GAN[pytorch].ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GeForce GTX 1080 Ti\n",
      "True\n",
      "cuda:2\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "6"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "import torch\n",
    "import torch.nn as nn \n",
    "import torch.optim as optim \n",
    "from torch.autograd import Variable \n",
    "from IPython.display import clear_output\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "print(torch.cuda.get_device_name(2))\n",
    "\n",
    "print(torch.cuda.is_available())\n",
    "\n",
    "device = torch.device('cuda:2' if torch.cuda.is_available() else 'cpu')\n",
    "print(device)\n",
    "torch.cuda.device_count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_mean = 4\n",
    "data_stddev = 1.25"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "(name, preprocess, d_input_func) = (\"Only 4 moments\", lambda data: get_moments(data), lambda x: 4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using data [Only 4 moments]\n"
     ]
    }
   ],
   "source": [
    "print(\"Using data [%s]\" % (name))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_distribution_sampler(mu, sigma):\n",
    "    return lambda n: torch.Tensor(np.random.normal(mu, sigma, (1, n))).to(device)  # Gaussian\n",
    "\n",
    "def get_generator_input_sampler():\n",
    "    return lambda m, n: torch.rand(m, n).to(device)  # Uniform-dist data into generator, _NOT_ Gaussian\n",
    "\n",
    "# ##### MODELS: Generator model and discriminator model\n",
    "\n",
    "class Generator(nn.Module):\n",
    "    def __init__(self, input_size, hidden_size, output_size, f):\n",
    "        super(Generator, self).__init__()\n",
    "        self.map1 = nn.Linear(input_size, hidden_size)\n",
    "        self.map2 = nn.Linear(hidden_size, hidden_size)\n",
    "        self.map3 = nn.Linear(hidden_size, output_size)\n",
    "        self.f = f\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.map1(x)\n",
    "        x = self.f(x)\n",
    "        x = self.map2(x)\n",
    "        x = self.f(x)\n",
    "        x = self.map3(x)\n",
    "        return x\n",
    "\n",
    "class Discriminator(nn.Module):\n",
    "    def __init__(self, input_size, hidden_size, output_size, f):\n",
    "        super(Discriminator, self).__init__()\n",
    "        self.map1 = nn.Linear(input_size, hidden_size)\n",
    "        self.map2 = nn.Linear(hidden_size, hidden_size)\n",
    "        self.map3 = nn.Linear(hidden_size, output_size)\n",
    "        self.f = f\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.f(self.map1(x))\n",
    "        x = self.f(self.map2(x))\n",
    "        return self.f(self.map3(x))\n",
    "\n",
    "def extract(v):\n",
    "    return v.data.storage().tolist()\n",
    "\n",
    "def stats(d):\n",
    "    return [np.mean(d), np.std(d)]\n",
    "\n",
    "def get_moments(d):\n",
    "    # Return the first 4 moments of the data provided\n",
    "    mean = torch.mean(d)\n",
    "    diffs = d - mean\n",
    "    var = torch.mean(torch.pow(diffs, 2.0))\n",
    "    std = torch.pow(var, 0.5)\n",
    "    zscores = diffs / std\n",
    "    skews = torch.mean(torch.pow(zscores, 3.0))\n",
    "    kurtoses = torch.mean(torch.pow(zscores, 4.0)) - 3.0  # excess kurtosis, should be 0 for Gaussian\n",
    "    final = torch.cat((mean.reshape(1,), std.reshape(1,), skews.reshape(1,), kurtoses.reshape(1,)))\n",
    "    return final\n",
    "\n",
    "def decorate_with_diffs(data, exponent, remove_raw_data=False):\n",
    "    mean = torch.mean(data.data, 1, keepdim=True)\n",
    "    mean_broadcast = torch.mul(torch.ones(data.size()), mean.tolist()[0][0])\n",
