04_DGL_07_整图预测.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 一.原理\n",
    "整图预测是针对图层面的学习任务，比如判断某药物分子是否具有某种理化性质，再比如判断某社团是否具有欺诈可能，这需要我们对整个图提取它的特征表示，然后再基于此构建我们的学习任务，图的整体特征无外乎来源于三部分：1）节点特征；2）边特征；3）结构信息，基于这些信息，我们可以通过许多方式来构建图特征，DGL提供了一些简单的API，比如对各节点特征求和/求平均/pooling等，这可以方便我们构建一些基准图预测模型，下面我们利用对节点特征求平均的方式构建图特征，这可以通过`dgl.mean_nodes`这个API很方便的实现，它相当于做了如下计算：    \n",
    "\n",
    "$$\n",
    "h_g=\\frac{1}{|V|}\\sum_{v\\in V }h_v\n",
    "$$  \n",
    "\n",
    "$h_v$表示节点$v$的特征，然后基于$h_g$特征向量，构建我们预测模型"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 二.实现  \n",
    "利用dgl自带的MiniGCDataset数据集，它包括如下的8种类别的图结构   \n",
    "![avatar](./pic/MiniGCDataset.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using backend: pytorch\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#1.导入数据\n",
    "import dgl\n",
    "import torch\n",
    "from dgl.data import MiniGCDataset\n",
    "import matplotlib.pyplot as plt\n",
    "import networkx as nx\n",
    "#这里，随机构造了80个图，每个图是少10条边，最多30条边\n",
    "dataset = MiniGCDataset(80, 10, 20)\n",
    "graph, label = dataset[0]\n",
    "\n",
    "#绘制图像\n",
    "%matplotlib inline\n",
    "fig, ax = plt.subplots()\n",
    "nx.draw(graph.to_networkx(), ax=ax)\n",
    "ax.set_title('Class: {:d}'.format(label))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "#2.定义模型\n",
    "from dgl.nn.pytorch import GraphConv\n",
    "import torch.nn.functional as F\n",
    "from torch import nn\n",
    "class Classifier(nn.Module):\n",
    "    def __init__(self, in_dim, hidden_dim, n_classes):\n",
    "        super(Classifier, self).__init__()\n",
    "        self.conv1 = GraphConv(in_dim, hidden_dim)\n",
    "        self.conv2 = GraphConv(hidden_dim, hidden_dim)\n",
    "        self.classify = nn.Linear(hidden_dim, n_classes)#线性分类器\n",
    "        \n",
    "    def forward(self, g):\n",
    "        # 以节点度作为初始节点特征。对于无向图，入度与外度相同。\n",
    "        h = g.in_degrees().view(-1, 1).float()\n",
    "        # 执行图形卷积和激活函数\n",
    "        h = F.relu(self.conv1(g,h))\n",
    "        h = F.relu(self.conv2(g,h))\n",
    "        g.ndata['h'] = h\n",
    "        \n",
    "        # 通过对所有节点表示求平均来计算图形表示。\n",
    "        hg = dgl.mean_nodes(g, 'h')\n",
    "        return self.classify(hg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x2658c2b9dd8>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "#3.训练\n",
    "import torch.optim as optim\n",
    "from torch.utils.data import DataLoader\n",
    "\n",
    "# 将多张图合并为一张图\n",
    "def collate(samples):\n",
    "    # The input `samples` is a list of pairs (graph, label).\n",
    "    graphs, labels = map(list, zip(*samples)) #把一批图 zip成 列表对象\n",
    "    batched_graph = dgl.batch(graphs)#合并为一张图\n",
    "    return batched_graph, torch.tensor(labels)\n",
    "\n",
    "\n",
    "# 训练集/测试集\n",
    "trainset = MiniGCDataset(320, 10, 20)\n",
    "testset = MiniGCDataset(80, 10, 20)\n",
    "\n",
    "#batch训练\n",
    "data_loader = DataLoader(trainset, batch_size=32, shuffle=True,\n",
    "                         collate_fn=collate)\n",
    "\n",
    "# 构建模型\n",
    "model = Classifier(1, 256, trainset.num_classes)\n",
    "\n",
    "loss_func = nn.CrossEntropyLoss()\n",
    "\n",
    "optimizer = optim.Adam(model.parameters(), lr=0.001)\n",
    "\n",
    "model.train()\n",
    "epoch_losses = []\n",
    "\n",
    "for epoch in range(100):\n",
    "    epoch_loss = 0\n",
    "    for i, (bg, label) in enumerate(data_loader):\n",
    "        prediction = model(bg)\n",
    "        loss = loss_func(prediction, label.long())\n",
    "        optimizer.zero_grad()\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        epoch_loss += loss.detach().item()\n",
    "    epoch_loss /= (i + 1)\n",
    "#     print('Epoch {}, loss {:.4f}'.format(epoch, epoch_loss))\n",
    "    epoch_losses.append(epoch_loss)\n",
    "\n",
    "plt.plot(epoch_losses)\n",
    "plt.legend([\"loss\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of argmax predictions on the test set: 72.500000%\n"
     ]
    }
   ],
   "source": [
    "#4.测试\n",
    "model.eval()\n",
    "test_X, test_Y = map(list, zip(*testset))\n",
    "\n",
    "test_bg = dgl.batch(test_X)\n",
    "test_Y = torch.tensor(test_Y).float().view(-1, 1)\n",
    "\n",
    "pred_Y = torch.max(model(test_bg), 1)[1].view(-1, 1)\n",
    "\n",
    "print('Accuracy of argmax predictions on the test set: {:4f}%'.format(\n",
    "    (test_Y == pred_Y.float()).sum().item() / len(test_Y) * 100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[10,  0,  0,  0,  0,  0,  0,  0],\n",
       "       [ 0,  5,  0,  0,  3,  2,  0,  0],\n",
       "       [ 0,  0, 10,  0,  0,  0,  0,  0],\n",
       "       [ 0,  0,  0,  7,  0,  0,  3,  0],\n",
       "       [ 0,  0,  3,  0,  0,  0,  0,  7],\n",
       "       [ 0,  0,  0,  0,  2,  6,  0,  2],\n",
       "       [ 0,  0,  0,  0,  0,  0, 10,  0],\n",
       "       [ 0,  0,  0,  0,  0,  0,  0, 10]], dtype=int64)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#5.查看混淆矩阵\n",
    "from sklearn.metrics import confusion_matrix\n",
    "confusion_matrix(test_Y, pred_Y)"
   ]
  },
  {
   "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": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}