    "    return output\n",
    "\n",
    "for i in range(n_confusion):\n",
    "    category, line, category_tensor, line_tensor = random_training_pair()\n",
    "    output = evaluate(line_tensor)\n",
    "    guess, guess_i = category_from_output(output)\n",
    "    category_i = all_categories.index(category)\n",
    "    confusion[category_i][guess_i] += 1\n",
    "\n",
    "for i in range(n_categories):\n",
    "    confusion[i] = confusion[i] / confusion[i].sum()\n",
    "\n",
    "fig = plt.figure()\n",
    "ax = fig.add_subplot(111)\n",
    "cax = ax.matshow(confusion.numpy())\n",
    "fig.colorbar(cax)\n",
    "\n",
    "\n",
    "ax.set_xticklabels([''] + all_categories, rotation=90)\n",
    "ax.set_yticklabels([''] + all_categories)\n",
    "\n",
    "ax.xaxis.set_major_locator(ticker.MultipleLocator(1))\n",
    "ax.yaxis.set_major_locator(ticker.MultipleLocator(1))\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "> Hilmi\n",
      "(-1.05) names\\Turkish\n",
      "(-1.29) names\\Italian\n",
      "(-2.28) names\\German\n",
      "\n",
      "> Jack\n",
      "(-1.58) names\\Korean\n",
      "(-1.86) names\\Czech\n",
      "(-1.87) names\\Polish\n",
      "\n",
      "> Cho\n",
      "(-0.64) names\\Vietnamese\n",
      "(-1.37) names\\Korean\n",
      "(-1.82) names\\Chinese\n"
     ]
    }
   ],
   "source": [
    "def predict(input_line, n_predictions=3):\n",
    "    print('\\n> %s' % input_line)\n",
    "    output = evaluate(Variable(line_to_tensor(input_line)))\n",
    "    \n",
    "    topv, topi = output.data.topk(n_predictions, 1, True)\n",
    "    predictions = []\n",
    "    \n",
    "    for i in range(n_predictions):\n",
    "        value = topv[0][i]\n",
    "        category_index = topi[0][i]\n",
    "        print('(%.2f) %s' % (value, all_categories[category_index]))\n",
    "        predictions.append([value, all_categories[category_index]])\n",
    "    \n",
    "predict('Hilmi')\n",
    "predict('Jack')\n",
    "predict('Cho')"
   ]
  }
 ],
 "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}