FUSENet.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "FUSENet.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "mJVLwWkc21pG",
        "colab_type": "code",
        "outputId": "ab6faa64-20b5-4b8e-f758-834a100fb182",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 153
        }
      },
      "source": [
        "!pip3 install keras==2.2.4"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: keras==2.2.4 in /usr/local/lib/python3.6/dist-packages (2.2.4)\n",
            "Requirement already satisfied: numpy>=1.9.1 in /usr/local/lib/python3.6/dist-packages (from keras==2.2.4) (1.17.4)\n",
            "Requirement already satisfied: scipy>=0.14 in /usr/local/lib/python3.6/dist-packages (from keras==2.2.4) (1.3.3)\n",
            "Requirement already satisfied: pyyaml in /usr/local/lib/python3.6/dist-packages (from keras==2.2.4) (3.13)\n",
            "Requirement already satisfied: h5py in /usr/local/lib/python3.6/dist-packages (from keras==2.2.4) (2.8.0)\n",
            "Requirement already satisfied: keras-applications>=1.0.6 in /usr/local/lib/python3.6/dist-packages (from keras==2.2.4) (1.0.8)\n",
            "Requirement already satisfied: keras-preprocessing>=1.0.5 in /usr/local/lib/python3.6/dist-packages (from keras==2.2.4) (1.1.0)\n",
            "Requirement already satisfied: six>=1.9.0 in /usr/local/lib/python3.6/dist-packages (from keras==2.2.4) (1.12.0)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "I1zvad66hMfI",
        "colab_type": "code",
        "outputId": "9a39859b-8047-4a5a-a59a-0dc9f221ead0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 80
        }
      },
      "source": [
        "import keras\n",
        "from keras.layers import Conv2D, Conv3D, Flatten, Dense, Reshape, BatchNormalization\n",
        "from keras.layers import Dropout, Input\n",
        "from keras.models import Model\n",
        "from keras.optimizers import Adam\n",
        "from keras.callbacks import ModelCheckpoint\n",
        "from keras.utils import np_utils\n",
        "\n",
        "from sklearn.decomposition import PCA\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import confusion_matrix, accuracy_score, classification_report, cohen_kappa_score\n",
        "\n",
        "from operator import truediv\n",
        "\n",
        "from plotly.offline import init_notebook_mode\n",
        "\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import scipy.io as sio\n",
        "import os\n",
        "\n",
        "init_notebook_mode(connected=True)\n",
        "%matplotlib inline\n",
        "import numpy as np\n",
        "import random \n",
        "#import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import scipy.io as sio\n",
        "from keras.models import Sequential, Model\n",
        "from keras.layers import Convolution2D, MaxPooling2D, Conv3D, MaxPooling3D, ZeroPadding3D\n",
        "from keras.layers import Activation, Dropout, Flatten, Dense, BatchNormalization, Input\n",
        "from keras.utils.np_utils import to_categorical\n",
        "from sklearn.decomposition import PCA\n",
        "from keras.optimizers import Adam, SGD, Adadelta, RMSprop, Nadam\n",
        "import keras.callbacks as kcallbacks\n",
        "from keras.regularizers import l2\n",
        "import time\n",
        "import collections\n",
