03-파이토치로_구현하는_ANN/basic_neural_network.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 파이토치로 구현하는 신경망\n",
    "\n",
    "## 신경망 모델 구현하기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import numpy\n",
    "from sklearn.datasets import make_blobs\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_dim = 2\n",
    "x_train, y_train = make_blobs(n_samples=80, n_features=n_dim, centers=[[1,1],[-1,-1],[1,-1],[-1,1]], shuffle=True, cluster_std=0.3)\n",
    "x_test, y_test = make_blobs(n_samples=20, n_features=n_dim, centers=[[1,1],[-1,-1],[1,-1],[-1,1]], shuffle=True, cluster_std=0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def label_map(y_, from_, to_):\n",
    "    y = numpy.copy(y_)\n",
    "    for f in from_:\n",
    "        y[y_ == f] = to_\n",
    "    return y\n",
    "\n",
    "y_train = label_map(y_train, [0, 1], 0)\n",
    "y_train = label_map(y_train, [2, 3], 1)\n",
    "y_test = label_map(y_test, [0, 1], 0)\n",
    "y_test = label_map(y_test, [2, 3], 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def vis_data(x,y = None, c = 'r'):\n",
    "    if y is None:\n",
    "        y = [None] * len(x)\n",
    "    for x_, y_ in zip(x,y):\n",
    "        if y_ is None:\n",
    "            plt.plot(x_[0], x_[1], '*',markerfacecolor='none', markeredgecolor=c)\n",
    "        else:\n",
    "            plt.plot(x_[0], x_[1], c+'o' if y_ == 0 else c+'+')\n",
    "\n",
    "plt.figure()\n",
    "vis_data(x_train, y_train, c='r')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([80, 2])\n"
     ]
    }
   ],
   "source": [
    "x_train = torch.FloatTensor(x_train)\n",
    "print(x_train.shape)\n",
    "x_test = torch.FloatTensor(x_test)\n",
    "y_train = torch.FloatTensor(y_train)\n",
    "y_test = torch.FloatTensor(y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "class NeuralNet(torch.nn.Module):\n",
    "        def __init__(self, input_size, hidden_size):\n",
    "            super(NeuralNet, self).__init__()\n",
    "            self.input_size = input_size\n",
    "            self.hidden_size  = hidden_size\n",
    "            self.linear_1 = torch.nn.Linear(self.input_size, self.hidden_size)\n",
    "            self.relu = torch.nn.ReLU()\n",
    "            self.linear_2 = torch.nn.Linear(self.hidden_size, 1)\n",
    "            self.sigmoid = torch.nn.Sigmoid()\n",
    "            \n",
    "        def forward(self, input_tensor):\n",
    "            linear1 = self.linear_1(input_tensor)\n",
    "            relu = self.relu(linear1)\n",
    "            linear2 = self.linear_2(relu)\n",
    "            output = self.sigmoid(linear2)\n",
    "            return output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = NeuralNet(2, 5)\n",
    "learning_rate = 0.03\n",
    "criterion = torch.nn.BCELoss()\n",
    "epochs = 2000\n",
    "optimizer = torch.optim.SGD(model.parameters(), lr = learning_rate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before Training, test loss is 0.6852480173110962\n"
     ]
    }
   ],
   "source": [
    "model.eval()\n",
    "test_loss_before = criterion(model(x_test).squeeze(), y_test)\n",
    "print('Before Training, test loss is {}'.format(test_loss_before.item()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "오차값이 0.73 이 나왔습니다. 이정도의 오차를 가진 모델은 사실상 분류하는 능력이 없다고 봐도 무방합니다.\n",
    "자, 이제 드디어 인공신경망을 학습시켜 퍼포먼스를 향상시켜 보겠습니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train loss at 0 is 0.6864233613014221\n",
      "Train loss at 100 is 0.6495805978775024\n",
      "Train loss at 200 is 0.6164270639419556\n",
      "Train loss at 300 is 0.5812116861343384\n",
      "Train loss at 400 is 0.5414608716964722\n",
      "Train loss at 500 is 0.4868636727333069\n",
      "Train loss at 600 is 0.4093613624572754\n",
      "Train loss at 700 is 0.3258974552154541\n",
      "Train loss at 800 is 0.25767043232917786\n",
      "Train loss at 900 is 0.2096645087003708\n",
      "Train loss at 1000 is 0.1747014969587326\n",
      "Train loss at 1100 is 0.14887328445911407\n",
      "Train loss at 1200 is 0.12930235266685486\n",
      "Train loss at 1300 is 0.11365455389022827\n",
      "Train loss at 1400 is 0.10088513791561127\n",
      "Train loss at 1500 is 0.0905335545539856\n",
      "Train loss at 1600 is 0.08191367983818054\n",
      "Train loss at 1700 is 0.07460639625787735\n",
      "Train loss at 1800 is 0.06843724846839905\n",
      "Train loss at 1900 is 0.06324858218431473\n"
     ]
    }
   ],
   "source": [
    "for epoch in range(epochs):\n",
    "    model.train()\n",
    "    optimizer.zero_grad()\n",
    "    train_output = model(x_train)\n",
    "    train_loss = criterion(train_output.squeeze(), y_train)\n",
    "    if epoch % 100 == 0:\n",
    "        print('Train loss at {} is {}'.format(epoch, train_loss.item()))\n",
    "    train_loss.backward()\n",
    "    optimizer.step()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "After Training, test loss is 0.04343646019697189\n"
     ]
    }
   ],
   "source": [
    "model.eval()\n",
    "test_loss = criterion(torch.squeeze(model(x_test)), y_test)\n",
    "print('After Training, test loss is {}'.format(test_loss.item()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "학습을 하기 전과 비교했을때 현저하게 줄어든 오차값을 확인 하실 수 있습니다.\n",
    "지금까지 인공신경망을 구현하고 학습시켜 보았습니다.\n",
    "이제 학습된 모델을 .pt 파일로 저장해 보겠습니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "state_dict format of the model: OrderedDict([('linear_1.weight', tensor([[-2.0236,  1.1915],\n",
      "        [-0.1490,  0.0405],\n",
      "        [-1.3762, -1.1183],\n",
      "        [-0.4763, -0.3839],\n",
      "        [ 1.2096, -1.8587]])), ('linear_1.bias', tensor([-0.3576, -0.5651, -0.0955, -0.0369, -0.1106])), ('linear_2.weight', tensor([[ 2.3306,  0.4274, -1.7503, -0.5550,  2.0900]])), ('linear_2.bias', tensor([-2.0816]))])\n"
     ]
    }
   ],
   "source": [
    "torch.save(model.state_dict(), './model.pt')\n",
    "print('state_dict format of the model: {}'.format(model.state_dict()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`save()` 를 실행하고 나면 학습된 신경망의 가중치를 내포하는 model.pt 라는 파일이 생성됩니다. 아래 코드처럼 새로운 신경망 객체에 model.pt 속의 가중치값을 입력시키는 것 또한 가능합니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "벡터 [-1, 1]이 레이블 1을 가질 확률은 0.9861199855804443\n"
     ]
    }
   ],
   "source": [
    "new_model = NeuralNet(2, 5)\n",
    "new_model.load_state_dict(torch.load('./model.pt'))\n",
    "new_model.eval()\n",
    "print('벡터 [-1, 1]이 레이블 1을 가질 확률은 {}'.format(new_model(torch.FloatTensor([-1,1])).item()))"
   ]
  }
 ],
 "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}