    "    diffs = torch.pow(data - Variable(mean_broadcast), exponent)\n",
    "    if remove_raw_data:\n",
    "        return torch.cat([diffs], 1)\n",
    "    else:\n",
    "        return torch.cat([data, diffs], 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Plotting the generated distribution...\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "def train():\n",
    "    # Model parameters\n",
    "    g_input_size = 1      # Random noise dimension coming into generator, per output vector\n",
    "    g_hidden_size = 5     # Generator complexity\n",
    "    g_output_size = 1     # Size of generated output vector\n",
    "    d_input_size = 500    # Minibatch size - cardinality of distributions\n",
    "    d_hidden_size = 10    # Discriminator complexity\n",
    "    d_output_size = 1     # Single dimension for 'real' vs. 'fake' classification\n",
    "    minibatch_size = d_input_size\n",
    "\n",
    "    d_learning_rate = 1e-3\n",
    "    g_learning_rate = 1e-3\n",
    "    sgd_momentum = 0.9\n",
    "\n",
    "    num_epochs = 5000\n",
    "    print_interval = 100\n",
    "    d_steps = 20\n",
    "    g_steps = 20\n",
    "\n",
    "    dfe, dre, ge = 0, 0, 0\n",
    "    d_real_data, d_fake_data, g_fake_data = None, None, None\n",
    "\n",
    "    discriminator_activation_function = torch.sigmoid\n",
    "    generator_activation_function = torch.tanh\n",
    "\n",
    "    d_sampler = get_distribution_sampler(data_mean, data_stddev)\n",
    "    gi_sampler = get_generator_input_sampler()\n",
    "    G = Generator(input_size=g_input_size,\n",
    "                  hidden_size=g_hidden_size,\n",
    "                  output_size=g_output_size,\n",
    "                  f=generator_activation_function).to(device)\n",
    "    D = Discriminator(input_size=d_input_func(d_input_size),\n",
    "                      hidden_size=d_hidden_size,\n",
    "                      output_size=d_output_size,\n",
    "                      f=discriminator_activation_function).to(device)\n",
    "    criterion = nn.BCELoss().to(device) # Binary cross entropy: http://pytorch.org/docs/nn.html#bceloss\n",
    "    d_optimizer = optim.SGD(D.parameters(), lr=d_learning_rate, momentum=sgd_momentum)\n",
    "    g_optimizer = optim.Adam(G.parameters(), lr=g_learning_rate )\n",
    "    \n",
    "    iteration , mean , std = 0 , 0 , 0\n",
    "    output = pd.DataFrame({\"iter\" : [iteration] , \"mean\" : [mean] , \"std\" : [std]})\n",
    "    for epoch in range(num_epochs):\n",
    "        for d_index in range(d_steps):\n",
    "            # 1. Train D on real+fake\n",
    "            D.zero_grad()\n",
    "\n",
    "            #  1A: Train D on real\n",
    "            d_real_data = Variable(d_sampler(d_input_size))\n",
    "            d_real_decision = D(preprocess(d_real_data))\n",
    "            d_real_error = criterion(d_real_decision, Variable(torch.ones([1,1]).to(device)))  # ones = true\n",
    "            d_real_error.backward() # compute/store gradients, but don't change params\n",
    "\n",
    "            #  1B: Train D on fake\n",
    "            d_gen_input = Variable(gi_sampler(minibatch_size, g_input_size))\n",
    "            d_fake_data = G(d_gen_input).detach()  # detach to avoid training G on these labels\n",
    "            d_fake_decision = D(preprocess(d_fake_data.t()))\n",
    "            d_fake_error = criterion(d_fake_decision, Variable(torch.zeros([1,1]).to(device)))  # zeros = fake\n",
    "            d_fake_error.backward()\n",