        "from sklearn import metrics, preprocessing\n",
        "from keras import backend as K"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<p style=\"color: red;\">\n",
              "The default version of TensorFlow in Colab will soon switch to TensorFlow 2.x.<br>\n",
              "We recommend you <a href=\"https://www.tensorflow.org/guide/migrate\" target=\"_blank\">upgrade</a> now \n",
              "or ensure your notebook will continue to use TensorFlow 1.x via the <code>%tensorflow_version 1.x</code> magic:\n",
              "<a href=\"https://colab.research.google.com/notebooks/tensorflow_version.ipynb\" target=\"_blank\">more info</a>.</p>\n"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "        <script type=\"text/javascript\">\n",
              "        window.PlotlyConfig = {MathJaxConfig: 'local'};\n",
              "        if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}\n",
              "        if (typeof require !== 'undefined') {\n",
              "        require.undef(\"plotly\");\n",
              "        requirejs.config({\n",
              "            paths: {\n",
              "                'plotly': ['https://cdn.plot.ly/plotly-latest.min']\n",
              "            }\n",
              "        });\n",
              "        require(['plotly'], function(Plotly) {\n",
              "            window._Plotly = Plotly;\n",
              "        });\n",
              "        }\n",
              "        </script>\n",
              "        "
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hUW0w1aIhdYY",
        "colab_type": "code",
        "outputId": "2365b5b8-dbe3-40f4-a625-38a74574aee5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive',force_remount=False)"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/gdrive; to attempt to forcibly remount, call drive.mount(\"/content/gdrive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2mKMExLghjyq",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def loadData(name):\n",
        "    data_path = '/content/gdrive/My Drive/GAMO_gen/'\n",
        "    if name == 'IP':\n",
        "        data = sio.loadmat(os.path.join(data_path, 'Indian_pines_corrected.mat'))['indian_pines_corrected']\n",
        "        labels = sio.loadmat(os.path.join(data_path, 'Indian_pines_gt.mat'))['indian_pines_gt']\n",
        "    elif name == 'SA':\n",
        "        data = sio.loadmat(os.path.join(data_path, 'Salinas_corrected.mat'))['salinas_corrected']\n",
        "        labels = sio.loadmat(os.path.join(data_path, 'Salinas_gt.mat'))['salinas_gt']\n",
        "    elif name == 'PU':\n",
        "        data = sio.loadmat(os.path.join(data_path, 'PaviaU.mat'))['paviaU']\n",
        "        labels = sio.loadmat(os.path.join(data_path, 'PaviaU_gt.mat'))['paviaU_gt']\n",
        "    elif name == 'KSC':\n",
        "        data = sio.loadmat(os.path.join(data_path, 'KSC.mat'))['KSC']\n",
        "        labels = sio.loadmat(os.path.join(data_path, 'KSC_gt.mat'))['KSC_gt']\n",
        "    \n",
        "    return data, labels"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Co8BW-WshtXI",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def splitTrainTestSet(X, y, testRatio, randomState=345):\n",
