14-基于自表达性的深度子空间聚类.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "${{ \\left\\Vert {X} \\right\\Vert }\\mathop{{}}\\nolimits_{{0}}}$  L0范数:表示向量x中非零元素的个数  \n",
    "${{ \\left\\Vert {X} \\right\\Vert }\\mathop{{}}\\nolimits_{{1}}}$  L1范数（列模）:表示向量x中非零元素的绝对值之和  \n",
    "${{ \\left\\Vert {X} \\right\\Vert }\\mathop{{}}\\nolimits_{{2}}}$  L2范数（谱模）:表示向量元素的平方和再开平方求$A^{T}A$的特征值，找出其中的最大特征值，求其平方根,相当于$max(sqrt(eig(A^{T}A)))$，也叫谱范数  \n",
    "${{ \\left\\Vert {X} \\right\\Vert }\\mathop{{}}\\nolimits_{{F}}}$  F范数:是把一个矩阵中每个元素的平方求和后开根号(用于表示矩阵量级)   \n",
    "${{ \\left\\Vert {X} \\right\\Vert }\\mathop{{}}\\nolimits_{{\\infty}}}$  无穷范数（行模）:度量向量元素的最大值   \n",
    "${{ \\left\\Vert {X} \\right\\Vert }\\mathop{{}}\\nolimits_{{*}}}$  核范数Nuclear Norm:矩阵奇异值的和(用于表示低秩矩阵)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 子空间聚类（Subspace clustering）\n",
    "[Subspace clustering](https://towardsdatascience.com/subspace-clustering-7b884e8fff73)  \n",
    "\n",
    "本教程解决以下问题：  \n",
    "1. 高维数据的处理存在什么挑战？\n",
    "+ 什么是子空间聚类？\n",
    "+ 如何用python进行子空间聚类"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "高维数据有从几十到甚至几千个维度。比如，OD数据  \n",
    "## 高维数据的处理存在困难：\n",
    "\n",
    "1. 可视化困难，很难理解数据长什么样，因此它需要降维。也导致了维度灾难，即很难对每个子维度进行枚举迭代\n",
    "+ 前面降维技术的选择会极大的影响后续的聚类效果\n",
    "+ 许多维度可能是不相关的，并且可以在有噪声的数据中屏蔽现有的聚类\n",
    "+ 一种常见的技术是执行特征选择（删除不相关的维度），但是在某些情况下，识别冗余维度并不容易\n",
    "\n",
    "## 什么是子空间聚类？\n",
    "\n",
    "> 子空间聚类是一种在不同子空间中发现聚类的技术。  \n",
    "\n",
    "*子空间即是数据里面一个或者多个维度的组合*\n",
    "\n",
    "基本的假设是，我们可以找到只由维度子集定义的有效聚类（不需要具有所有N个特征的一致性）。  \n",
    "> **举例:** 如果我们输入病人的基因数据(每个病人的基因有20000个属性，数据维度有20000个)，有一簇病人患了帕金森病，这些病人只需要看100个基因就可以知道，那么我们称这个子集合存在于100维里。\n",
    "\n",
    "\n",
    "换句话说，子空间聚类是传统N维聚类分析的扩展，它允许通过创建**行**和**列**同时进行聚类。（传统的只是对行聚类，子空间聚类是同时对行和列聚类）\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-17T14:17:45.165029Z",
     "start_time": "2019-09-17T14:17:45.157085Z"
    }
   },
   "source": [
    "<img alt=\"\"  width=\"500\" height=\"360\" src=\"./resource/1_x8UldsjTQjyfx_m23nx_QA.png\">\n",
    "子空间的聚类结果可能在属性（行）和观测值（列）之间有重叠，如上图，出自此[paper](https://www.kdd.org/exploration_files/parsons.pdf)。  \n",
    "可以看到，这里竟然可以把数据聚成4类，两个簇之间的元素可以离得很近，但也不会干扰到子空间聚类。传统的聚类方法则很容易被干扰"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-17T14:22:51.295101Z",
     "start_time": "2019-09-17T14:22:51.289151Z"
    }
   },
   "source": [
    "<img alt=\"\"  width=\"1000\" height=\"360\" src=\"./resource/1_Ee99wDxlommERhG4QwsOeQ.png\">\n",
    "上面的数据，如果你从三个维度观看，会发现每个维度中，都有不同簇的数据糅合在一起"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 子空间聚类的种类"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "基于搜索策略，我们可以区分两类子空间聚类算法：  \n",
    "1. 由下至上算法从找到低维度的（1D）聚类开始，逐渐融合，以进行高维度的处理。\n",
    "+ 由上至下算法从全部维度开始找到聚类，如何开始评估每个聚类的子空间。\n",
    "下图展示的是常见的子空间聚类算法\n",
    "<img alt=\"\"  width=\"1000\" height=\"360\" src=\"./resource/1_1JgMfAGmIYbeInlT5LCxLQ.png\">\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 派系算法（Clique algorithm）"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "简单来说，算法经过了以下步骤：\n",
    "> 对每个维度(属性，feature)，我们将空间分割为nBins个格子（第一个参数），并对每个格子计算直方图（数据量）。我们只考虑密度大的单元(dense units)，即格子里数据量大于某个给定的阈值nPoints（第二个参数）。每个dense units又带着属性：\n",
    "1. 它所在的维度(属性，feature)\n",
    "2. 格子的编号\n",
    "3. 格子中的数据"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[代码](https://github.com/ciortanmadalina/medium/blob/master/clique_clustering.ipynb)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-18T01:21:57.598245Z",
     "start_time": "2019-09-18T01:21:57.312970Z"
    }
   },
   "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": [
    "from sklearn.datasets.samples_generator import make_blobs\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "n_components = 4\n",
    "data, truth = make_blobs(n_samples=100, centers=n_components, random_state=42, n_features=2)\n",
    "data = preprocessing.MinMaxScaler().fit_transform(data)\n",
    "plt.scatter(data[:, 0], data[:, 1], s=50, c = truth)\n",
    "plt.title(f\"Example of a mixture of {n_components} distributions\")\n",
    "plt.xlabel(\"Feature 1\")\n",
    "plt.ylabel(\"Feature 2\");"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "该数据集中，有2个维度，存在4个簇。算法选择参数nBins = 8,nPoints = 2。  \n",
    "算法选择由下至上的思路，从1D开始。如果2个以上的dense units为邻居，则将它们合并为一个更大的bin。如果将这些格子转换为网络图，这个操作就可以很容易地进行，创建图的时候，每个dense units为节点，如果两个节点属于同一维度，而且他们相邻（他们之间的距离不超过1），则在两个节点之间生成边。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-18T02:01:02.617401Z",
     "start_time": "2019-09-18T02:01:02.611418Z"
    }
   },
   "source": [
    "<img alt=\"\"  width=\"500\" height=\"360\" src=\"./resource/1_IEhLyuCflUA-XIwcPJqjHw.png\">\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-18T02:02:31.576167Z",
     "start_time": "2019-09-18T02:02:31.569183Z"
    }
   },
   "source": [
    "横轴维度区分两簇\n",
    "<img alt=\"\"  width=\"500\" height=\"360\" src=\"./resource/1_QlGgbOK9_3W1LnuJKibbOg.png\">\n",
    "纵轴维度区分三簇\n",
    "<img alt=\"\"  width=\"500\" height=\"360\" src=\"./resource/1_MmMK1SlBqzdwgCg8U-v7mQ.png\">\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "接下来，我们要从2D开始到所有D，计算所有可行的的簇。这个操作可以转化为计算k个维度的dense units的融合，并只保留有重叠的且连续的dense bins的结果。在计算完k-1维度的dense units后，我们可以通过计算所有k-1维度的dense units的融合而扩展到第k维。  \n",
    "\n",
    "<img alt=\"\"  width=\"1000\" height=\"360\" src=\"./resource/c75c10385343fbf224b69b11929e698464388f4a.jpeg\">\n",
    "在上面的数据中，我们可以得到下图的聚类结果。  \n",
    "紫色点不属于任何类簇，因为他们所属的栅格中数据个数小于2个（nPoints）\n",
    "<img alt=\"\"  width=\"500\" height=\"360\" src=\"./resource/1_434TdPdpi1eEZWno4Ez4Vg.png\">\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "派系算法对它参数的选择非常敏感（nBins和nPoints）。不过，它是由下至上算法家族中最基础的算法。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 基于谱聚类的子空间聚类"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "实际上，谱聚类实现的就是子空间聚类  \n",
    "回想一下，我们在谱聚类最后一步的时候,$L$矩阵有$n$个维度，我们只通过最小的$k$个特征值所对应的特征向量对数据进行kmeans聚类，就只考虑了$k$个维度，这不就是子空间吗    \n",
    "前面我们讲到，谱聚类中需要构建相似矩阵(affinity matrix)，以$W$或者$A$表示\n",
    "\n",
    "## Self-Expressiveness affinity（基于自表达性的相似度矩阵）\n",
    "\n",
    "概念：从线性子空间中提取的一个数据点$x_i$可以由同一子空间中其他点的线性组合表示将所有点叠加到数据矩阵$X$的列中，其自表达性可以简单地描述为$$X = XC$$其中$C$为Self-Expressiveness系数矩阵，$X$矩阵为$(n*m)$，$C$矩阵为$(m*m)$，$XC$矩阵为$(n*m)$，$n$为数据记录数，$m$为每条数据的维度。在这里，以OD矩阵为例，$n$为OD对数，$m$为时间，即  \n",
    "\n",
    "|X矩阵|1日|2日|...|m日|\n",
    "|----|||||\n",
    "|A到B|20|30|...|40|\n",
    "|A到C|20|30|...|40|\n",
    "|...|...|...|...|...|\n",
    "|n|20|30|...|40|\n",
    "\n",
    "|C矩阵|1日|2日|...|m日|\n",
    "|----|||||\n",
    "|1日|0|相似度|...|相似度|\n",
    "|2日|相似度|0|...|相似度|\n",
    "|...|...|...|...|...|\n",
    "|m日|相似度|相似度|...|0|\n",
    "\n",
    "假设子空间都是独立的，最小化$C$矩阵的范数，就可以保证$C$在以某些顺序排列的时候出现块状对角线结构，也因此可以以$C$矩阵来构建相似度矩阵用于谱聚类[P. Ji, M. Salzmann, and H. Li. Efficient dense subspace clustering. In WACV, pages 461–468. IEEE, 2014.]  \n",
    "在数学上为：  \n",
    "$$min{{ \\left\\Vert {C} \\right\\Vert }_{p}}$$  \n",
    "$$s.t. diag(C)=0$$  \n",
    "$$X = XC$$  \n",
    "$$C\\geqslant0$$  \n",
    "其中，${ \\left\\Vert {C} \\right\\Vert }_{p}$为矩阵$C$的任意范数\n",
    "找到这样子的$C$,就可以构建出谱聚类所需的相似度矩阵$A$:\n",
    "$$A=C^T+C$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-info\"><h3>*常见的矩阵范数总结  </h3><p>  \n",
    "${{ \\left\\Vert {X} \\right\\Vert }\\mathop{{}}\\nolimits_{{0}}}$  L0范数:表示向量x中非零元素的个数  <br>  \n",
    "${{ \\left\\Vert {X} \\right\\Vert }\\mathop{{}}\\nolimits_{{1}}}$  L1范数（列模）:表示向量x中非零元素的绝对值之和  <br>  \n",
    "${{ \\left\\Vert {X} \\right\\Vert }\\mathop{{}}\\nolimits_{{2}}}$  L2范数（谱模）:表示向量元素的平方和再开平方求$A^{T}A$的特征值，找出其中的最大特征值，求其平方根,相当于$max(sqrt(eig(A^{T}A)))$，也叫谱范数  <br>  \n",
    "${{ \\left\\Vert {X} \\right\\Vert }\\mathop{{}}\\nolimits_{{F}}}$  F范数:是把一个矩阵中每个元素的平方求和后开根号(用于表示矩阵量级)   <br>  \n",
    "${{ \\left\\Vert {X} \\right\\Vert }\\mathop{{}}\\nolimits_{{\\infty}}}$  无穷范数（行模）:度量向量元素的最大值   <br>  \n",
    "${{ \\left\\Vert {X} \\right\\Vert }\\mathop{{}}\\nolimits_{{*}}}$  核范数Nuclear Norm:矩阵奇异值的和(用于表示低秩矩阵)<br>  \n",
    "\n",
    "</p></div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "而在实际的数据集中，存在数据噪声，因此将问题调整为：  \n",
    "$$min{{ \\left\\Vert {C} \\right\\Vert }\\mathop{{}}\\nolimits_{{F}}}+\\frac{\\lambda}{2}{{ \\left\\Vert {X-XC} \\right\\Vert }\\mathop{{}}\\nolimits_{{F}}}^2$$  \n",
    "$$s.t. diag(C)=0$$  \n",
    "$$C\\geqslant0$$  \n",
    "找到这样子的$C$,就可以用谱聚类"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "上面的目标函数，其实相当于同时达成两个目标，即：\n",
    "1. ${ \\left\\Vert {C} \\right\\Vert }\\mathop{{}}\\nolimits_{{F}}$最小，这个是为了满足Self-Expressiveness系数矩阵定义\n",
    "2. ${ \\left\\Vert {X-XC} \\right\\Vert }\\mathop{{}}\\nolimits_{{F}}$最小，实际上$XC$是我们构建出来的矩阵，$X-XC$是噪声矩阵，即噪声最小  \n",
    "\n",
    "那么，这个目标函数相当于我们能够同时完成去噪声和降维的工作,例如：  \n",
    "<img alt=\"\"  width=\"500\" height=\"360\" src=\"./resource/20150314105633589.jpg\">\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 那么，如何求解，$C$矩阵怎么获得?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Deep Adversarial Subspace Clustering（2018年机器学习领域论文）"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[【论文阅读】Deep Adversarial Subspace Clustering](https://www.cnblogs.com/EstherLjy/p/9840016.html)  \n",
