autoencoder/pytorch_basic_experiment_autoencoder_myself.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "pytorch basic experiment autoencoder myself",
      "provenance": [],
      "collapsed_sections": [],
      "machine_shape": "hm",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/manishiitg/ML_Experiments/blob/master/autoencoder/pytorch_basic_experiment_autoencoder_myself.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dtG1b7UWHSK5",
        "colab_type": "text"
      },
      "source": [
        "# Autoencoder in pytorch\n",
        "\n",
        "just trying to simple FF autoencoder using pytorch"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Sv7z92ycAP1I",
        "colab_type": "code",
        "outputId": "b7348500-b673-4b53-ffe0-9925bf139522",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "import torch\n",
        "import torch.nn as nn\n",
        "import torchvision  \n",
        "import torchvision.transforms as transforms\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline  \n",
        "\n",
        "device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
        "print(device)\n",
        "\n"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "cuda\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PVuYJlnXM_ES",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "Simple Encoder/Decoder Network"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "97QcSO9JAXtO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class AutoEncoderNN(nn.Module):\n",
        "    def __init__(self, input_size = 784, encoding_dim = 10):\n",
        "        super(AutoEncoderNN, self).__init__()\n",
        "        self.encoder = nn.Sequential(\n",
        "            nn.Linear(input_size, 500),\n",
        "            nn.ReLU(),\n",
        "            nn.Linear(500, 500),\n",
        "            nn.ReLU(),\n",
        "            nn.Linear(500, 2000),\n",
        "            nn.ReLU(),\n",
        "            nn.Linear(2000, encoding_dim),\n",
        "            nn.ReLU()\n",
        "        )   \n",
        "\n",
        "        self.decoder = nn.Sequential(\n",
        "            nn.Linear(encoding_dim, 2000),\n",
        "            nn.ReLU(),\n",
        "            nn.Linear(2000, 500),\n",
        "            nn.ReLU(),\n",
        "            nn.Linear(500, 500),\n",
        "            nn.ReLU(),\n",
        "            nn.Linear(500, input_size),\n",
        "            nn.Tanh()\n",
        "        )\n",
        "\n",
        "    \n",
        "    def forward(self, x):\n",
        "        out = self.encoder(x)\n",
        "        out = self.decoder(out)\n",
        "        return out\n",
        "\n",
        "    def encode(self, x):\n",
        "      return self.encoder(x)\n",
        "\n",
        "    def decode(self, x):\n",
        "      return self.decoder(x)\n",
        "\n",
        "model = AutoEncoderNN().to(device)\n",
        "\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eIfZ1xhzNCfu",
        "colab_type": "text"
      },
      "source": [
        "MNIST Dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2WOZMa6hEXS_",
        "colab_type": "code",
        "outputId": "fe2c676b-82b9-4d73-9626-a13fbba2f95f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 331
        }
      },
      "source": [
        "\n",
        "# Hyper-parameters \n",
        "batch_size = 250\n",
        "\n",
        "transform = transforms.Compose([transforms.ToTensor(),\n",
        "                              transforms.Normalize((0.5,), (0.5,)),\n",
        "                              ])\n",
        "\n",
        "# MNIST dataset \n",
        "train_dataset = torchvision.datasets.MNIST(root='data',\n",
        "                                           train=True,\n",
        "                                           transform=transform,  \n",
        "                                           download=True)\n",
        "\n",
        "\n",
        "test_dataset = torchvision.datasets.MNIST(root='data', \n",
        "                                          train=False, \n",
        "                                          transform=transform)\n",
        "\n",
        "combine_dataset = torch.utils.data.ConcatDataset([train_dataset, test_dataset])\n",
        "\n",
        "\n",
        "# Data loader\n",
        "train_loader = torch.utils.data.DataLoader(dataset=train_dataset, \n",