        "    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=testRatio, random_state=randomState,\n",
        "                                                        stratify=y)\n",
        "    return X_train, X_test, y_train, y_test"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "llqxvI-uhyT6",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def applyPCA(X, numComponents=75):\n",
        "    newX = np.reshape(X, (-1, X.shape[2]))\n",
        "    pca = PCA(n_components=numComponents, whiten=True)\n",
        "    newX = pca.fit_transform(newX)\n",
        "    newX = np.reshape(newX, (X.shape[0],X.shape[1], numComponents))\n",
        "    return newX, pca"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GLiYZYqXh07S",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def padWithZeros(X, margin=2):\n",
        "    newX = np.zeros((X.shape[0] + 2 * margin, X.shape[1] + 2* margin, X.shape[2]))\n",
        "    x_offset = margin\n",
        "    y_offset = margin\n",
        "    newX[x_offset:X.shape[0] + x_offset, y_offset:X.shape[1] + y_offset, :] = X\n",
        "    return newX"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U5q0SW9HmsXK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def createImageCubes(X, y, windowSize=5, removeZeroLabels = True):\n",
        "    margin = int((windowSize - 1) / 2)\n",
        "    zeroPaddedX = padWithZeros(X, margin=margin)\n",
        "    # split patches\n",
        "    patchesData = np.zeros((X.shape[0] * X.shape[1], windowSize, windowSize, X.shape[2]))\n",
        "    patchesLabels = np.zeros((X.shape[0] * X.shape[1]))\n",
        "    patchIndex = 0\n",
        "    for r in range(margin, zeroPaddedX.shape[0] - margin):\n",
        "        for c in range(margin, zeroPaddedX.shape[1] - margin):\n",
        "            patch = zeroPaddedX[r - margin:r + margin + 1, c - margin:c + margin + 1]   \n",
        "            patchesData[patchIndex, :, :, :] = patch\n",
        "            patchesLabels[patchIndex] = y[r-margin, c-margin]\n",
        "            patchIndex = patchIndex + 1\n",
        "    if removeZeroLabels:\n",
        "        patchesData = patchesData[patchesLabels>0,:,:,:]\n",
        "        patchesLabels = patchesLabels[patchesLabels>0]\n",
        "        patchesLabels -= 1\n",
        "    return patchesData, patchesLabels"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZTbZ-IKlivZP",
        "colab_type": "code",
        "outputId": "f78f4abf-d1d0-4a06-a4ff-119ab2a90719",
        "colab": {
          "resources": {
            "http://localhost:8080/nbextensions/google.colab/files.js": {
              "data": "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",
              "ok": true,
              "headers": [
                [
                  "content-type",
                  "application/javascript"
                ]
              ],
              "status": 200,
              "status_text": ""
            }
          },
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "from google.colab import files\n",
        "fl = files.upload() #upload the .py here from utils\n",
        "import se_resnet8_max\n",
        "\n",
        "def res4_model_ss():\n",
        "    model_res4 = se_resnet8_max.SE_ResNet8((1, w, w, K),64, mp[d])\n",
        "    # Set the hyper-parameters\n",