    "[Deep-subspace-clustering-networks github](https://github.com/panji1990/Deep-subspace-clustering-networks)  \n",
    "给出的解决方案是搭建Deep Convolutional Auto-Encoder  \n",
    "<img src = \"./resource/Deepautoencoder.png\" style=\"width:800px\">\n",
    "损失函数为  \n",
    "<img src = \"./resource/lossfunction1.png\" style=\"width:800px\">\n",
    "其中"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 246,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-23T16:03:22.715477Z",
     "start_time": "2019-09-23T16:03:22.711488Z"
    }
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.utils.data as Data\n",
    "import torchvision\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "EPOCH = 100\n",
    "BATCH_SIZE = 500\n",
    "LR = 0.01\n",
    "DOWNLOAD_MNIST = False\n",
    "\n",
    "train_data = torchvision.datasets.MNIST(\n",
    "    root = './mnist',\n",
    "    train = True,\n",
    "    transform = torchvision.transforms.ToTensor(), #从0-255压缩到0-1\n",
    "    download =DOWNLOAD_MNIST\n",
    ")\n",
    "\n",
    "# 先转换成 torch 能识别的 Dataset\n",
    "torch_dataset = Data.TensorDataset(train_data.train_data[:BATCH_SIZE], train_data.train_labels[:BATCH_SIZE])\n",
    "\n",
    "# 把 dataset 放入 DataLoader\n",
    "loader = Data.DataLoader(\n",
    "    dataset=torch_dataset,      # torch TensorDataset format\n",
    "    batch_size=BATCH_SIZE,      # mini batch size\n",
    "    shuffle=False,               # 要不要打乱数据 (打乱比较好)\n",
    "    num_workers=2,              # 多线程来读数据\n",
    ")\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 250,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-23T16:03:27.485491Z",
     "start_time": "2019-09-23T16:03:27.386239Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\torchvision\\datasets\\mnist.py:53: UserWarning: train_data has been renamed data\n",
      "  warnings.warn(\"train_data has been renamed data\")\n",
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\torchvision\\datasets\\mnist.py:43: UserWarning: train_labels has been renamed targets\n",
      "  warnings.warn(\"train_labels has been renamed targets\")\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": [
    "i = 0\n",
    "def printdata(x,y):\n",
    "    plt.imshow(x,cmap = 'gray')\n",
    "    plt.title(y)\n",
    "    plt.show()\n",
    "printdata(train_data.train_data[i],str(train_data.train_labels[i].numpy()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 251,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-23T16:03:28.094149Z",
     "start_time": "2019-09-23T16:03:28.084177Z"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "class DSC(nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        self.encoder = nn.Sequential(\n",
    "            nn.Linear(28*28,128),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(128,64),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(64,12),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(12,5),\n",
    "\n",
    "        )\n",
    "        self.selfexpr = nn.Linear(BATCH_SIZE,BATCH_SIZE,bias=False)\n",
    "        \n",
    "        \n",
    "        self.decoder = nn.Sequential(\n",
    "            nn.Linear(5,12),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(12,64),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(64,128),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(128,28*28),\n",
    "            nn.Sigmoid()\n",
    "        )\n",
    "        \n",
    "        \n",
    "    def forward(self,x):\n",
    "        z = self.encoder(x)\n",
    "        z = torch.transpose(z, 1, 0)\n",
    "        z_ = self.selfexpr(z)\n",
    "        z_ = torch.transpose(z_, 1, 0)\n",
    "        x_ = self.decoder(z_)\n",
    "        return z,z_,x_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 297,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-23T16:30:58.209482Z",
     "start_time": "2019-09-23T16:20:20.300723Z"
    },
    "scrolled": false
   },
   "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"
    }
   ],
   "source": [
    "dsc = DSC()\n",
    "\n",
    "dsc.cuda()\n",
    "optimizer = torch.optim.Adam(dsc.parameters(),lr = LR,betas = (0.9,0.99))    \n",
    "loss_func = torch.nn.MSELoss()\n",
    "\n",
    "loss_list = [] \n",
    "lambda1 = 1\n",
    "lambda2 = 10**(epoch/10-3)\n",
    "\n",
    "steps = []\n",
    "for epoch in range(1000):   # 训练所有!整套!数据 EPOCH 次\n",
    "    for step, (batch_x, batch_y) in enumerate(loader):  # 每一步 loader 释放一小批数据用来学习\n",
    "        steps.append(step)\n",
    "        batch_x = (batch_x.view(-1,28*28).type(torch.FloatTensor)/255).cuda()\n",
    "        batch_y = batch_y.cuda()\n",
    "        \n",
    "        z,z_,x_ = dsc(batch_x)\n",
    "        \n",
    "        C = dsc.selfexpr.state_dict()['weight']\n",
    "        \n",
    "        loss = 1/2*torch.norm(x_-batch_x)**2+lambda1*torch.norm(C)+lambda2/2*torch.norm(torch.transpose(z, 1, 0)-z_)**2\n",
    "        \n",
    "        loss_list.append(loss)\n",
    "        \n",
    "        optimizer.zero_grad()      #初始化梯度\n",
    "        loss.backward()         #计算梯度\n",
    "        optimizer.step()        #对参数学习\n",
    "\n",
    "\n",
    "        import IPython\n",
    "        IPython.display.clear_output(wait=True)\n",
    "        plt.plot(range(len(steps[-50:])),loss_list[-50:])\n",
    "        plt.ylabel('loss')\n",
    "        plt.xlabel('train step')\n",
    "\n",
    "        plt.show()\n",
    "\n",
    "\n",
    "C = dsc.selfexpr.state_dict()['weight'].cpu().numpy()\n",
    "plt.imshow(C,cmap = 'Greys')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "start_time": "2019-09-23T16:40:04.146Z"
    }
   },
   "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": [
    "for epoch in range(1000):   # 训练所有!整套!数据 EPOCH 次\n",
    "    for step, (batch_x, batch_y) in enumerate(loader):  # 每一步 loader 释放一小批数据用来学习\n",
    "        steps.append(step)\n",
    "        batch_x = (batch_x.view(-1,28*28).type(torch.FloatTensor)/255).cuda()\n",
    "        batch_y = batch_y.cuda()\n",
    "        \n",
    "        z,z_,x_ = dsc(batch_x)\n",
    "        \n",
    "        C = dsc.selfexpr.state_dict()['weight']\n",
    "        \n",
    "        loss = 1/2*torch.norm(x_-batch_x)**2+lambda1*torch.norm(C)+lambda2/2*torch.norm(torch.transpose(z, 1, 0)-z_)**2\n",
    "        \n",
    "        loss_list.append(loss)\n",
    "        \n",
    "        optimizer.zero_grad()      #初始化梯度\n",
    "        loss.backward()         #计算梯度\n",
    "        optimizer.step()        #对参数学习\n",
    "\n",
    "\n",
    "        import IPython\n",
    "        IPython.display.clear_output(wait=True)\n",
    "        plt.plot(range(len(steps[-50:])),loss_list[-50:])\n",
    "        plt.ylabel('loss')\n",
    "        plt.xlabel('train step')\n",
    "\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 305,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-23T16:39:45.031641Z",
     "start_time": "2019-09-23T16:39:44.778270Z"
    }
   },
   "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"
    }
   ],
   "source": [
    "i = 10\n",
    "def printdata(x,y):\n",
    "    plt.imshow(x,cmap = 'gray')\n",
    "    plt.title(y)\n",
    "    plt.show()\n",
    "printdata(x_[i].view(28,28).cpu().data.numpy(),'test')\n",
    "def printdata(x,y):\n",
    "    plt.imshow(x,cmap = 'gray')\n",
    "    plt.title(y)\n",
    "    plt.show()\n",
    "printdata(batch_x[i].view(28,28).cpu().data.numpy(),'test')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 302,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-23T16:39:36.833512Z",
     "start_time": "2019-09-23T16:39:36.685913Z"
    }
   },
   "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": [
    "A = C+C.T\n",
    "plt.imshow(A,cmap = 'Greys')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 303,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-23T16:39:37.731112Z",
     "start_time": "2019-09-23T16:39:37.588498Z"
    }
   },
   "outputs": [],
   "source": [
    "D = np.diag(A.sum(axis=1))\n",
    "L = D - A\n",
    "#计算特征值和特征向量\n",
    "vals, vecs = np.linalg.eig(L)\n",
    "#重新排序\n",
    "vecs = vecs[:,np.argsort(vals)]\n",
    "vals = vals[np.argsort(vals)]\n",
    "n = 10\n",
    "#用kmeans对第2到第4个特征向量聚成4类\n",
    "from sklearn.cluster import KMeans\n",
    "kmeans = KMeans(n_clusters=n)\n",
    "kmeans.fit(vecs[:,:n])\n",
    "spectral_labels = kmeans.labels_\n",
    "spectral_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 257,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-23T16:04:51.624479Z",
     "start_time": "2019-09-23T16:04:51.449256Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([5, 0, 4, 1, 9, 2, 1, 3, 1, 4, 3, 5, 3, 6, 1, 7, 2, 8, 6, 9, 4, 0, 9, 1,\n",
       "        1, 2, 4, 3, 2, 7, 3, 8, 6, 9, 0, 5, 6, 0, 7, 6, 1, 8, 7, 9, 3, 9, 8, 5,\n",