        "                                           batch_size=batch_size, \n",
        "                                           shuffle=True)\n",
        "\n",
        "test_loader = torch.utils.data.DataLoader(dataset=test_dataset, \n",
        "                                          batch_size=20, \n",
        "                                          shuffle=False)\n",
        "\n",
        "combin_dataloader = torch.utils.data.DataLoader(dataset=combine_dataset, batch_size=batch_size, \n",
        "                                          shuffle=False)\n",
        "\n",
        "\n",
        "print(len(combine_dataset))\n",
        "\n",
        "print(len(combine_dataset)/batch_size)\n",
        "\n",
        "print(combine_dataset[0][0].shape)\n",
        "\n",
        "# # for epoch in range(1):\n",
        "# for i, (images, labels) in enumerate(combin_dataloader):  \n",
        "#     # Move tensors to the configured device\n",
        "#     images = images.reshape(-1, 28*28)\n",
        "#     labels = labels\n",
        "\n",
        "#     print(images.data.numpy())\n",
        "\n",
        "#     if i > 1:\n",
        "#       break\n"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "  0%|          | 0/9912422 [00:00<?, ?it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Downloading http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz to data/MNIST/raw/train-images-idx3-ubyte.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "9920512it [00:00, 28048425.68it/s]                            \n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Extracting data/MNIST/raw/train-images-idx3-ubyte.gz to data/MNIST/raw\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "32768it [00:00, 448051.67it/s]\n",
            "  1%|          | 16384/1648877 [00:00<00:11, 145431.58it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Downloading http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz to data/MNIST/raw/train-labels-idx1-ubyte.gz\n",
            "Extracting data/MNIST/raw/train-labels-idx1-ubyte.gz to data/MNIST/raw\n",
            "Downloading http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz to data/MNIST/raw/t10k-images-idx3-ubyte.gz\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "1654784it [00:00, 7180800.29it/s]                            \n",
            "8192it [00:00, 184386.78it/s]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Extracting data/MNIST/raw/t10k-images-idx3-ubyte.gz to data/MNIST/raw\n",
            "Downloading http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz to data/MNIST/raw/t10k-labels-idx1-ubyte.gz\n",
            "Extracting data/MNIST/raw/t10k-labels-idx1-ubyte.gz to data/MNIST/raw\n",
            "Processing...\n",
            "Done!\n",
            "70000\n",
            "280.0\n",
            "torch.Size([1, 28, 28])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Z3wivifdFNZy",
        "colab_type": "code",
        "outputId": "85a408c2-a966-473b-c905-83b68fbd3bd7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 129
        }
      },
      "source": [
        "n = 10\n",
        "plt.figure(figsize=(20, 4))\n",
        "for i in range(n):\n",
        "    # display original\n",
        "    ax = plt.subplot(2, n, i + 1)\n",
        "    plt.imshow(combine_dataset[i][0].numpy().squeeze())\n",
        "    plt.gray()\n",
        "    ax.get_xaxis().set_visible(False)\n",
        "    ax.get_yaxis().set_visible(False)\n",
        "\n",
        "plt.show()"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAABG0AAABwCAYAAACkaY2RAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nq2S2rgAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 1440x288 with 10 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OtRGd9XUNFeY",
        "colab_type": "text"
      },
      "source": [
        "Training the network"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ve8sWKodi3Ud",
        "colab_type": "code",
        "outputId": "e3eb990c-a0ab-4309-fbca-96f64ef2504b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 363
        }
      },
      "source": [
        "# Train the model\n",
        "total_step = len(combin_dataloader)\n",
        "num_epochs = 10\n",
        "\n",
        "learning_rate = .0005\n",
        "\n",
        "# Loss and optimizer\n",
        "# optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate, momentum=.9)\n",
        "optimizer = torch.optim.Adam(\n",
        "    model.parameters(), lr=learning_rate, weight_decay=1e-5)\n",