        "    # m set according to paper to 2 * number of bits\n",
        "    #my_custom_loss = lambda x,y: hashing_loss(x, y, m=2*32, alpha=0.001)\n",
        "\n",
        "    # Compile the model\n",
        "    #opt = keras.optimizers.adam(lr=0.1)\n",
        "    #model_res4.compile(loss=my_custom_loss, optimizer=opt,metrics=['accuracy'])\n",
        "    RMS = RMSprop(lr=0.001)\n",
        "    # Let's train the model using RMSprop\n",
        "    model_res4.compile(loss='categorical_crossentropy', optimizer=RMS, metrics=['accuracy'])\n",
        "\n",
        "    return model_res4"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "     <input type=\"file\" id=\"files-a02740ce-5649-497c-87a0-8cdb618c33dd\" name=\"files[]\" multiple disabled />\n",
              "     <output id=\"result-a02740ce-5649-497c-87a0-8cdb618c33dd\">\n",
              "      Upload widget is only available when the cell has been executed in the\n",
              "      current browser session. Please rerun this cell to enable.\n",
              "      </output>\n",
              "      <script src=\"/nbextensions/google.colab/files.js\"></script> "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:74: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:517: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:4185: The name tf.truncated_normal is deprecated. Please use tf.random.truncated_normal instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:133: The name tf.placeholder_with_default is deprecated. Please use tf.compat.v1.placeholder_with_default instead.\n",
            "\n",
            "input shape: (None, 15, 15, 5, 32)\n",
            "(?, 1, 1, 32)\n",
            "input shape: (None, 13, 13, 1, 32)\n",
            "(?, 1, 1, 32)\n",
            "(None, 13, 13, 32)\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:4138: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.\n",
            "\n",
            "call\n",
            "output\n",
            "(None, 13, 13, 32)\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            (None, 15, 15, 15, 1 0                                            \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_1 (Conv3D)               (None, 15, 15, 5, 32 256         input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_1 (BatchNor (None, 15, 15, 5, 32 128         conv3d_1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_1 (Activation)       (None, 15, 15, 5, 32 0           batch_normalization_1[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_2 (Conv3D)               (None, 15, 15, 5, 32 9248        activation_1[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_2 (BatchNor (None, 15, 15, 5, 32 128         conv3d_2[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_2 (Activation)       (None, 15, 15, 5, 32 0           batch_normalization_2[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_3 (Conv3D)               (None, 15, 15, 5, 32 9248        activation_2[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling3d_1 (Glo (None, 32)           0           conv3d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling3d_1 (GlobalM (None, 32)           0           conv3d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "reshape_1 (Reshape)             (None, 1, 1, 32)     0           