       "        9, 3, 3, 0, 7, 4, 9, 8, 0, 9, 4, 1, 4, 4, 6, 0, 4, 5, 6, 1, 0, 0, 1, 7,\n",
       "        1, 6, 3, 0, 2, 1, 1, 7, 9, 0, 2, 6, 7, 8, 3, 9, 0, 4, 6, 7, 4, 6, 8, 0,\n",
       "        7, 8, 3, 1, 5, 7, 1, 7, 1, 1, 6, 3, 0, 2, 9, 3, 1, 1, 0, 4, 9, 2, 0, 0,\n",
       "        2, 0, 2, 7, 1, 8, 6, 4, 1, 6, 3, 4, 5, 9, 1, 3, 3, 8, 5, 4, 7, 7, 4, 2,\n",
       "        8, 5, 8, 6, 7, 3, 4, 6, 1, 9, 9, 6, 0, 3, 7, 2, 8, 2, 9, 4, 4, 6, 4, 9,\n",
       "        7, 0, 9, 2, 9, 5, 1, 5, 9, 1, 2, 3, 2, 3, 5, 9, 1, 7, 6, 2, 8, 2, 2, 5,\n",
       "        0, 7, 4, 9, 7, 8, 3, 2, 1, 1, 8, 3, 6, 1, 0, 3, 1, 0, 0, 1, 7, 2, 7, 3,\n",
       "        0, 4, 6, 5, 2, 6, 4, 7, 1, 8, 9, 9, 3, 0, 7, 1, 0, 2, 0, 3, 5, 4, 6, 5,\n",
       "        8, 6, 3, 7, 5, 8, 0, 9, 1, 0, 3, 1, 2, 2, 3, 3, 6, 4, 7, 5, 0, 6, 2, 7,\n",
       "        9, 8, 5, 9, 2, 1, 1, 4, 4, 5, 6, 4, 1, 2, 5, 3, 9, 3, 9, 0, 5, 9, 6, 5,\n",
       "        7, 4, 1, 3, 4, 0, 4, 8, 0, 4, 3, 6, 8, 7, 6, 0, 9, 7, 5, 7, 2, 1, 1, 6,\n",
       "        8, 9, 4, 1, 5, 2, 2, 9, 0, 3, 9, 6, 7, 2, 0, 3, 5, 4, 3, 6, 5, 8, 9, 5,\n",
       "        4, 7, 4, 2, 7, 3, 4, 8, 9, 1, 9, 2, 8, 7, 9, 1, 8, 7, 4, 1, 3, 1, 1, 0,\n",
       "        2, 3, 9, 4, 9, 2, 1, 6, 8, 4, 7, 7, 4, 4, 9, 2, 5, 7, 2, 4, 4, 2, 1, 9,\n",
       "        7, 2, 8, 7, 6, 9, 2, 2, 3, 8, 1, 6, 5, 1, 1, 0, 2, 6, 4, 5, 8, 3, 1, 5,\n",
       "        1, 9, 2, 7, 4, 4, 4, 8, 1, 5, 8, 9, 5, 6, 7, 9, 9, 3, 7, 0, 9, 0, 6, 6,\n",
       "        2, 3, 9, 0, 7, 5, 4, 8, 0, 9, 4, 1, 2, 8, 7, 1, 2, 6, 1, 0, 3, 0, 1, 1,\n",
       "        8, 2, 0, 3, 9, 4, 0, 5, 0, 6, 1, 7, 7, 8, 1, 9, 2, 0, 5, 1, 2, 2, 7, 3,\n",
       "        5, 4, 9, 7, 1, 8, 3, 9, 6, 0, 3, 1, 1, 2, 6, 3, 5, 7, 6, 8],\n",
       "       device='cuda:0')"
      ]
     },
     "execution_count": 257,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "batch_y"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 実験を始まります！"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## pre training strategies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-24T08:23:05.133207Z",
     "start_time": "2019-09-24T08:22:43.863626Z"
    }
   },
   "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": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.utils.data as Data\n",
    "import torchvision\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "EPOCH = 2\n",
    "BATCH_SIZE = 50\n",
    "LR = 0.01\n",
    "DOWNLOAD_MNIST = False\n",
    "\n",
    "train_data = torchvision.datasets.MNIST(\n",
    "    root = './mnist',\n",
    "    train = True,\n",
    "    transform = torchvision.transforms.ToTensor(), #从0-255压缩到0-1\n",
    "    download =DOWNLOAD_MNIST\n",
    ")\n",
    "\n",
    "# 先转换成 torch 能识别的 Dataset\n",
    "torch_dataset = Data.TensorDataset(train_data.train_data, train_data.train_labels)\n",
    "\n",
    "# 把 dataset 放入 DataLoader\n",
    "loader = Data.DataLoader(\n",
    "    dataset=torch_dataset,      # torch TensorDataset format\n",
    "    batch_size=BATCH_SIZE,      # mini batch size\n",
    "    shuffle=True,               # 要不要打乱数据 (打乱比较好)\n",
    "    num_workers=2,              # 多线程来读数据\n",
    ")\n",
    "\n",
    "\n",
    "class AutoEncoder(nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        self.encoder = nn.Sequential(\n",
    "            nn.Linear(28*28,128),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(128,64),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(64,12),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(12,10),\n",
    "\n",
    "        )\n",
    "        self.decoder = nn.Sequential(\n",
    "            nn.Linear(10,12),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(12,64),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(64,128),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(128,28*28),\n",
    "            nn.Sigmoid()\n",
    "        )\n",
    "        \n",
    "    def forward(self,x):\n",
    "        encoded = self.encoder(x)\n",
    "        decoded = self.decoder(encoded)\n",
    "        return encoded,decoded\n",
    "\n",
    "    \n",
    "autoencoder = AutoEncoder()\n",
    "\n",
    "autoencoder.cuda()\n",
    "optimizer = torch.optim.Adam(autoencoder.parameters(),lr = LR,betas = (0.9,0.99))    \n",
    "loss_func = torch.nn.MSELoss()\n",
    "\n",
    "loss_list = [] \n",
    "\n",
    "\n",
    "for epoch in range(EPOCH):   # 训练所有!整套!数据 EPOCH 次\n",
    "    for step, (batch_x, batch_y) in enumerate(loader):  # 每一步 loader 释放一小批数据用来学习\n",
    "        batch_x = (batch_x.view(-1,28*28).type(torch.FloatTensor)/255).cuda()\n",
    "        batch_y = batch_y.cuda()\n",
    "        \n",
    "        encoded,decoded = autoencoder(batch_x)\n",
    "\n",
    "        loss = loss_func(decoded,batch_x)\n",
    "        \n",
    "        loss_list.append(loss)\n",
    "        \n",
    "        optimizer.zero_grad()      #初始化梯度\n",
    "        loss.backward()         #计算梯度\n",
    "        optimizer.step()        #对参数学习\n",
    "\n",
    "import IPython\n",
    "IPython.display.clear_output(wait=True)\n",
    "\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('train step')\n",
    "plt.plot(range(len(loss_list)),loss_list,label='Adam')\n",
    "plt.legend()\n",
    "plt.show()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-24T08:23:13.606547Z",
     "start_time": "2019-09-24T08:23:13.164762Z"
    }
   },
   "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": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAEThJREFUeJzt3XuMnNV5x/Hf48vaxl7Axhfs9eLFxmBDkE1lGYQrLkqCIEICKhHBH5WrRnXUBtSg/lGEECC1kWjVpEWqGmkjUIyUkCABASGUgFCFyz/GF4zxhYAx1CwYrwHba8C3tZ/+seNmMfs+Z5l5Z96xz/cjoZ2dZ87s2Vl+ft+Z855zzN0FID9jqu4AgGoQfiBThB/IFOEHMkX4gUwRfiBThB/IFOHHiMysx8xeMLN9Zvaxmf2nmY2rul8oD+FHkf+S1C9ptqSlkq6V9HeV9gilIvwocqGkJ939sLt/LOn3ki6ruE8oEeFHkUck3WFmZ5lZl6SbNPQPAM4QhB9FXtHQkX5AUp+k9ZJ+V2mPUCrCj68xszGS/iDpaUmTJU2XNFXSv1TZL5TLmNWHU5nZdEl7JZ3r7gdq990q6Z/d/VuVdg6l4ciPr3H3TyS9J+lvzWycmZ0raaWkN6rtGcpE+FHkLyTdqKEzgB2SBiXdU2mPUCpO+4FMceQHMkX4gUwRfiBThB/IVEtnaZkZny4CTebuNprHNXTkN7MbzeyPZrbDzO5t5LkAtFbdQ31mNlbS25K+q6Frv9dJutPdtwVtOPIDTdaKI/9ySTvcfae7H5X0G0m3NPB8AFqokfB3Sfpg2Pd9tfu+wsxWmdl6M1vfwM8CULJGPvAb6dTia6f17t4rqVfitB9oJ40c+fskdQ/7fq6kjxrrDoBWaST86yQtNLMLzaxD0h2SniunWwCare7TfncfNLO7NLTow1hJj7n71tJ6BqCpWjqrj/f8QPO15CIfAKcvwg9kivADmSL8QKYIP5Apwg9kivADmSL8QKYIP5Apwg9kivADmSL8QKYIP5Cpli7djdOPWTxBbMyY+PjR0dFRWDtx4kTYduzYsQ397Oj5J0yYELYdGBio+7kl6XTYA5MjP5Apwg9kivADmSL8QKYIP5Apwg9kivADmWKc/wyXGgufNGlSWF+8eHFY/853vhPW58yZU1i75JJLwrbjx48P66lrEDZu3FhYe+ONN8K2mzdvDuvbt28P6ylHjhxpqH0ZOPIDmSL8QKYIP5Apwg9kivADmSL8QKYIP5ApxvnPAOPGFf8Zzz777LDtkiVLwvq1114b1i+//PKwPm/evLpqUrwWgJQe5+/s7Cys9ff3h203bNgQ1mfOnBnW9+zZE9ajvrdqLYCGwm9m70s6KOm4pEF3X1ZGpwA0XxlH/uvd/ZMSngdAC/GeH8hUo+F3SS+a2QYzWzXSA8xslZmtN7P1Df4sACVq9LR/hbt/ZGYzJb1kZm+5+5rhD3D3Xkm9kmRm7b+qIZCJho787v5R7Wu/pGckLS+jUwCar+7wm9lkM+s8eVvSDZK2lNUxAM3VyGn/LEnP1MYrx0n6tbv/vpRe4StSc/KjsfwVK1aEbW+77bawPmvWrLA+ffr0sB6tj3/48OG6247G8ePHC2v79+8P26bqBw4cCOuDg4NhvR3W9a87/O6+U1J8hQiAtsVQH5Apwg9kivADmSL8QKYIP5AppvS2gdRQXmoJ6/nz5xfWUktrp6b8pqbVpqauHj16tLCWWjb8rLPOCutbtsSXlbz44ouFtdTS26mhvuj3ktJbeLcDjvxApgg/kCnCD2SK8AOZIvxApgg/kCnCD2SKcf4WSC0xnRrnT01tnTt3bmHt0KFDYdu9e/eG9XfeeSesp7a6jq5RuOiii8K2F1xwQVjftm1bWI/G8lO/d2pKbjRd+HTBkR/IFOEHMkX4gUwRfiBThB/IFOEHMkX4gUwxzt8GUuP8M2bMCOtdXV2FtdSc+NS8861bt4b1ffv2hfVp06bVVZPircclaezYsWE9+t2OHTsWtj0dlt5uFEd+IFOEH8gU4QcyRfiBTBF+IFOEH8gU4QcyxTh/G0jN9584cWJYj9btv+aaa8K2u3btCus7d+4M61OmTAnrCxcuLKxF1ydI6bUIUuv+R+P8qesbTod19xuVPPKb2WNm1m9mW4bdN83MXjKzd2pfpza3mwDKNprT/l9KuvGU++6V9LK7L5T0cu17AKeRZPjdfY2kz065+xZJq2u3V0u6teR+AWiyet/zz3L33ZLk7rvNbGbRA81slaRVdf4cAE3S9A/83L1XUq8kmdnpPxsCOEPUO9S3x8xmS1Lta395XQLQCvWG/zlJK2u3V0p6tpzuAGiV5Gm/mT0h6TpJ082sT9KDkh6W9KSZ/UDSLkm3N7OTZ7rUvPVFixaF9RtuuKGwNmfOnLDt5MmTw3rqGoSOjo6wHl2jcPjw4bBtaqz97bffDuuRo0eP1t1WOjPm8yfD7+53FpS+XXJfALQQl/cCmSL8QKYIP5Apwg9kivADmWJKbwukhstSy2tfeeWVYb27u7uwlhqKSy2fvXz58rCe2qr6wIEDhbWPP/44bNvfH187lhpuSy3P3chznwk48gOZIvxApgg/kCnCD2SK8AOZIvxApgg/kCnG+VsgtQX3vHnzwvqyZcvCerRVdeoag1TfJkyYENZT026j8fIvv/wybLtgwYKwPnv27LDe2dlZWEtdY5ADjvxApgg/kCnCD2SK8AOZIvxApgg/kCnCD2SKcf4WSI2lp5bXHj9+fFj/7LNTt1L8k4GBgbBtypEjR8L64OBgWI/m+6fWAli8eHFYv/7668P6hg0bCmupdQ5S24OfCTjyA5ki/ECmCD+QKcIPZIrwA5ki/ECmCD+QKcb5S9DoNtap9eXfeuutsL5jx47C2gcffBC2fffdd8P6rl27wnpPT09Yv/nmmwtrF198cdg2dX3DZZddFtbPPffcsJ675JHfzB4zs34z2zLsvofM7EMz21T773vN7SaAso3mtP+Xkm4c4f5/d/eltf9eKLdbAJotGX53XyOp+PpRAKelRj7wu8vMNtfeFkwtepCZrTKz9Wa2voGfBaBk9Yb/55IWSFoqabeknxY90N173X2Zu8erUAJoqbrC7+573P24u5+Q9AtJ8VauANpOXeE3s+FrJt8maUvRYwG0p+Q4v5k9Iek6SdPNrE/Sg5KuM7OlklzS+5J+2MQ+tkS09r0Uj+WnxqPPP//8sD5//vywnlrfftOmTYW1NWvWhG337dsX1lPz+d97772wPnfu3MLapZdeGradNGlSWE+97tGc/NR+AzlIht/d7xzh7keb0BcALcTlvUCmCD+QKcIPZIrwA5ki/ECmzpgpvY1Oq00NG02cOLGwNmXKlLDt0qVLw/r06dPD+qeffhrW161bV1jbu3dv2PaLL74I69EW25J08ODBsD558uTC2nnnnRe2jV5zSdq/f39Yj353hvo48gPZIvxApgg/kCnCD2SK8AOZIvxApgg/kKlsxvlTY8apsfpoyu+CBQvCtkuWLAnr0bRXSXrttdfCejQtNzUlNzWOP25c/L9IV1dXWL/qqqsKa+ecc07Y9ujRo2F97dq1Yf3AgQOFNcb5OfID2SL8QKYIP5Apwg9kivADmSL8QKYIP5CpM2acf8yYxv4di+adS/F1BNFYtiRdffXVYX1gYCCsp5YV7+zsLKyl5tunlse+4oorwvr9998f1lO/e+TDDz8M688//3xYj5Y8T13fkAOO/ECmCD+QKcIPZIrwA5ki/ECmCD+QKcIPZGo0W3R3S3pc0vmSTkjqdfdHzGyapN9K6tHQNt3fd/d4v+cGRWPtqbHwY8eOhfXjx4+H9Wgsvbu7O2w7derUsJ6az5/awvumm24qrH3++ecNPfe8efPCeup3i/4uW7duDds+8MADYX3z5s1hPfqbM84/uiP/oKR/cPfFkq6S9CMzu1TSvZJedveFkl6ufQ/gNJEMv7vvdveNtdsHJW2X1CXpFkmraw9bLenWZnUSQPm+0Xt+M+uRdIWktZJmuftuaegfCEkzy+4cgOYZ9bX9ZjZF0lOSfuzuA6k184a1WyVpVX3dA9Asozrym9l4DQX/V+7+dO3uPWY2u1afLal/pLbu3uvuy9x9WRkdBlCOZPht6BD/qKTt7v6zYaXnJK2s3V4p6dnyuwegWUZz2r9C0l9KetPMNtXuu0/Sw5KeNLMfSNol6fbmdPFPouGZwcHBsG1qym9qWu2iRYsKa9OmTQvbpobDUtNqZ8yYEdYXLlxYWEstvZ2qp6SGzKLhvHvuuSds++qrr4b11PBsVGeobxThd/dXJRW9wf92ud0B0Cpc4QdkivADmSL8QKYIP5Apwg9kivADmTpjlu5OjfmmLkc+dOhQWN+7d29h7fXXXw/bpqbNpuqp7cU7OjoKa6mpzqnx7uj3ltLbh999992Ftb6+vrBt6m/KWH1jOPIDmSL8QKYIP5Apwg9kivADmSL8QKYIP5Apa+VYqZlVNjCbGudPzfePxtJnzoyXL5w1a1ZY7+rqCus9PT11t0/97G3btoX1V155Jaynlt+O1klgnL453H1Ua+xx5AcyRfiBTBF+IFOEH8gU4QcyRfiBTBF+IFPZjPM302i3LqtX6hqESOrv22gd7YdxfgAhwg9kivADmSL8QKYIP5Apwg9kivADmUqO85tZt6THJZ0v6YSkXnd/xMwekvQ3kk4u7H6fu7+QeC4GjYEmG+04/2jCP1vSbHffaGadkjZIulXS9yV97u7/NtpOEX6g+UYb/uSOPe6+W9Lu2u2DZrZdUrz0DIC2943e85tZj6QrJK2t3XWXmW02s8fMbGpBm1Vmtt7M1jfUUwClGvW1/WY2RdIrkn7i7k+b2SxJn0hySf+kobcGf514Dk77gSYr7T2/JJnZeEnPS/qDu/9shHqPpOfd/VuJ5yH8QJOVNrHHhqasPSpp+/Dg1z4IPOk2SVu+aScBVGc0n/b/uaT/kfSmhob6JOk+SXdKWqqh0/73Jf2w9uFg9Fwc+YEmK/W0vyyEH2g+5vMDCBF+IFOEH8gU4QcyRfiBTBF+IFOEH8gU4QcyRfiBTBF+IFOEH8gU4QcyRfiBTBF+IFPJBTxL9omk/x32/fTafe2oXfvWrv2S6Fu9yuzbvNE+sKXz+b/2w83Wu/uyyjoQaNe+tWu/JPpWr6r6xmk/kCnCD2Sq6vD3VvzzI+3at3btl0Tf6lVJ3yp9zw+gOlUf+QFUhPADmaok/GZ2o5n90cx2mNm9VfShiJm9b2ZvmtmmqvcXrO2B2G9mW4bdN83MXjKzd2pfR9wjsaK+PWRmH9Zeu01m9r2K+tZtZv9tZtvNbKuZ/X3t/kpfu6BflbxuLX/Pb2ZjJb0t6buS+iStk3Snu29raUcKmNn7kpa5e+UXhJjZNZI+l/T4ya3QzOxfJX3m7g/X/uGc6u7/2CZ9e0jfcNv2JvWtaFv5v1KFr12Z292XoYoj/3JJO9x9p7sflfQbSbdU0I+25+5rJH12yt23SFpdu71aQ//ztFxB39qCu+9294212wclndxWvtLXLuhXJaoIf5ekD4Z936cKX4ARuKQXzWyDma2qujMjmHVyW7Ta15kV9+dUyW3bW+mUbeXb5rWrZ7v7slUR/pG2Emqn8cYV7v5nkm6S9KPa6S1G5+eSFmhoD8fdkn5aZWdq28o/JenH7j5QZV+GG6FflbxuVYS/T1L3sO/nSvqogn6MyN0/qn3tl/SMht6mtJM9J3dIrn3tr7g//8/d97j7cXc/IekXqvC1q20r/5SkX7n707W7K3/tRupXVa9bFeFfJ2mhmV1oZh2S7pD0XAX9+Bozm1z7IEZmNlnSDWq/rcefk7SydnulpGcr7MtXtMu27UXbyqvi167dtruv5Aq/2lDGf0gaK+kxd/9JyzsxAjObr6GjvTQ03fnXVfbNzJ6QdJ2GpnzukfSgpN9JelLSBZJ2Sbrd3Vv+wVtB367TN9y2vUl9K9pWfq0qfO3K3O6+lP5weS+QJ67wAzJF+IFMEX4gU4QfyBThBzJF+IFMEX4gU/8H2IcOelYVSaYAAAAASUVORK5CYII=\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"
    }
   ],
   "source": [
    "def printdata(x,y):\n",
    "    plt.imshow(x,cmap = 'gray')\n",
    "    plt.title(y)\n",
    "    plt.show()\n",
    "for i in range(3):\n",
    "    printdata(decoded[i].view(28,28).cpu().data.numpy(),str(batch_y[i].cpu().numpy()))\n",
    "    printdata(batch_x[i].view(28,28).cpu().numpy(),str(batch_y[i].cpu().numpy()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## fine-tuning strategies"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 提取参数"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-24T08:23:22.400030Z",
     "start_time": "2019-09-24T08:23:22.377124Z"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "latent_features = 10\n",
    "\n",
    "class DSC(nn.Module):\n",
    "    def __init__(self,autoencoder,BATCH_SIZE):\n",
    "        super().__init__()\n",
    "        self.encoder = nn.Sequential(\n",
    "            nn.Linear(28*28,128),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(128,64),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(64,12),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(12,10),\n",
    "\n",
    "        )\n",
    "        self.encoder.load_state_dict(autoencoder.encoder.state_dict())\n",
    "        \n",
    "        kill_matrix = np.ones((BATCH_SIZE,BATCH_SIZE))\n",
    "        for i in range(BATCH_SIZE):\n",
    "            kill_matrix[i][i]=0\n",
    "        self.k = torch.tensor(kill_matrix,dtype=torch.float, requires_grad=False)\n",
    "        self.m = torch.zeros([BATCH_SIZE,BATCH_SIZE],dtype=torch.float ,requires_grad=True)\n",
    "        \n",
    "        self.decoder = nn.Sequential(\n",
    "            nn.Linear(10,12),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(12,64),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(64,128),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(128,28*28),\n",
    "            nn.Sigmoid()\n",
    "        )\n",
    "        self.decoder.load_state_dict(autoencoder.decoder.state_dict())\n",
    "        \n",
    "        ### //不训练某些层\n",
    "        frozen_layers = [self.encoder, self.decoder,]\n",
    "        for layer in frozen_layers:\n",
    "            for name, value in layer.named_parameters():\n",
    "                value.requires_grad = False\n",
    "        \n",
    "    def forward(self,x):\n",
    "        z = self.encoder(x)\n",
    "        z = torch.transpose(z, 1, 0)\n",
    "        c = self.k.mul(self.m)\n",
    "        z_ = z.mm(c)\n",
    "        x_ = torch.transpose(z_, 1, 0)\n",
    "        x_ = self.decoder(x_)\n",
    "        return z,z_,x_,c"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 开始炼丹"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-24T08:23:25.675273Z",
     "start_time": "2019-09-24T08:23:25.625406Z"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\torchvision\\datasets\\mnist.py:53: UserWarning: train_data has been renamed data\n",
      "  warnings.warn(\"train_data has been renamed data\")\n",
      "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\torchvision\\datasets\\mnist.py:43: UserWarning: train_labels has been renamed targets\n",
      "  warnings.warn(\"train_labels has been renamed targets\")\n"
     ]
    }
   ],
   "source": [
    "DOWNLOAD_MNIST = False\n",
    "\n",
    "train_data = torchvision.datasets.MNIST(\n",
    "    root = './mnist',\n",
    "    train = True,\n",
    "    transform = torchvision.transforms.ToTensor(), #从0-255压缩到0-1\n",
    "    download =DOWNLOAD_MNIST\n",
    ")\n",
    "\n",
    "# 先转换成 torch 能识别的 Dataset\n",
    "torch_dataset = Data.TensorDataset(train_data.train_data[:BATCH_SIZE], train_data.train_labels[:BATCH_SIZE])\n",
    "\n",
    "# 把 dataset 放入 DataLoader\n",
    "loader = Data.DataLoader(\n",
    "    dataset=torch_dataset,      # torch TensorDataset format\n",
    "    batch_size=BATCH_SIZE,      # mini batch size\n",
    "    shuffle=False,               # 要不要打乱数据 (打乱比较好)\n",
    "    num_workers=2,              # 多线程来读数据\n",
    ")\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-24T10:09:03.684056Z",
     "start_time": "2019-09-24T10:05:50.715311Z"
    }
   },
   "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"
    }
   ],
   "source": [
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.utils.data as Data\n",
    "import torchvision\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "K = 10\n",
    "EPOCH = 50+25*K\n",
    "BATCH_SIZE = 64\n",
    "LR = 0.01\n",
    "DOWNLOAD_MNIST = False\n",
    "\n",
    "train_data = torchvision.datasets.MNIST(\n",
    "    root = './mnist',\n",
    "    train = True,\n",
    "    transform = torchvision.transforms.ToTensor(), #从0-255压缩到0-1\n",
    "    download =DOWNLOAD_MNIST\n",
    ")\n",
    "\n",
    "# 先转换成 torch 能识别的 Dataset\n",
    "torch_dataset = Data.TensorDataset(train_data.train_data[:BATCH_SIZE], train_data.train_labels[:BATCH_SIZE])\n",
    "\n",
    "# 把 dataset 放入 DataLoader\n",
    "loader = Data.DataLoader(\n",
    "    dataset=torch_dataset,      # torch TensorDataset format\n",
    "    batch_size=BATCH_SIZE,      # mini batch size\n",
    "    shuffle=False,               # 要不要打乱数据 (打乱比较好)\n",
    "    num_workers=2,              # 多线程来读数据\n",
    ")\n",
    "\n",
    "\n",
    "dsc = DSC(autoencoder,BATCH_SIZE)\n",
    "dsc\n",
    "\n",
    "\n",
    "optimizer = torch.optim.Adam([dsc.m],lr = LR,betas = (0.9,0.99),weight_decay=1e-5)    \n",
    "loss_func = torch.nn.MSELoss()\n",
    "\n",
    "loss_list = [] \n",
    "lambda1 = 1\n",
    "lambda2 = 10**(K/10-3)\n",
    "\n",
    "steps = []\n",
    "t = 0\n",
    "for epoch in range(EPOCH):   # 训练所有!整套!数据 EPOCH 次\n",
    "    for step, (batch_x, batch_y) in enumerate(loader):  # 每一步 loader 释放一小批数据用来学习\n",
    "        t+=1\n",
    "        steps.append(t)\n",
    "        batch_x = (batch_x.view(-1,28*28).type(torch.FloatTensor)/255)\n",
    "        batch_y = batch_y\n",
    "        \n",
    "        batch_y,index_ = batch_y.sort()\n",
    "        batch_x = batch_x[index_]\n",
    "        \n",
    "        z,z_,x_,c = dsc(batch_x)\n",
    "        \n",
    "        \n",
    "        \n",
    "        loss = 1/2*torch.norm(x_-batch_x)**2+lambda1*torch.norm(c)+\\\n",
    "        lambda2/2*torch.norm(z-z_)**2\n",
    "        \n",
    "        loss_list.append(loss)\n",
    "        \n",
    "        optimizer.zero_grad()      #初始化梯度\n",
    "        loss.backward()         #计算梯度\n",