    "            d_optimizer.step()     # Only optimizes D's parameters; changes based on stored gradients from backward()\n",
    "\n",
    "            dre, dfe = extract(d_real_error)[0], extract(d_fake_error)[0]\n",
    "\n",
    "        for g_index in range(g_steps):\n",
    "            # 2. Train G on D's response (but DO NOT train D on these labels)\n",
    "            G.zero_grad()\n",
    "\n",
    "            gen_input = Variable(gi_sampler(minibatch_size, g_input_size))\n",
    "            g_fake_data = G(gen_input)\n",
    "            dg_fake_decision = D(preprocess(g_fake_data.t()))\n",
    "            g_error = criterion(dg_fake_decision, Variable(torch.ones([1,1]).to(device)))  # Train G to pretend it's genuine\n",
    "\n",
    "            g_error.backward()\n",
    "            g_optimizer.step()  # Only optimizes G's parameters\n",
    "            ge = extract(g_error)[0]\n",
    "\n",
    "        if epoch % print_interval == 0:\n",
    "            print(\"Epoch %s: D (%s real_err, %s fake_err) G (%s err); Real Dist (%s),  Fake Dist (%s) \" %\n",
    "                  (epoch, dre, dfe, ge, stats(extract(d_real_data)), stats(extract(d_fake_data))))\n",
    "        \n",
    "        \n",
    "        iteration +=1\n",
    "        values = extract(g_fake_data)\n",
    "        mean = np.mean(values)\n",
    "        std = np.std(values)\n",
    "        output1 = pd.DataFrame({\"iter\" : [iteration] , \"mean\" : [mean] , \"std\" : [std]})\n",
    "        output = output.append(output1)\n",
    "        if epoch % 100 == 0 :\n",
    "            clear_output(wait= True)\n",
    "            fig = plt.figure(figsize = (15, 15))\n",
    "            ax1 = fig.add_subplot(3,1,1)\n",
    "            ax2 = fig.add_subplot(3,1,2)\n",
    "            ax3 = fig.add_subplot(3,1,3)\n",
    "            print(\"Plotting the generated distribution...\")\n",
    "            ax2.plot( output.iter , output[\"mean\"], 'ro')\n",
    "            ax2.set_xlabel(\"Iteration\")\n",
    "            ax3.set_xlabel(\"Iteration\")\n",
    "            ax2.set_ylabel(\"mean\")\n",
    "            ax3.set_ylabel(\"std\")\n",
    "            ax2.set_title('Iteration : {}  Mean of Generated Distribution'.format(iteration))\n",
    "            ax3.set_title('Iteration : {}  std of Generated Distribution'.format(iteration))\n",
    "            ax3.plot( output.iter , output[\"std\"], 'go')\n",
    "            #print(\" Values: %s\" % (str(values)))\n",
    "            sns.kdeplot( values , ax = ax1)\n",
    "            ax1.set_xlabel('Value')\n",
    "            ax1.set_ylabel('Count')\n",
    "            ax1.set_title(\"Epoch %s: D (%s real_err, %s fake_err) G (%s err);  \" %\n",
    "                  (epoch, round(dre,4), round(dfe,4), round(ge,4)))\n",
    "            plt.grid(False)\n",
    "            plt.show()\n",
    "\n",
    "\n",
    "train()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 평균이 4 표준편차는 1.5 \n",
    "    * 목표로 하는 분포는 평균이 4 표준편차는 1.5를 나타낸다\n",
    "    * 그러나 kernel density 로 파악을 해보면 약간 쌍봉 형태로 나타나게 된다.\n",
    "    * 여기서 D 는 데이터 1.평균 , 2.잔차 3.왜도 4.첨도 기준으로 판단하게 된다.\n",
    "        * 그래서 이 4가지 조건에서만 G가 달성하면 D는 속게 된다.\n",
    "\n",
    "## GPU\n",
    "    * 사용하는 TENSOR에 .to(device) 이런식으로 넣어주면 돌아가는 것 같다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[ 1.3771, -0.8693,  0.7400, -0.2607, -0.8174, -0.9412,  0.7873, -0.7265,\n",
       "          1.3371, -0.2109]], device='cuda:0')"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "n = 10 \n",
    "mu = 0 \n",
    "sigma = 1 \n",
    "torch.Tensor(np.random.normal(mu, sigma, (1, n))).cuda()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}