global_average_pooling3d_1[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "reshape_2 (Reshape)             (None, 1, 1, 32)     0           global_max_pooling3d_1[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "dense_1 (Dense)                 (None, 1, 1, 2)      64          reshape_1[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "dense_3 (Dense)                 (None, 1, 1, 2)      64          reshape_2[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "dense_2 (Dense)                 (None, 1, 1, 32)     64          dense_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dense_4 (Dense)                 (None, 1, 1, 32)     64          dense_3[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "multiply_1 (Multiply)           (None, 15, 15, 5, 32 0           conv3d_3[0][0]                   \n",
            "                                                                 dense_2[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "multiply_2 (Multiply)           (None, 15, 15, 5, 32 0           conv3d_3[0][0]                   \n",
            "                                                                 dense_4[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "maximum_1 (Maximum)             (None, 15, 15, 5, 32 0           multiply_1[0][0]                 \n",
            "                                                                 multiply_2[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "add_1 (Add)                     (None, 15, 15, 5, 32 0           maximum_1[0][0]                  \n",
            "                                                                 conv3d_3[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_3 (BatchNor (None, 15, 15, 5, 32 128         add_1[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "activation_3 (Activation)       (None, 15, 15, 5, 32 0           batch_normalization_3[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_4 (Conv3D)               (None, 15, 15, 1, 12 20608       activation_3[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_4 (BatchNor (None, 15, 15, 1, 12 512         conv3d_4[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_4 (Activation)       (None, 15, 15, 1, 12 0           batch_normalization_4[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "reshape_3 (Reshape)             (None, 15, 15, 128,  0           activation_4[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_5 (Conv3D)               (None, 13, 13, 1, 32 36896       reshape_3[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_5 (BatchNor (None, 13, 13, 1, 32 128         conv3d_5[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_5 (Activation)       (None, 13, 13, 1, 32 0           batch_normalization_5[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_8 (Conv3D)               (None, 13, 13, 1, 32 1056        activation_5[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling3d_2 (Glo (None, 32)           0           conv3d_8[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling3d_2 (GlobalM (None, 32)           0           conv3d_8[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "reshape_4 (Reshape)             (None, 1, 1, 32)     0           global_average_pooling3d_2[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "reshape_5 (Reshape)             (None, 1, 1, 32)     0           global_max_pooling3d_2[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "dense_5 (Dense)                 (None, 1, 1, 2)      64          reshape_4[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "dense_7 (Dense)                 (None, 1, 1, 2)      64          reshape_5[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_6 (Conv3D)               (None, 13, 13, 1, 32 9248        activation_5[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dense_6 (Dense)                 (None, 1, 1, 32)     64          dense_5[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dense_8 (Dense)                 (None, 1, 1, 32)     64          dense_7[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_6 (BatchNor (None, 13, 13, 1, 32 128         conv3d_6[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "multiply_3 (Multiply)           (None, 13, 13, 1, 32 0           conv3d_8[0][0]                   \n",
            "                                                                 dense_6[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "multiply_4 (Multiply)           (None, 13, 13, 1, 32 0           conv3d_8[0][0]                   \n",
            "                                                                 dense_8[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "activation_6 (Activation)       (None, 13, 13, 1, 32 0           batch_normalization_6[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "maximum_2 (Maximum)             (None, 13, 13, 1, 32 0           multiply_3[0][0]                 \n",
            "                                                                 multiply_4[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_7 (Conv3D)               (None, 13, 13, 1, 32 9248        activation_6[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "add_2 (Add)                     (None, 13, 13, 1, 32 0           maximum_2[0][0]                  \n",
            "                                                                 conv3d_7[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_7 (BatchNor (None, 13, 13, 1, 32 128         add_2[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "activation_7 (Activation)       (None, 13, 13, 1, 32 0           batch_normalization_7[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_8 (BatchNor (None, 13, 13, 1, 32 128         activation_7[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "activation_8 (Activation)       (None, 13, 13, 1, 32 0           batch_normalization_8[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "reshape_6 (Reshape)             (None, 13, 13, 32)   0           activation_8[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "bo_f__pooling_1 (BoF_Pooling)   (None, 64)           2112        reshape_6[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "dropout_1 (Dropout)             (None, 64)           0           bo_f__pooling_1[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "dense_9 (Dense)                 (None, 16)           1040        dropout_1[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 100,880\n",
            "Trainable params: 100,176\n",
            "Non-trainable params: 704\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_MVm89fVi9PI",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "98c778d7-7126-4442-d2fa-36c334c93e75"