    "        optimizer.step()        #对参数学习\n",
    "\n",
    "\n",
    "        import IPython\n",
    "        IPython.display.clear_output(wait=True)\n",
    "        plt.plot(steps[-30:],loss_list[-30:])\n",
    "        plt.ylabel('loss')\n",
    "        plt.xlabel('train step')\n",
    "\n",
    "        plt.show()\n",
    "\n",
    "\n",
    "C = c.data.numpy()\n",
    "plt.imshow(C,cmap = 'Greys')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-24T10:09:50.588935Z",
     "start_time": "2019-09-24T10:09:50.109948Z"
    },
    "scrolled": false
   },
   "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"
    }
   ],
   "source": [
    "def printdata(x,y):\n",
    "    plt.imshow(x,cmap = 'gray')\n",
    "    plt.title(y)\n",
    "    plt.show()\n",
    "for i in range(3):\n",
    "    printdata(x_[i].view(28,28).cpu().data.numpy(),str(batch_y[i].cpu().numpy()))\n",
    "    printdata(batch_x[i].view(28,28).cpu().numpy(),str(batch_y[i].cpu().numpy()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-24T10:09:58.777789Z",
     "start_time": "2019-09-24T10:09:58.751894Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([[ 1.20576978e+00,  2.66972721e-01,  3.83623391e-02,\n",
       "         -4.57601517e-01, -5.09580970e-02,  1.13464499e+00,\n",
       "         -8.31992984e-01, -1.40674603e+00, -9.70266521e-01,\n",
       "         -1.08433700e+00, -1.60010302e+00, -9.58125114e-01,\n",
       "         -1.34503126e+00, -1.12922454e+00, -1.88359007e-01,\n",
       "          3.49861884e+00,  1.34379268e-02,  4.44763690e-01,\n",
       "          7.17804193e-01,  8.03960681e-01,  7.37778962e-01,\n",
       "          2.66724920e+00,  9.26708877e-01,  1.24766326e+00,\n",
       "          7.32824624e-01,  1.59421110e+00, -3.46159548e-01,\n",
       "         -1.54587662e+00,  2.00167418e+00,  8.80079567e-01,\n",
       "          4.91081774e-01, -8.43377888e-01, -2.19111276e+00,\n",
       "         -1.56400949e-01, -1.61205411e+00, -3.74720931e-01,\n",
       "          8.71458173e-01, -1.64163733e+00, -7.37363875e-01,\n",
       "         -1.28580141e+00, -1.37574649e+00, -2.31410074e+00,\n",
       "         -1.25875688e+00, -1.06051290e+00, -9.86639678e-01,\n",
       "         -8.36536467e-01,  1.57057926e-01, -8.88472795e-01,\n",
       "          1.71555385e-01,  1.18819916e+00,  4.03335951e-02,\n",
       "          3.51119787e-01,  3.15987498e-01,  2.13724986e-01,\n",
       "          3.28801543e-01, -6.87676549e-01,  2.79107779e-01,\n",
       "         -5.26402056e-01, -1.43812180e+00, -7.34946281e-02,\n",
       "          1.46607292e+00, -4.43791062e-01, -9.58267599e-02,\n",
       "         -1.24433887e+00],\n",
       "        [-1.53637934e+00, -3.26452911e-01, -1.44210780e+00,\n",
       "          3.61863188e-02, -1.45871028e-01, -7.72992134e-01,\n",
       "         -8.87607872e-01, -3.24674547e-01, -1.12445906e-01,\n",
       "         -4.58072305e-01, -2.07485124e-01, -3.40234429e-01,\n",
       "          6.94968691e-03, -3.54889125e-01,  9.35774207e-01,\n",
       "          2.35551095e+00,  1.03669906e+00,  1.48135662e-01,\n",
       "          9.75225985e-01,  1.29655913e-01,  5.58992922e-01,\n",
       "          4.92419511e-01, -1.10127680e-01,  2.19131798e-01,\n",
       "          7.24888504e-01,  1.35333312e+00,  7.92553723e-01,\n",
       "         -9.36971724e-01, -1.82500422e+00, -4.05021071e-01,\n",
       "          5.91606498e-01, -1.25745118e+00, -2.88882399e+00,\n",
       "          3.03133875e-01, -1.06892681e+00, -1.08105803e+00,\n",
       "          1.42980084e-01,  3.54400635e-01, -1.74360573e-01,\n",
       "          1.26469567e-01,  3.97091329e-01, -9.99127507e-01,\n",
       "          1.12590510e-02, -7.82069623e-01,  9.37135890e-03,\n",
       "         -1.93295979e+00, -9.87101555e-01,  8.35053772e-02,\n",
       "          2.43190885e-01, -1.30707574e+00, -5.10589778e-01,\n",
       "         -1.94535398e+00, -3.78996164e-01, -6.79644287e-01,\n",
       "         -3.46225232e-01, -1.92661834e+00, -2.69203043e+00,\n",
       "         -2.23360610e+00, -2.66669178e+00, -3.36359769e-01,\n",
       "         -2.66594744e+00, -7.88027346e-01, -1.76830329e-02,\n",
       "         -8.01376164e-01],\n",
       "        [ 2.64835477e+00,  2.90864992e+00,  4.28994685e-01,\n",
       "          8.03604126e-01,  8.50472987e-01,  2.12450075e+00,\n",
       "          5.06810367e-01, -2.83824682e+00, -9.29626405e-01,\n",
       "         -4.30528224e-01, -4.39169824e-01, -1.46659124e+00,\n",
       "         -2.58811116e+00, -3.07650113e+00, -3.92622232e-01,\n",
       "          2.20631337e+00,  2.49598131e-01,  1.86313823e-01,\n",
       "          4.40987408e-01, -4.43532199e-01,  8.60524535e-01,\n",
       "         -1.76481470e-01, -1.29603297e-01, -2.24995598e-01,\n",
       "         -6.58290207e-01,  9.89546120e-01,  5.70179045e-01,\n",
       "         -1.74436557e+00,  1.71401107e+00,  1.82448590e+00,\n",
       "          1.01318729e+00,  1.82141550e-02, -8.82256210e-01,\n",
       "          1.11263692e+00, -1.32262063e+00, -1.62484503e+00,\n",
       "         -9.21248198e-02, -9.09131646e-01, -1.57656264e+00,\n",
       "          1.18302524e+00,  1.14988470e+00,  1.47366774e+00,\n",
       "          7.58784771e-01,  6.54255688e-01,  1.04570341e+00,\n",
       "         -6.91512227e-01, -5.50657153e-01, -7.83704162e-01,\n",
       "          2.09330663e-01,  9.65028405e-01, -9.79222655e-01,\n",
       "         -1.62185287e+00, -1.60680473e+00, -6.01254225e-01,\n",
       "         -8.00916851e-02, -2.16640067e+00, -3.33631128e-01,\n",
       "          2.64200687e-01, -3.06745076e+00, -1.33377838e+00,\n",
       "          1.11900330e+00, -6.82639301e-01,  4.85281646e-01,\n",
       "         -7.99498558e-01],\n",
       "        [-4.41689610e-01,  1.20135093e+00, -1.43017936e+00,\n",
       "          2.94523060e-01,  4.08277631e-01,  5.54685473e-01,\n",
       "         -5.22367418e-01,  7.60558844e-01,  2.07115728e-02,\n",
       "          2.42932916e-01, -1.39947391e+00,  3.26372415e-01,\n",
       "          1.51791382e+00,  1.15226865e+00,  1.79477885e-01,\n",
       "          1.96269825e-01, -9.53857601e-01, -6.89081609e-01,\n",
       "         -9.13116708e-02, -6.87426865e-01, -4.56654746e-03,\n",
       "         -1.25129759e+00,  3.45556468e-01, -4.90070283e-01,\n",
       "         -1.50166881e+00, -7.93142140e-01, -6.94697976e-01,\n",
       "         -3.32228124e-01,  1.49739265e-01,  3.25790524e-01,\n",
       "          6.33017540e-01,  3.38078260e-01, -1.58111608e+00,\n",
       "          7.11130142e-01, -1.75727338e-01,  9.54844594e-01,\n",
       "         -2.51816642e-02,  9.24648702e-01,  8.62500012e-01,\n",
       "         -4.76209491e-01, -5.83075106e-01, -2.15512919e+00,\n",
       "          1.34316593e-01, -6.10531867e-01, -1.35143626e+00,\n",
       "          1.54401565e+00,  8.60791951e-02,  2.20718801e-01,\n",
       "          1.88643485e-01,  8.11999202e-01,  1.09634876e-01,\n",
       "          4.14274275e-01, -5.86182773e-02, -3.21635664e-01,\n",
       "          5.19856930e-01, -9.20138597e-01,  6.24632478e-01,\n",
       "          1.01631474e+00, -5.50679505e-01, -1.10332131e+00,\n",
       "         -6.74622059e-01,  2.40310550e-01, -1.53346602e-02,\n",
       "          3.30040157e-02],\n",
       "        [ 7.41578579e-01,  6.43011406e-02,  1.12837005e+00,\n",
       "          5.27395494e-03,  1.88177422e-01, -4.19344664e-01,\n",
       "          5.83835065e-01,  1.43654382e+00, -9.77280736e-02,\n",
       "         -5.64848125e-01,  1.19709365e-01,  1.05044985e+00,\n",
       "          1.59211969e+00,  1.60402167e+00,  1.70538172e-01,\n",
       "         -1.58245051e+00,  2.43444607e-01,  7.64669478e-01,\n",
       "          1.07734084e+00,  1.49668291e-01, -8.05985749e-01,\n",
       "         -8.19481909e-01,  4.41834778e-02,  2.39883050e-01,\n",
       "         -6.30809844e-01, -3.68043929e-01, -2.60184675e-01,\n",
       "         -3.45288128e-01,  7.69951284e-01, -1.84810951e-01,\n",
       "         -4.15973812e-01,  6.24444783e-01,  9.80769575e-01,\n",
       "         -1.82144630e+00,  1.13188612e+00,  3.60486686e-01,\n",
       "         -6.73030317e-01,  8.69706273e-01,  7.35295177e-01,\n",
       "         -1.25755265e-01, -2.98802048e-01, -1.32798240e-01,\n",
       "         -3.40021193e-01, -2.96966642e-01, -3.31500433e-02,\n",
       "         -4.75336127e-02,  4.65570390e-01,  1.41013384e+00,\n",
       "         -6.04635477e-01, -2.39416933e+00,  1.01916957e+00,\n",
       "         -2.48706192e-02,  1.20725012e+00,  4.50819910e-01,\n",
       "          9.67718840e-01,  3.19457918e-01, -4.77407038e-01,\n",
       "          1.74123645e-01,  6.97485268e-01,  1.11682773e+00,\n",
       "          4.12907749e-01, -2.28069261e-01, -1.88960209e-01,\n",
       "          1.59164876e-01],\n",
       "        [-3.10501695e+00, -1.06418109e+00, -1.26612782e+00,\n",