        "\n",
        "criterion = nn.MSELoss() \n",
        "\n",
        "for epoch in range(num_epochs):\n",
        "    for i, (images, labels) in enumerate(combin_dataloader):  \n",
        "        # Move tensors to the configured device\n",
        "        images = images.reshape(-1, 28*28).to(device)\n",
        "        \n",
        "        # Forward pass\n",
        "        outputs = model(images)\n",
        "        loss = criterion(outputs, images)\n",
        "        \n",
        "        # Backward and optimize\n",
        "        optimizer.zero_grad()\n",
        "        loss.backward()\n",
        "        optimizer.step()\n",
        "        \n",
        "        if (i+1) % 100 == 0:\n",
        "            print ('Epoch [{}/{}], Step [{}/{}], Loss: {:.4f}' \n",
        "                   .format(epoch+1, num_epochs, i+1, total_step, loss.item()))"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch [1/10], Step [100/280], Loss: 0.2194\n",
            "Epoch [1/10], Step [200/280], Loss: 0.2086\n",
            "Epoch [2/10], Step [100/280], Loss: 0.1684\n",
            "Epoch [2/10], Step [200/280], Loss: 0.1676\n",
            "Epoch [3/10], Step [100/280], Loss: 0.1350\n",
            "Epoch [3/10], Step [200/280], Loss: 0.1479\n",
            "Epoch [4/10], Step [100/280], Loss: 0.1231\n",
            "Epoch [4/10], Step [200/280], Loss: 0.1379\n",
            "Epoch [5/10], Step [100/280], Loss: 0.1160\n",
            "Epoch [5/10], Step [200/280], Loss: 0.1307\n",
            "Epoch [6/10], Step [100/280], Loss: 0.1104\n",
            "Epoch [6/10], Step [200/280], Loss: 0.1268\n",
            "Epoch [7/10], Step [100/280], Loss: 0.1076\n",
            "Epoch [7/10], Step [200/280], Loss: 0.1236\n",
            "Epoch [8/10], Step [100/280], Loss: 0.1046\n",
            "Epoch [8/10], Step [200/280], Loss: 0.1219\n",
            "Epoch [9/10], Step [100/280], Loss: 0.1028\n",
            "Epoch [9/10], Step [200/280], Loss: 0.1194\n",
            "Epoch [10/10], Step [100/280], Loss: 0.1010\n",
            "Epoch [10/10], Step [200/280], Loss: 0.1182\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "S2TreyF0rajJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import os.path\n",
        "if os.path.isfile('basic_autoencoer_pytorch_25_epoch.json'):\n",
        "  model.load_state_dict(torch.load(\"basic_autoencoer_pytorch_25_epoch.json\"))\n",
        "  model.eval()\n",
        "else:\n",
        "  torch.save(model.state_dict(), \"basic_autoencoer_pytorch_25_epoch.json\")"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "o-76Sd6uNb9_",
        "colab_type": "text"
      },
      "source": [
        "Comparing encoded images with original images"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EdxQhIPtto01",
        "colab_type": "code",
        "outputId": "2978d47d-1508-477c-8aec-bbd8aa0c3398",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 200
        }
      },
      "source": [
        "with torch.no_grad():\n",
        "  for i, (images, labels) in enumerate(test_loader):  \n",
        "    images = images.reshape(-1, 28*28).to(device)\n",
        "    outputs = model(images)\n",
        "\n",
        "    \n",
        "    n = len(images)\n",
        "    plt.figure(figsize=(20, 4))\n",
        "    for i in range(n):\n",
        "        # display original\n",
        "        ax = plt.subplot(2, n, i + 1)\n",
        "        plt.imshow(images[i].reshape(28,28).cpu().data.numpy())\n",
        "        plt.gray()\n",
        "        ax.get_xaxis().set_visible(False)\n",
        "        ax.get_yaxis().set_visible(False)\n",
        "\n",
        "        # display reconstruction\n",
        "        ax = plt.subplot(2, n, i + 1 + n)\n",
        "        plt.imshow(outputs[i].reshape(28, 28).cpu().data.numpy())\n",
        "        plt.gray()\n",
        "        ax.get_xaxis().set_visible(False)\n",
        "        ax.get_yaxis().set_visible(False)\n",
        "    plt.show()\n",
        "\n",
        "    break\n",
        "\n"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAABG0AAAC3CAYAAACohKOdAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/D+jrALph8Ww1QAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 1440x288 with 40 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tp6lpsl8zvRQ",
        "colab_type": "code",
        "outputId": "38ef068c-2392-4274-a534-c38df00a0ce8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "encoded_data = None\n",
