py2.[실습01]MLP-행렬곱 구현.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "da7569c6",
   "metadata": {},
   "source": [
    "### 1. 라이브러리 임포트"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "53a380cc",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "import torch\n",
    "import torchvision.datasets as dset\n",
    "import torchvision.transforms as transforms"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "25de98e8",
   "metadata": {},
   "source": [
    "### 2. MNIST test dataset 가져오기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "cd9e4ee9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# \"\":현재 폴더에 MNIST 있음\n",
    "mnist_test = dset.MNIST(\"\", train=False, transform=transforms.ToTensor(),    # test용으로 사용\n",
    "                        target_transform=None, download=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "51f1a0be",
   "metadata": {},
   "source": [
    "### 3. 대략적인 데이터 형태"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "9170cfc1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mnist_test 길이:  10000\n",
      "image data 형태:  torch.Size([1, 28, 28])\n",
      "label:  7\n"
     ]
    },
    {
     "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": [
    "print \"mnist_test 길이: \", len(mnist_test)\n",
    "\n",
    "# 데이터 하나 형태\n",
    "image, label = mnist_test.__getitem__(0)    # 0번째 데이터\n",
    "print \"image data 형태: \", image.size()\n",
    "print \"label: \", label\n",
    "\n",
    "# 그리기\n",
    "img = image.numpy()    # image 타입을 numpy로 변환(1,28,28)\n",
    "plt.title(\"label: %d\" %label)\n",
    "plt.imshow(img[0], cmap='gray')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "069ae904",
   "metadata": {},
   "source": [
    "### 4. sigmoid, softmax 함수 구현"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bb82316d",
   "metadata": {},
   "source": [
    "$ g(z) = {1 \\over 1+e^{-z}} $"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "04298ae9",
   "metadata": {},
   "outputs": [],
   "source": [
    "def sigmoid(x):\n",
    "    return 1/(1+np.exp(-x))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "190ede17",
   "metadata": {},
   "source": [
    "$ softmax(x)_i = {e^{x_i} \\over \\sum_{j} e^{x_j}} $"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "eb526bf1",
   "metadata": {},
   "outputs": [],
   "source": [
    "def softmax(x):\n",
    "    exp_x = np.exp(x)\n",
    "    return exp_x/np.sum(exp_x)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9f49678f",
   "metadata": {},
   "source": [
    "### 5. 모델 선언"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "b087a249",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Multi-layered perceptron\n",
    "# # of units in each layer : 28*28 - 256 - 128 - 10\n",
    "class MyMLP:\n",
    "    def __init__(self, n_input, n_hidden1, n_hidden2, n_output):\n",
    "        # W^(1): layer1->layer2에 매핑되는 Weight\n",
    "        self.W1 = np.zeros((n_hidden1, n_input), dtype=np.float32)    # W1(256,28*28)\n",
    "        self.b1 = np.zeros((n_hidden1,), dtype=np.float32)\n",
    "        \n",
    "        self.W2 = np.zeros((n_hidden2, n_hidden1), dtype=np.float32)    # W2(128,256)\n",
    "        self.b2 = np.zeros((n_hidden2,), dtype=np.float32)\n",
    "        \n",
    "        self.W3 = np.zeros((n_output, n_hidden2), dtype=np.float32)    #W3(10,128)\n",
    "        self.b3 = np.zeros((n_output,), dtype=np.float32)\n",
    "        \n",
    "    def __call__(self, x):\n",
    "        # (1,28,28) -> (28*28)\n",
    "        x = x.reshape(-1)    # 일렬로 피기\n",
    "        \n",
    "        h1 = sigmoid(np.dot(self.W1, x) + self.b1)    # W1(256,28*28), X(28*28), b1(256) -> h1(256)\n",
    "        h2 = np.dot(self.W2, h1) + self.b2    # W2(128,256), h1(256), b2(128) -> h2(128)\n",
    "        out = np.dot(self.W3, h2) + self.b3    # W3(10,128), h2(128), b3(10) -> out(10)\n",
    "        \n",
    "        return softmax(out)    # (10)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "70e113e1",
   "metadata": {},
   "source": [
    "### 6. 모델 생성"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "ec7ec763",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = MyMLP(28*28, 256, 128, 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "1987c3d4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(256, 784) (256,)\n",
      "(128, 256) (128,)\n",
      "(10, 128) (10,)\n"
     ]
    }
   ],
   "source": [
    "print model.W1.shape, model.b1.shape\n",
    "print model.W2.shape, model.b2.shape\n",
    "print model.W3.shape, model.b3.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a1d938e6",
   "metadata": {},
   "source": [
    "### 7. 미리 학습된 weight 로드"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "f3435da5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(256, 784) (256,)\n",
      "(128, 256) (128,)\n",
      "(10, 128) (10,)\n"
     ]
    }
   ],
   "source": [
    "weights = np.load('./nets/mlp_weight.npz')\n",
    "model.W1 = weights['W1']\n",
    "model.b1 = weights['b1']\n",
    "model.W2 = weights['W2']\n",
    "model.b2 = weights['b2']\n",
    "model.W3 = weights['W3']\n",
    "model.b3 = weights['b3']\n",
    "\n",
    "print model.W1.shape, model.b1.shape\n",
    "print model.W2.shape, model.b2.shape\n",
    "print model.W3.shape, model.b3.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "95809192",
   "metadata": {},
   "source": [
    "### 8. 테스트"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "33517901",
   "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"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "정확도: 91.91\n"
     ]
    }
   ],
   "source": [
    "mysum = 0\n",
    "\n",
    "m = len(mnist_test)\n",
    "cnt = 0\n",
    "for i in range(m):\n",
    "    image, label = mnist_test.__getitem__(i)    # 0번째 데이터\n",
    "    output = model(image)\n",
    "    \n",
    "    if(i%1000==0):\n",
    "        img = image.numpy()    # image 타입을 numpy로 변환(1,28,28)\n",
    "        pred_label = np.argmax(output)\n",
    "        plt.title(\"pred: %d, label: %d\" %(pred_label, label))\n",
    "        plt.imshow(img[0], cmap='gray')\n",
    "        plt.show()\n",
    "        \n",
    "    cnt+=1\n",
    "    mysum+=(np.argmax(output)==label)\n",
    "    \n",
    "print \"정확도: %.2f\" %((float(mysum)/cnt)*100.0)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "conda_pytorch_p27",
   "language": "python",
   "name": "conda_pytorch_p27"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}