       "         -1.20259917e+00, -1.31597459e+00, -1.57932854e+00,\n",
       "         -1.34545851e+00,  7.94509172e-01, -2.49727458e-01,\n",
       "          2.00352490e-01,  5.53584576e-01,  1.18848872e+00,\n",
       "          8.88490021e-01,  1.32936442e+00, -4.57847536e-01,\n",
       "          9.76855218e-01,  3.32487524e-01, -3.52548927e-01,\n",
       "          6.22779489e-01, -8.45891312e-02, -6.43273890e-01,\n",
       "          2.49311849e-01,  2.07798317e-01,  4.56946313e-01,\n",
       "          2.07124695e-01,  5.98459721e-01,  7.81880766e-02,\n",
       "         -6.54424489e-01,  1.27969003e+00,  8.74073148e-01,\n",
       "         -3.90658081e-01,  1.38344228e+00, -1.30460978e+00,\n",
       "          3.03815812e-01, -2.06806168e-01, -1.02789807e+00,\n",
       "         -1.60556287e-01, -8.19075644e-01, -1.12261367e+00,\n",
       "          1.17695093e-01,  1.14687815e-01,  1.06188071e+00,\n",
       "          1.44112945e-01,  2.30062613e-03, -1.04755811e-01,\n",
       "          1.24085546e+00, -6.08109713e-01, -3.03971898e-02,\n",
       "          2.93651730e-01, -1.13032952e-01, -1.29882634e-01,\n",
       "          1.41727340e+00, -3.42151374e-01,  4.72365499e-01,\n",
       "          7.63570905e-01,  1.45702207e+00,  2.62833714e-01,\n",
       "          1.27544081e+00, -3.09168428e-01, -5.53236127e-01,\n",
       "          8.61956656e-01, -4.69603948e-03,  7.77259767e-01,\n",
       "          4.40159082e-01],\n",
       "        [ 1.85111022e+00,  1.02466512e+00, -1.83053792e-01,\n",
       "         -5.24419844e-01, -7.14983344e-02,  7.08348036e-01,\n",
       "         -2.50701249e-01, -5.97667098e-01, -2.49935538e-01,\n",
       "          1.61786377e-02,  8.70760679e-01, -4.96860221e-02,\n",
       "         -1.28511071e+00, -9.95024323e-01,  5.68405330e-01,\n",
       "         -3.25531173e+00, -9.51934040e-01, -4.21411574e-01,\n",
       "          7.03984797e-01,  7.86031067e-01,  1.17059767e+00,\n",
       "          8.68298948e-01,  6.62951112e-01,  4.65269715e-01,\n",
       "          9.12378907e-01,  2.09568501e+00,  4.01887178e-01,\n",
       "         -8.85115504e-01,  6.27390623e-01,  4.31420982e-01,\n",
       "         -8.63632381e-01, -1.25357056e+00, -2.93576384e+00,\n",
       "          7.58220926e-02, -1.27880490e+00,  2.65064806e-01,\n",
       "          9.35754031e-02, -1.04788685e+00,  3.72882277e-01,\n",
       "         -3.42676602e-02,  2.28498831e-01,  2.18568778e+00,\n",
       "          8.91303837e-01,  1.70900738e+00, -1.27015993e-01,\n",
       "         -1.29722655e+00, -1.72161496e+00, -1.78967047e+00,\n",
       "         -4.20879066e-01, -2.15989327e+00, -3.99425000e-01,\n",
       "         -1.11498415e+00, -7.75636315e-01,  4.98645693e-01,\n",
       "          1.33001283e-01, -2.01372886e+00, -1.32651258e+00,\n",
       "         -7.70484805e-01, -2.78165889e+00, -1.64439285e+00,\n",
       "         -2.55370879e+00, -8.32349241e-01,  4.49993700e-01,\n",
       "         -1.26906574e+00],\n",
       "        [ 1.45974815e+00,  6.57970548e-01,  5.91422021e-01,\n",
       "         -4.65693414e-01,  3.11934426e-02,  2.85756320e-01,\n",
       "         -1.62986085e-01,  1.09339179e-02, -2.65607715e-01,\n",
       "         -2.62695789e-01,  5.51554263e-02, -1.26724109e-01,\n",
       "          2.98893452e-01,  1.69044629e-01, -6.04048848e-01,\n",
       "         -5.15941441e-01,  2.71337211e-01,  5.05611956e-01,\n",
       "          2.63329625e-01,  1.15278743e-01,  3.81775677e-01,\n",
       "         -1.95782155e-01,  2.12574482e-01,  9.28035006e-02,\n",
       "         -3.62556219e-01, -8.64410866e-03, -8.46366405e-01,\n",
       "         -1.48440581e-02,  2.38430977e+00,  1.30157626e+00,\n",
       "          9.05478060e-01,  4.36699092e-01,  4.74985123e-01,\n",
       "          3.10067028e-01,  2.73911685e-01, -3.64027023e-01,\n",
       "         -3.68208647e-01,  2.61862874e-01,  3.27270091e-01,\n",
       "         -1.57124564e-01,  1.20637722e-01,  3.01630169e-01,\n",
       "         -3.47308248e-01,  7.36891776e-02,  1.29243010e-03,\n",
       "          3.31659287e-01,  2.99322188e-01,  1.25254020e-01,\n",
       "          3.97898287e-01, -4.27846283e-01,  1.13533534e-01,\n",
       "         -3.82536314e-02, -2.54003316e-01, -4.09882069e-02,\n",
       "          2.81475008e-01, -2.38438517e-01,  3.76699179e-01,\n",
       "          6.51694536e-01, -2.13126957e-01,  7.55274668e-02,\n",
       "          1.05281961e+00,  1.07844032e-01, -7.52443016e-01,\n",
       "         -7.12221190e-02],\n",
       "        [ 1.77383792e+00,  5.15362680e-01,  1.47449529e+00,\n",
       "          9.23164308e-01,  8.11702311e-01,  9.64652777e-01,\n",
       "          1.34450066e+00,  4.25562173e-01, -2.59813070e+00,\n",
       "         -2.49037910e+00, -4.18392038e+00, -1.73102736e+00,\n",
       "          1.87492892e-01,  3.79689783e-01, -3.73100370e-01,\n",
       "          5.68507671e-01,  2.11024135e-01,  7.16932058e-01,\n",
       "          7.92723775e-01,  4.02009755e-01, -1.18807995e+00,\n",
       "          1.03271997e+00, -1.04666984e+00,  1.40211582e+00,\n",
       "         -1.23355523e-01, -9.21891928e-01,  1.95613548e-01,\n",
       "         -3.83399010e-01,  1.50982842e-01,  1.77649871e-01,\n",
       "         -1.29929483e-01, -1.44371533e+00,  1.31478572e+00,\n",
       "         -5.43268740e-01,  2.70901799e-01,  7.94134319e-01,\n",
       "          2.40477353e-01,  2.28683084e-01,  5.91673292e-02,\n",
       "          9.63128328e-01, -3.20639580e-01,  1.22484408e-01,\n",
       "          1.07403204e-01,  5.98123848e-01,  1.75284415e-01,\n",
       "         -2.14401245e+00, -1.56231713e-03,  7.05310330e-02,\n",
       "         -9.22997773e-01, -1.00409710e+00,  7.42404997e-01,\n",
       "          1.39740169e+00,  1.49403310e+00,  8.28097284e-01,\n",
       "         -2.94080973e-01, -3.82588416e-01, -3.93927842e-01,\n",
       "         -8.54224145e-01, -6.92209750e-02,  2.08615765e-01,\n",
       "          1.02582455e-01, -2.15107217e-01, -6.50059700e-01,\n",
       "          2.03124478e-01],\n",
       "        [ 1.52254030e-01,  6.61334217e-01,  1.02505937e-01,\n",
       "          7.84455061e-01,  3.51903945e-01,  1.05365729e+00,\n",
       "          1.92704499e-01, -3.72643441e-01,  3.75697970e-01,\n",
       "          6.11036956e-01, -6.50737062e-02,  6.55276537e-01,\n",
       "         -8.74697983e-01, -4.50317204e-01, -6.11528873e-01,\n",
       "          6.39576256e-01, -1.91306844e-01, -7.08935499e-01,\n",
       "         -1.38561890e-01,  4.35051411e-01, -7.24567294e-01,\n",
       "          5.83073735e-01, -1.98783316e-02,  4.83738959e-01,\n",
       "         -1.10753544e-01,  9.19252694e-01,  9.23968613e-01,\n",
       "         -1.15204275e+00, -2.22409892e+00, -1.54198015e+00,\n",
       "         -7.41932333e-01,  4.23220217e-01, -1.42543197e+00,\n",
       "         -1.90805882e-01, -7.25302041e-01,  2.42202934e-02,\n",
       "          4.19550855e-03, -7.04881430e-01, -5.00047445e-01,\n",
       "         -9.56675649e-01, -6.76625967e-02,  2.33991548e-01,\n",
       "          6.16010964e-01, -2.83120692e-01, -2.10817292e-01,\n",
       "          8.24909747e-01, -3.81851435e-01, -3.50119054e-01,\n",
       "         -7.04626381e-01,  6.35532439e-01,  7.32299313e-02,\n",
       "         -4.15704399e-01,  4.01315331e-01, -9.12481397e-02,\n",
       "          4.33694333e-01,  2.26282135e-01,  1.31649658e-01,\n",
       "         -2.70260107e-02,  2.89101183e-01, -4.41871345e-01,\n",
       "          2.88382947e-01, -2.22098440e-01,  4.31008130e-01,\n",
       "          5.96854053e-02]], dtype=float32),\n",
       " array([[ 3.41718346e-01,  4.22755741e-02, -5.34886420e-01,\n",
       "         -2.06218421e-01,  7.41492659e-02,  8.88241589e-01,\n",
       "         -6.11987948e-01, -8.13162208e-01, -8.67097795e-01,\n",
       "         -5.68636954e-01, -7.27556407e-01, -8.51799488e-01,\n",
       "         -8.73999953e-01, -9.03976142e-01, -9.42166269e-01,\n",
       "          1.10270131e+00, -8.96947533e-02,  1.20155722e-01,\n",
       "          3.51196885e-01,  5.05135298e-01,  6.14153922e-01,\n",
       "          1.12975633e+00,  5.43587387e-01,  7.91878760e-01,\n",
       "         -7.28624314e-03,  7.45334327e-01, -2.10898384e-01,\n",
       "         -9.96496379e-01,  7.22119212e-01,  6.82471454e-01,\n",
       "          5.63415170e-01, -1.25748411e-01, -8.29806268e-01,\n",
       "          5.03456414e-01, -1.04323041e+00, -1.94250029e-02,\n",
       "          3.16854984e-01, -5.96258998e-01,  8.44098628e-04,\n",
       "         -1.39654660e+00, -1.56400502e+00, -1.09747136e+00,\n",
       "         -1.46173215e+00, -1.16036904e+00, -1.45367002e+00,\n",
       "         -6.34906173e-01,  2.55640060e-01, -1.42464626e+00,\n",
       "          2.15452269e-01,  5.84112227e-01, -4.25473541e-01,\n",