        "Y = None\n",
        "with torch.no_grad():\n",
        "  for i, (images, labels) in enumerate(combin_dataloader):  \n",
        "    images = images.reshape(-1, 28*28).to(device)\n",
        "    outputs = model.encode(images)\n",
        "    if encoded_data is None:\n",
        "      encoded_data = outputs.cpu().data.numpy()\n",
        "      Y = labels\n",
        "    else:\n",
        "      encoded_data = np.append(encoded_data, outputs.cpu().data.numpy(), axis = 0)\n",
        "      Y = np.append(Y, labels)\n",
        "\n",
        "    # if i > 10:\n",
        "    #   break\n",
        "\n",
        "\n",
        "print(encoded_data.shape)\n",
        "\n"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(70000, 10)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HEv48W4RI3LK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def acc(y_true, y_pred):\n",
        "    \"\"\"\n",
        "    Calculate clustering accuracy. Require scikit-learn installed\n",
        "    # Arguments\n",
        "        y: true labels, numpy.array with shape `(n_samples,)`\n",
        "        y_pred: predicted labels, numpy.array with shape `(n_samples,)`\n",
        "    # Return\n",
        "        accuracy, in [0,1]\n",
        "    \"\"\"\n",
        "    y_true = y_true.astype(np.int64)\n",
        "    assert y_pred.size == y_true.size\n",
        "    D = max(y_pred.max(), y_true.max()) + 1\n",
        "    w = np.zeros((D, D), dtype=np.int64)\n",
        "    for i in range(y_pred.size):\n",
        "        w[y_pred[i], y_true[i]] += 1\n",
        "    from sklearn.utils.linear_assignment_ import linear_assignment\n",
        "    ind = linear_assignment(w.max() - w)\n",
        "    return sum([w[i, j] for i, j in ind]) * 1.0 / y_pred.size"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AHx3Ej4YNfpq",
        "colab_type": "text"
      },
      "source": [
        "Clustering of Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PocuKYOM1PZT",
        "colab_type": "code",
        "outputId": "c197098f-bd36-4104-c3e5-2e9864cc2129",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 357
        }
      },
      "source": [
        "from sklearn.cluster import KMeans\n",
        "X = encoded_data\n",
        "kmeans = KMeans(n_clusters=10, random_state=0).fit(X)\n",
        "Y_pred = kmeans.predict(X)\n",
        "\n",
        "print(\"kmean accuracy \" ,  acc(Y, Y_pred))\n",
        "\n",
        "plt.scatter(X[:, 0], X[:, 1], c=Y, cmap='viridis')\n",
        "\n",
        "centers = kmeans.cluster_centers_\n",
        "plt.scatter(centers[:, 0], centers[:, 1], c='black', s=100, alpha=0.5);\n",
        "\n",
        "print(centers.shape)\n"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/sklearn/utils/linear_assignment_.py:127: DeprecationWarning: The linear_assignment function is deprecated in 0.21 and will be removed from 0.23. Use scipy.optimize.linear_sum_assignment instead.\n",
            "  DeprecationWarning)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "kmean accuracy  0.45015714285714287\n",
            "(10, 10)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYkAAAD7CAYAAACfQGjDAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "So-eELsT6atW",
        "colab_type": "code",
        "outputId": "c0075605-008e-46eb-c41d-de539b5dde91",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 129
        }
      },
      "source": [
        "n = len(centers)\n",
        "\n",
        "outputs = model.decode(torch.from_numpy(centers).to(device))\n",
        "plt.figure(figsize=(20, 4))\n",
        "for i in range(n):\n",
        "    # display original\n",
        "    ax = plt.subplot(2, n, i + 1)\n",
        "    \n",
        "    plt.imshow(outputs[i].reshape(28, 28).cpu().data.numpy())\n",
        "    plt.gray()\n",
        "    ax.get_xaxis().set_visible(False)\n",
        "    ax.get_yaxis().set_visible(False)\n"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAABG0AAABwCAYAAACkaY2RAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAIAK4ksbAAAAAACACvo/8TQHado+1vEAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 1440x288 with 10 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OBDZGXz_N6tS",
        "colab_type": "text"
      },
      "source": [
        "Accuracy is very low for clustering. This is was just an experiment, for images its better to go with CNN"
      ]
    }
  ]
}