fizzbuzz w/ maxout dense

fizzbuzz.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "import keras\n",
    "from keras.layers import Dense, MaxoutDense\n",
    "from keras.models import Sequential\n",
    "from keras.utils import np_utils\n",
    "from __future__ import print_function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "N = 10000\n",
    "\n",
    "def _encode(num):\n",
    "    digits = [np_utils.to_categorical(c, 10) for c in '%04d' % num]\n",
    "    return np.hstack(digits)[0]\n",
    "        \n",
    "\n",
    "def _fizzdata():\n",
    "    while True:\n",
    "        batch_x = []\n",
    "        batch_y = []\n",
    "\n",
    "        idx = np.arange(N)\n",
    "        np.random.shuffle(idx)\n",
    "        for i in idx:\n",
    "            batch_x.append(_encode(i))\n",
    "            if   i % 15 == 0: batch_y.append(np.asarray([0, 0, 0, 1]))\n",
    "            elif i % 5  == 0: batch_y.append(np.asarray([0, 0, 1, 0]))\n",
    "            elif i % 3  == 0: batch_y.append(np.asarray([0, 1, 0, 0]))\n",
    "            else: batch_y.append(np.asarray([1, 0, 0, 0]))\n",
    "\n",
    "            if len(batch_x) >= 32:\n",
    "                yield np.asarray(batch_x), np.asarray(batch_y)\n",
    "                batch_x = []\n",
    "                batch_y = []\n",
    "\n",
    "        yield np.asarray(batch_x), np.asarray(batch_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "def model():\n",
    "    m = Sequential()\n",
    "    m.add(MaxoutDense(20, input_shape=(40,)))\n",
    "    m.add(Dense(4, activation='softmax'))\n",
    "    m.compile(loss='categorical_crossentropy', optimizer='rmsprop', metrics=['accuracy'])\n",
    "    return m\n",
    "\n",
    "training_model = model()\n",
    "training_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "for i in range(500):\n",
    "    training_model.fit_generator(_fizzdata(), samples_per_epoch=N, verbose=1 if i % 10 == 0 else 5, nb_epoch=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.5.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}