       "          1.43759817e-01, -3.83295566e-02, -3.13475847e-01,\n",
       "          3.84521425e-01, -5.55281579e-01,  8.51181448e-02,\n",
       "         -2.15215340e-01, -5.59164584e-01, -2.63682127e-01,\n",
       "          8.87984157e-01, -4.25674021e-01, -4.12819445e-01,\n",
       "         -9.01401162e-01],\n",
       "        [ 1.06647268e-01,  2.45140135e-01, -4.09668833e-01,\n",
       "         -8.95880684e-02, -7.27093145e-02, -4.16963309e-01,\n",
       "         -3.16507146e-02,  1.30536675e-01,  8.25130194e-02,\n",
       "         -6.22113869e-02,  3.52787748e-02,  1.07944906e-02,\n",
       "         -6.75670728e-02, -5.80408499e-02,  7.55248010e-01,\n",
       "          4.36551094e-01,  1.09886646e+00,  1.79575682e-01,\n",
       "          8.19212079e-01,  3.29497039e-01,  4.44863170e-01,\n",
       "          6.62547171e-01,  8.39023143e-02,  6.56069741e-02,\n",
       "          2.22019076e-01,  6.96216047e-01,  8.36209834e-01,\n",
       "         -5.67907393e-01, -8.90831709e-01, -4.76412848e-02,\n",
       "          5.61872244e-01, -1.10592532e+00, -5.48050046e-01,\n",
       "         -3.37617338e-01, -6.98956013e-01, -7.23759413e-01,\n",
       "         -8.89682397e-02,  2.80550599e-01,  2.04108879e-02,\n",
       "          4.03516054e-01,  8.47369879e-02,  6.54169917e-01,\n",
       "         -1.88838422e-01, -1.86536938e-01, -1.58785909e-01,\n",
       "         -1.07482421e+00, -1.03002524e+00, -2.44075239e-01,\n",
       "          1.35308474e-01, -5.99993467e-01, -7.51372218e-01,\n",
       "         -8.55324447e-01, -2.34411657e-01, -6.92938924e-01,\n",
       "         -8.69269550e-01, -9.85275269e-01, -1.57754421e+00,\n",
       "         -1.19909322e+00, -6.01614416e-01, -1.65387064e-01,\n",
       "         -8.79894137e-01, -1.31956792e+00, -1.83292240e-01,\n",
       "         -8.41044188e-01],\n",
       "        [ 1.10557234e+00,  1.67449927e+00,  1.74225345e-02,\n",
       "          1.03319848e+00,  7.32962370e-01,  9.50570762e-01,\n",
       "          4.21901941e-01, -1.71621895e+00, -7.45535731e-01,\n",
       "         -5.15958190e-01, -2.45596573e-01, -1.28962314e+00,\n",
       "         -1.84991014e+00, -1.86041045e+00, -4.11846936e-01,\n",
       "          6.57745481e-01,  3.06376696e-01,  1.42549306e-01,\n",
       "          1.57869652e-01, -5.75327158e-01,  8.48332167e-01,\n",
       "         -8.82854462e-02,  5.52257560e-02, -4.29404795e-01,\n",
       "         -4.76715922e-01,  3.80316913e-01,  4.20895755e-01,\n",
       "         -1.03849494e+00,  1.23583746e+00,  1.21150184e+00,\n",
       "          1.09515142e+00,  8.43190402e-03, -3.51176023e-01,\n",
       "          1.64531612e+00, -9.20648456e-01, -1.31117594e+00,\n",
       "         -4.56493616e-01, -5.30010521e-01, -1.05896258e+00,\n",
       "          1.46186161e+00,  1.34376228e+00,  4.70771015e-01,\n",
       "          1.10985374e+00,  5.71684957e-01,  1.41789067e+00,\n",
       "         -9.85049903e-02, -7.25190639e-01, -6.97963834e-01,\n",
       "          2.06633449e-01,  8.54707599e-01, -1.24595094e+00,\n",
       "         -8.84679437e-01, -1.19566059e+00, -3.98945510e-01,\n",
       "          1.68197364e-01, -1.06569982e+00, -2.80519962e-01,\n",
       "          2.57308185e-01, -1.24782681e+00, -1.45580459e+00,\n",
       "          1.48576006e-01, -8.15264106e-01,  6.70832753e-01,\n",
       "         -8.28230262e-01],\n",
       "        [ 1.15954190e-01,  7.38097668e-01, -3.49341542e-01,\n",
       "          9.04901624e-02,  3.62759888e-01,  5.12144685e-01,\n",
       "         -2.66854912e-01,  2.91071773e-01, -7.73216188e-02,\n",
       "         -2.24635631e-01, -3.59087318e-01,  3.47093940e-01,\n",
       "          5.17990351e-01,  4.67174858e-01, -4.24579769e-01,\n",
       "         -2.86670625e-02, -5.20042896e-01, -6.07591629e-01,\n",
       "         -2.31651872e-01, -7.74003267e-01, -2.54243344e-01,\n",
       "         -5.95768332e-01, -3.42723340e-01, -7.96699405e-01,\n",
       "         -1.13232780e+00, -2.31785625e-01, -4.77597684e-01,\n",
       "         -1.17166340e-01, -6.73532784e-02,  1.80899680e-01,\n",
       "          7.23218799e-01, -2.34591395e-01, -3.87965411e-01,\n",
       "          2.18901768e-01, -1.89937204e-01,  1.18671790e-01,\n",
       "         -9.02425408e-01,  7.69292831e-01, -1.02860212e-01,\n",
       "         -5.52495360e-01, -5.86298108e-01, -7.99914122e-01,\n",
       "         -4.28470939e-01, -2.45800048e-01, -7.21025109e-01,\n",
       "          2.43163988e-01,  4.15900350e-03,  2.13538393e-01,\n",
       "          1.67981684e-02,  1.31228358e-01, -2.40391463e-01,\n",
       "          2.45309189e-01,  1.01308331e-01, -6.63450122e-01,\n",
       "         -5.84311187e-02, -4.04492646e-01, -1.71527833e-01,\n",
       "          1.82881221e-01, -3.87517482e-01, -5.84873080e-01,\n",
       "         -1.99533552e-01, -1.59120589e-01, -2.37230569e-01,\n",
       "         -2.56649345e-01],\n",
       "        [-6.90095872e-02,  3.33694786e-01, -2.64696836e-01,\n",
       "         -1.81558415e-01,  3.19593877e-01,  9.31505412e-02,\n",
       "         -2.86977112e-01,  7.45600700e-01, -4.28953767e-03,\n",
       "         -1.81065604e-01, -2.23117694e-01,  3.68232369e-01,\n",
       "          8.71444166e-01,  8.46673012e-01,  4.74808663e-02,\n",
       "         -4.08476114e-01, -2.48549089e-01,  1.97953641e-01,\n",
       "          7.47092009e-01, -1.42100200e-01, -4.56301868e-01,\n",
       "         -9.31150764e-02, -1.51477531e-01, -5.33890873e-02,\n",
       "         -3.46812010e-01, -5.91068715e-02, -3.21795523e-01,\n",
       "          5.75225949e-01, -3.70376885e-01, -2.05228329e-02,\n",
       "         -1.16656765e-01,  3.41601342e-01,  1.15430787e-01,\n",
       "         -6.28145099e-01,  5.33068895e-01,  8.63550246e-01,\n",
       "         -1.79688469e-01,  7.51413643e-01,  4.89771962e-01,\n",
       "         -1.04989871e-01,  6.97959960e-03, -2.22656652e-01,\n",
       "          9.61696431e-02,  8.71647373e-02, -4.11232054e-01,\n",
       "          1.15360238e-01,  6.11091405e-02,  3.20463628e-01,\n",
       "          1.34722680e-01, -1.00148737e+00,  7.48626888e-01,\n",
       "          6.39903545e-01,  7.07747698e-01,  1.85856059e-01,\n",
       "          4.98793662e-01,  3.17672431e-01,  2.13101119e-01,\n",
       "          2.16848627e-01,  3.51167440e-01,  2.23911777e-01,\n",
       "         -2.66025186e-01,  4.71688867e-01, -2.45574847e-01,\n",
       "          3.55377913e-01],\n",
       "        [-1.39738119e+00, -4.36093509e-01, -5.79220176e-01,\n",
       "         -1.06693339e+00, -1.54484713e+00, -9.82826531e-01,\n",
       "         -8.19679677e-01,  9.47337389e-01,  2.11863965e-01,\n",
       "         -1.32893920e-01,  2.97114909e-01,  9.94575858e-01,\n",
       "          7.25727439e-01,  7.48325408e-01, -3.86389017e-01,\n",
       "          8.77806246e-01,  7.19628036e-01,  3.86266232e-01,\n",
       "          6.07025862e-01, -3.48192334e-01,  3.74607146e-01,\n",
       "          2.33253822e-01,  7.38068938e-01,  9.01021957e-02,\n",
       "          5.60516194e-02,  5.82463503e-01, -9.47479010e-02,\n",
       "         -4.87305939e-01,  4.90320921e-01,  6.26800418e-01,\n",
       "         -2.74646342e-01,  9.98904943e-01, -3.18889320e-01,\n",
       "          4.03255403e-01, -1.29402548e-01, -6.10594451e-01,\n",
       "         -1.90669045e-01, -4.92995977e-01, -8.92024398e-01,\n",
       "         -1.97604850e-01,  7.83426166e-02,  4.45059538e-02,\n",
       "         -7.41694421e-02, -9.95739400e-02,  6.15620613e-03,\n",
       "          2.44071156e-01, -7.33394206e-01, -4.70228791e-01,\n",
       "          5.19550681e-01, -4.03852820e-01, -2.24167988e-01,\n",
       "          7.23715425e-02, -2.08692893e-01,  5.01895130e-01,\n",
       "          1.15943336e+00,  4.49223459e-01,  2.95482606e-01,\n",
       "          6.21937752e-01,  1.28915071e-01, -5.12221217e-01,\n",
       "          5.61199963e-01,  1.12802565e-01,  6.99729025e-01,\n",
       "          3.51895511e-01],\n",
       "        [ 7.20200419e-01,  5.21494865e-01, -5.38027287e-03,\n",
       "          7.84482062e-03,  6.09184027e-01, -8.23111683e-02,\n",
       "          1.07386082e-01, -5.59995532e-01, -1.43242881e-01,\n",
       "         -1.82529926e-01,  5.08223474e-02, -1.23277202e-01,\n",
       "         -1.06794381e+00, -1.00012326e+00,  4.26361650e-01,\n",
       "         -7.40256548e-01, -5.97633600e-01, -2.13949978e-01,\n",
       "          7.90008307e-01,  7.10333228e-01,  7.61699319e-01,\n",
       "          4.39831525e-01,  4.68188375e-01,  6.48396254e-01,\n",
       "          5.43767214e-01,  9.75668550e-01,  9.77327943e-01,\n",
       "         -4.49864239e-01,  2.87436724e-01,  1.36011809e-01,\n",
       "         -1.01937020e+00, -3.39724481e-01, -9.67661023e-01,\n",
       "         -8.07763457e-01, -7.05381393e-01,  1.55587643e-01,\n",
       "          4.88098949e-01, -6.89562917e-01, -1.38118759e-01,\n",
       "          1.21141672e-02,  6.91606879e-01,  7.84439683e-01,\n",
       "          1.03583193e+00,  9.61062908e-01,  5.63130230e-02,\n",
       "         -6.88561380e-01, -1.58052385e+00, -1.18323493e+00,\n",
       "         -7.67976165e-01, -1.18561625e+00,  2.50739306e-02,\n",
       "         -6.66324317e-01, -5.96608400e-01,  2.95108110e-01,\n",
       "          4.25528973e-01, -1.04570472e+00, -6.28497660e-01,\n",
       "         -8.26500416e-01, -1.08718538e+00, -1.12778246e+00,\n",
       "         -1.30361843e+00, -9.27351832e-01,  8.10976982e-01,\n",
       "         -1.12059331e+00],\n",
       "        [-7.96116218e-02,  1.40770316e-01, -4.47627246e-01,\n",
       "         -3.64522249e-01,  7.26740956e-02,  9.20895264e-02,\n",
       "         -6.12619281e-01, -2.00723082e-01, -2.20153451e-01,\n",
       "         -2.02174336e-01, -1.45168304e-01, -2.26719171e-01,\n",
       "         -1.99135154e-01, -2.06595242e-01, -1.08756296e-01,\n",
       "          3.08185667e-02, -2.60287791e-01,  3.87882233e-01,\n",
       "          1.78616449e-01, -4.07740831e-01,  1.94020838e-01,\n",
       "         -1.99588805e-01,  6.03408329e-02, -1.63816214e-01,\n",
       "         -3.23529989e-01, -8.87889639e-02, -6.55030251e-01,\n",
       "          1.23795800e-01,  8.49182546e-01,  7.99350500e-01,\n",
       "          4.40715671e-01,  3.94323796e-01,  5.97746298e-02,\n",
       "          5.39458930e-01,  2.02957422e-01,  2.99876958e-01,\n",
       "         -1.66060716e-01,  2.86244512e-01,  2.68807352e-01,\n",
       "         -1.29307777e-01, -8.03412125e-02, -7.48176053e-02,\n",
       "         -8.14494267e-02, -6.69525340e-02, -3.75038773e-01,\n",
       "          1.37762666e-01,  1.94214076e-01, -1.61303461e-01,\n",
       "          4.93975073e-01, -4.23171148e-02,  3.07322264e-01,\n",
       "          2.60776639e-01,  4.00845706e-03,  4.82801348e-03,\n",
       "          2.81823963e-01,  7.74095505e-02,  4.33229059e-01,\n",
       "          4.03299868e-01, -7.27460161e-02, -2.60430872e-01,\n",
       "          4.62886393e-01,  2.32380658e-01, -1.85640365e-01,\n",
       "          8.93109515e-02],\n",
       "        [ 5.54273427e-01,  3.71593654e-01,  9.17983353e-01,\n",
       "          1.23729241e+00,  5.21534860e-01,  5.79299927e-01,\n",
       "          1.02276564e+00,  1.60399079e-03, -1.90623152e+00,\n",
       "         -2.05471539e+00, -1.86283648e+00, -1.35090303e+00,\n",
       "         -1.28754362e-01, -1.11878537e-01, -5.79747558e-01,\n",
       "         -1.68241560e-01,  5.76368868e-01,  1.17595911e+00,\n",
       "          9.22814608e-01,  5.11528671e-01, -1.10590184e+00,\n",
       "          5.48343122e-01, -1.11897886e+00,  8.68454397e-01,\n",
       "         -9.16886255e-02, -4.66942042e-01,  7.31483325e-02,\n",
       "         -4.32539910e-01, -5.10418832e-01, -2.36692727e-01,\n",
       "         -7.15100989e-02, -5.59358656e-01,  5.97497463e-01,\n",
       "          9.66057554e-03,  2.77470574e-02,  4.83996749e-01,\n",
       "          4.22197819e-01, -4.19120230e-02,  4.90164831e-02,\n",
       "          4.37342912e-01, -1.15815818e-01, -1.55877441e-01,\n",
       "          2.62902752e-02,  4.42592353e-01,  2.12401420e-01,\n",
       "         -1.39036667e+00, -3.03650554e-02, -3.20852578e-01,\n",
       "         -8.91454816e-01, -8.74456346e-01,  5.00892639e-01,\n",
       "          7.31961966e-01,  1.07775676e+00,  7.54423440e-01,\n",
       "         -5.03944792e-03, -3.22737813e-01, -5.34719050e-01,\n",
       "         -5.34312665e-01, -2.22762764e-01,  7.18657151e-02,\n",
       "         -3.89049292e-01, -3.84499669e-01, -7.69101322e-01,\n",
       "          1.00627370e-01],\n",
       "        [ 5.30019939e-01,  7.83223748e-01,  1.53205380e-01,\n",
       "          7.64565945e-01,  4.75135267e-01,  8.45216513e-01,\n",
       "          3.22204709e-01, -6.69054568e-01, -2.12469101e-02,\n",
       "         -2.66999006e-04,  1.18895516e-01,  3.61764431e-02,\n",
       "         -8.53535593e-01, -8.20358813e-01, -6.09593332e-01,\n",
       "          2.91976333e-03, -6.09971762e-01, -1.43537140e+00,\n",
       "         -4.30435613e-02,  1.83662817e-01, -3.88680622e-02,\n",
       "          9.47095901e-02,  1.93843231e-01,  7.68513232e-02,\n",
       "         -1.61456451e-01,  2.54403234e-01,  1.02285850e+00,\n",
       "         -8.01979542e-01, -6.80148542e-01, -9.12805378e-01,\n",
       "         -8.26512218e-01, -3.99008185e-01, -9.09648538e-01,\n",
       "         -7.92777896e-01, -8.16475987e-01,  4.67394739e-02,\n",
       "         -2.95787990e-01, -9.60703492e-01, -7.39527106e-01,\n",
       "         -4.90981460e-01, -6.16882965e-02, -7.68847883e-01,\n",
       "          2.35398605e-01,  3.31324339e-03,  1.66246548e-01,\n",
       "          1.05191246e-01, -6.49412036e-01, -8.32777083e-01,\n",
       "         -1.03983498e+00,  3.42427552e-01, -2.58635044e-01,\n",
       "          2.23071590e-01, -2.40498334e-02,  3.87221724e-02,\n",
       "          2.02146173e-03, -8.64610225e-02, -3.32282662e-01,\n",
       "         -3.42075765e-01, -2.75683761e-01, -4.58397031e-01,\n",
       "         -4.81705666e-01, -3.68180752e-01,  4.21429813e-01,\n",
       "         -4.07815337e-01]], dtype=float32))"
      ]
     },
     "execution_count": 117,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.matmul(z.numpy(),C),z.data.numpy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-24T08:29:36.476390Z",
     "start_time": "2019-09-24T08:29:36.470440Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0.0000000e+00,  1.0113709e-02,  8.4769595e-03, ...,\n",
       "         5.1099188e-03,  1.1498177e-02, -4.0232805e-03],\n",
       "       [ 8.8272048e-03,  0.0000000e+00,  1.2509774e-02, ...,\n",
       "         1.1770651e-02, -1.2279133e-02, -4.0153102e-03],\n",
       "       [ 1.1084562e-02,  1.8770481e-02,  0.0000000e+00, ...,\n",
       "         1.3138597e-02,  7.5483383e-03,  1.9067800e-02],\n",
       "       ...,\n",
       "       [ 5.7351333e-03,  1.2529346e-02,  1.1859529e-02, ...,\n",
       "         0.0000000e+00, -1.3995894e-02, -3.6405106e-03],\n",
       "       [ 1.6392682e-02, -9.6394122e-03,  7.3988824e-03, ...,\n",
       "        -1.9019835e-02,  0.0000000e+00,  4.8323264e-03],\n",
       "       [-2.8756424e-03, -9.9099125e-06,  1.5107048e-02, ...,\n",
       "         8.3220308e-04,  4.5808102e-03,  0.0000000e+00]], dtype=float32)"
      ]
     },
     "execution_count": 101,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "C"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-24T10:10:12.040543Z",
     "start_time": "2019-09-24T10:10:11.928833Z"
    }
   },
   "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": [
    "C = c.data.numpy()\n",
    "plt.imshow(abs(C),cmap = 'Greys')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-24T10:10:16.595721Z",
     "start_time": "2019-09-24T10:10:16.429163Z"
    }
   },
   "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": [
    "A = abs(C)+abs(C.T)\n",
    "plt.imshow(A,cmap = 'Greys')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-24T10:10:20.422574Z",
     "start_time": "2019-09-24T10:10:20.390692Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n",
       "       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0])"
      ]
     },
     "execution_count": 120,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "D = np.diag(A.sum(axis=1))\n",
    "L = D - A\n",
    "#计算特征值和特征向量\n",
    "vals, vecs = np.linalg.eig(L)\n",
    "#重新排序\n",
    "vecs = vecs[:,np.argsort(vals)]\n",
    "vals = vals[np.argsort(vals)]\n",
    "n = 2\n",
    "#用kmeans对第2到第4个特征向量聚成4类\n",
    "from sklearn.cluster import KMeans\n",
    "kmeans = KMeans(n_clusters=n)\n",
    "kmeans.fit(vecs[:,:n])\n",
    "spectral_labels = kmeans.labels_\n",
    "spectral_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-24T10:12:02.795241Z",
     "start_time": "2019-09-24T10:12:02.692495Z"
    }
   },
   "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": [
    "plt.plot(range(len(vals)),vals)\n",
    "\n",
    "import pandas as pd\n",
    "a = pd.DataFrame(vals)\n",
    "eigngap = a-a.shift().fillna(0)\n",
    "plt.plot(range(len(eigngap)),eigngap)\n",
    "plt.xlim(0,50)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-09-24T10:12:08.462309Z",
     "start_time": "2019-09-24T10:12:08.456358Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([10, 64])"
      ]
     },
     "execution_count": 127,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "z.size()"
   ]
  }
 ],
 "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": "3.6.5"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {
    "height": "calc(100% - 180px)",
    "left": "10px",
    "top": "150px",
    "width": "203px"
   },
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}