      },
      "source": [
        "import os\n",
        "data_path = '/content/gdrive/My Drive/FuseNet' #Enter Desired output path here from utils\n",
        "os.listdir(data_path)"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['1_NEW_KERNEL32_B_32_hashing15_19.hdf5']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DAOKQ0lajDlX",
        "colab_type": "code",
        "colab": {
          "resources": {
            "http://localhost:8080/nbextensions/google.colab/files.js": {
              "data": "Ly8gQ29weXJpZ2h0IDIwMTcgR29vZ2xlIExMQwovLwovLyBMaWNlbnNlZCB1bmRlciB0aGUgQXBhY2hlIExpY2Vuc2UsIFZlcnNpb24gMi4wICh0aGUgIkxpY2Vuc2UiKTsKLy8geW91IG1heSBub3QgdXNlIHRoaXMgZmlsZSBleGNlcHQgaW4gY29tcGxpYW5jZSB3aXRoIHRoZSBMaWNlbnNlLgovLyBZb3UgbWF5IG9idGFpbiBhIGNvcHkgb2YgdGhlIExpY2Vuc2UgYXQKLy8KLy8gICAgICBodHRwOi8vd3d3LmFwYWNoZS5vcmcvbGljZW5zZXMvTElDRU5TRS0yLjAKLy8KLy8gVW5sZXNzIHJlcXVpcmVkIGJ5IGFwcGxpY2FibGUgbGF3IG9yIGFncmVlZCB0byBpbiB3cml0aW5nLCBzb2Z0d2FyZQovLyBkaXN0cmlidXRlZCB1bmRlciB0aGUgTGljZW5zZSBpcyBkaXN0cmlidXRlZCBvbiBhbiAiQVMgSVMiIEJBU0lTLAovLyBXSVRIT1VUIFdBUlJBTlRJRVMgT1IgQ09ORElUSU9OUyBPRiBBTlkgS0lORCwgZWl0aGVyIGV4cHJlc3Mgb3IgaW1wbGllZC4KLy8gU2VlIHRoZSBMaWNlbnNlIGZvciB0aGUgc3BlY2lmaWMgbGFuZ3VhZ2UgZ292ZXJuaW5nIHBlcm1pc3Npb25zIGFuZAovLyBsaW1pdGF0aW9ucyB1bmRlciB0aGUgTGljZW5zZS4KCi8qKgogKiBAZmlsZW92ZXJ2aWV3IEhlbHBlcnMgZm9yIGdvb2dsZS5jb2xhYiBQeXRob24gbW9kdWxlLgogKi8KKGZ1bmN0aW9uKHNjb3BlKSB7CmZ1bmN0aW9uIHNwYW4odGV4dCwgc3R5bGVBdHRyaWJ1dGVzID0ge30pIHsKICBjb25zdCBlbGVtZW50ID0gZG9jdW1lbnQuY3JlYXRlRWxlbWVudCgnc3BhbicpOwogIGVsZW1lbnQudGV4dENvbnRlbnQgPSB0ZXh0OwogIGZvciAoY29uc3Qga2V5IG9mIE9iamVjdC5rZXlzKHN0eWxlQXR0cmlidXRlcykpIHsKICAgIGVsZW1lbnQuc3R5bGVba2V5XSA9IHN0eWxlQXR0cmlidXRlc1trZXldOwogIH0KICByZXR1cm4gZWxlbWVudDsKfQoKLy8gTWF4IG51bWJlciBvZiBieXRlcyB3aGljaCB3aWxsIGJlIHVwbG9hZGVkIGF0IGEgdGltZS4KY29uc3QgTUFYX1BBWUxPQURfU0laRSA9IDEwMCAqIDEwMjQ7Ci8vIE1heCBhbW91bnQgb2YgdGltZSB0byBibG9jayB3YWl0aW5nIGZvciB0aGUgdXNlci4KY29uc3QgRklMRV9DSEFOR0VfVElNRU9VVF9NUyA9IDMwICogMTAwMDsKCmZ1bmN0aW9uIF91cGxvYWRGaWxlcyhpbnB1dElkLCBvdXRwdXRJZCkgewogIGNvbnN0IHN0ZXBzID0gdXBsb2FkRmlsZXNTdGVwKGlucHV0SWQsIG91dHB1dElkKTsKICBjb25zdCBvdXRwdXRFbGVtZW50ID0gZG9jdW1lbnQuZ2V0RWxlbWVudEJ5SWQob3V0cHV0SWQpOwogIC8vIENhY2hlIHN0ZXBzIG9uIHRoZSBvdXRwdXRFbGVtZW50IHRvIG1ha2UgaXQgYXZhaWxhYmxlIGZvciB0aGUgbmV4dCBjYWxsCiAgLy8gdG8gdXBsb2FkRmlsZXNDb250aW51ZSBmcm9tIFB5dGhvbi4KICBvdXRwdXRFbGVtZW50LnN0ZXBzID0gc3RlcHM7CgogIHJldHVybiBfdXBsb2FkRmlsZXNDb250aW51ZShvdXRwdXRJZCk7Cn0KCi8vIFRoaXMgaXMgcm91Z2hseSBhbiBhc3luYyBnZW5lcmF0b3IgKG5vdCBzdXBwb3J0ZWQgaW4gdGhlIGJyb3dzZXIgeWV0KSwKLy8gd2hlcmUgdGhlcmUgYXJlIG11bHRpcGxlIGFzeW5jaHJvbm91cyBzdGVwcyBhbmQgdGhlIFB5dGhvbiBzaWRlIGlzIGdvaW5nCi8vIHRvIHBvbGwgZm9yIGNvbXBsZXRpb24gb2YgZWFjaCBzdGVwLgovLyBUaGlzIHVzZXMgYSBQcm9taXNlIHRvIGJsb2NrIHRoZSBweXRob24gc2lkZSBvbiBjb21wbGV0aW9uIG9mIGVhY2ggc3RlcCwKLy8gdGhlbiBwYXNzZXMgdGhlIHJlc3VsdCBvZiB0aGUgcHJldmlvdXMgc3RlcCBhcyB0aGUgaW5wdXQgdG8gdGhlIG5leHQgc3RlcC4KZnVuY3Rpb24gX3VwbG9hZEZpbGVzQ29udGludWUob3V0cHV0SWQpIHsKICBjb25zdCBvdXRwdXRFbGVtZW50ID0gZG9jdW1lbnQuZ2V0RWxlbWVudEJ5SWQob3V0cHV0SWQpOwogIGNvbnN0IHN0ZXBzID0gb3V0cHV0RWxlbWVudC5zdGVwczsKCiAgY29uc3QgbmV4dCA9IHN0ZXBzLm5leHQob3V0cHV0RWxlbWVudC5sYXN0UHJvbWlzZVZhbHVlKTsKICByZXR1cm4gUHJvbWlzZS5yZXNvbHZlKG5leHQudmFsdWUucHJvbWlzZSkudGhlbigodmFsdWUpID0+IHsKICAgIC8vIENhY2hlIHRoZSBsYXN0IHByb21pc2UgdmFsdWUgdG8gbWFrZSBpdCBhdmFpbGFibGUgdG8gdGhlIG5leHQKICAgIC8vIHN0ZXAgb2YgdGhlIGdlbmVyYXRvci4KICAgIG91dHB1dEVsZW1lbnQubGFzdFByb21pc2VWYWx1ZSA9IHZhbHVlOwogICAgcmV0dXJuIG5leHQudmFsdWUucmVzcG9uc2U7CiAgfSk7Cn0KCi8qKgogKiBHZW5lcmF0b3IgZnVuY3Rpb24gd2hpY2ggaXMgY2FsbGVkIGJldHdlZW4gZWFjaCBhc3luYyBzdGVwIG9mIHRoZSB1cGxvYWQKICogcHJvY2Vzcy4KICogQHBhcmFtIHtzdHJpbmd9IGlucHV0SWQgRWxlbWVudCBJRCBvZiB0aGUgaW5wdXQgZmlsZSBwaWNrZXIgZWxlbWVudC4KICogQHBhcmFtIHtzdHJpbmd9IG91dHB1dElkIEVsZW1lbnQgSUQgb2YgdGhlIG91dHB1dCBkaXNwbGF5LgogKiBAcmV0dXJuIHshSXRlcmFibGU8IU9iamVjdD59IEl0ZXJhYmxlIG9mIG5leHQgc3RlcHMuCiAqLwpmdW5jdGlvbiogdXBsb2FkRmlsZXNTdGVwKGlucHV0SWQsIG91dHB1dElkKSB7CiAgY29uc3QgaW5wdXRFbGVtZW50ID0gZG9jdW1lbnQuZ2V0RWxlbWVudEJ5SWQoaW5wdXRJZCk7CiAgaW5wdXRFbGVtZW50LmRpc2FibGVkID0gZmFsc2U7CgogIGNvbnN0IG91dHB1dEVsZW1lbnQgPSBkb2N1bWVudC5nZXRFbGVtZW50QnlJZChvdXRwdXRJZCk7CiAgb3V0cHV0RWxlbWVudC5pbm5lckhUTUwgPSAnJzsKCiAgY29uc3QgcGlja2VkUHJvbWlzZSA9IG5ldyBQcm9taXNlKChyZXNvbHZlKSA9PiB7CiAgICBpbnB1dEVsZW1lbnQuYWRkRXZlbnRMaXN0ZW5lcignY2hhbmdlJywgKGUpID0+IHsKICAgICAgcmVzb2x2ZShlLnRhcmdldC5maWxlcyk7CiAgICB9KTsKICB9KTsKCiAgY29uc3QgY2FuY2VsID0gZG9jdW1lbnQuY3JlYXRlRWxlbWVudCgnYnV0dG9uJyk7CiAgaW5wdXRFbGVtZW50LnBhcmVudEVsZW1lbnQuYXBwZW5kQ2hpbGQoY2FuY2VsKTsKICBjYW5jZWwudGV4dENvbnRlbnQgPSAnQ2FuY2VsIHVwbG9hZCc7CiAgY29uc3QgY2FuY2VsUHJvbWlzZSA9IG5ldyBQcm9taXNlKChyZXNvbHZlKSA9PiB7CiAgICBjYW5jZWwub25jbGljayA9ICgpID0+IHsKICAgICAgcmVzb2x2ZShudWxsKTsKICAgIH07CiAgfSk7CgogIC8vIENhbmNlbCB1cGxvYWQgaWYgdXNlciBoYXNuJ3QgcGlja2VkIGFueXRoaW5nIGluIHRpbWVvdXQuCiAgY29uc3QgdGltZW91dFByb21pc2UgPSBuZXcgUHJvbWlzZSgocmVzb2x2ZSkgPT4gewogICAgc2V0VGltZW91dCgoKSA9PiB7CiAgICAgIHJlc29sdmUobnVsbCk7CiAgICB9LCBGSUxFX0NIQU5HRV9USU1FT1VUX01TKTsKICB9KTsKCiAgLy8gV2FpdCBmb3IgdGhlIHVzZXIgdG8gcGljayB0aGUgZmlsZXMuCiAgY29uc3QgZmlsZXMgPSB5aWVsZCB7CiAgICBwcm9taXNlOiBQcm9taXNlLnJhY2UoW3BpY2tlZFByb21pc2UsIHRpbWVvdXRQcm9taXNlLCBjYW5jZWxQcm9taXNlXSksCiAgICByZXNwb25zZTogewogICAgICBhY3Rpb246ICdzdGFydGluZycsCiAgICB9CiAgfTsKCiAgaWYgKCFmaWxlcykgewogICAgcmV0dXJuIHsKICAgICAgcmVzcG9uc2U6IHsKICAgICAgICBhY3Rpb246ICdjb21wbGV0ZScsCiAgICAgIH0KICAgIH07CiAgfQoKICBjYW5jZWwucmVtb3ZlKCk7CgogIC8vIERpc2FibGUgdGhlIGlucHV0IGVsZW1lbnQgc2luY2UgZnVydGhlciBwaWNrcyBhcmUgbm90IGFsbG93ZWQuCiAgaW5wdXRFbGVtZW50LmRpc2FibGVkID0gdHJ1ZTsKCiAgZm9yIChjb25zdCBmaWxlIG9mIGZpbGVzKSB7CiAgICBjb25zdCBsaSA9IGRvY3VtZW50LmNyZWF0ZUVsZW1lbnQoJ2xpJyk7CiAgICBsaS5hcHBlbmQoc3BhbihmaWxlLm5hbWUsIHtmb250V2VpZ2h0OiAnYm9sZCd9KSk7CiAgICBsaS5hcHBlbmQoc3BhbigKICAgICAgICBgKCR7ZmlsZS50eXBlIHx8ICduL2EnfSkgLSAke2ZpbGUuc2l6ZX0gYnl0ZXMsIGAgKwogICAgICAgIGBsYXN0IG1vZGlmaWVkOiAkewogICAgICAgICAgICBmaWxlLmxhc3RNb2RpZmllZERhdGUgPyBmaWxlLmxhc3RNb2RpZmllZERhdGUudG9Mb2NhbGVEYXRlU3RyaW5nKCkgOgogICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAgICAnbi9hJ30gLSBgKSk7CiAgICBjb25zdCBwZXJjZW50ID0gc3BhbignMCUgZG9uZScpOwogICAgbGkuYXBwZW5kQ2hpbGQocGVyY2VudCk7CgogICAgb3V0cHV0RWxlbWVudC5hcHBlbmRDaGlsZChsaSk7CgogICAgY29uc3QgZmlsZURhdGFQcm9taXNlID0gbmV3IFByb21pc2UoKHJlc29sdmUpID0+IHsKICAgICAgY29uc3QgcmVhZGVyID0gbmV3IEZpbGVSZWFkZXIoKTsKICAgICAgcmVhZGVyLm9ubG9hZCA9IChlKSA9PiB7CiAgICAgICAgcmVzb2x2ZShlLnRhcmdldC5yZXN1bHQpOwogICAgICB9OwogICAgICByZWFkZXIucmVhZEFzQXJyYXlCdWZmZXIoZmlsZSk7CiAgICB9KTsKICAgIC8vIFdhaXQgZm9yIHRoZSBkYXRhIHRvIGJlIHJlYWR5LgogICAgbGV0IGZpbGVEYXRhID0geWllbGQgewogICAgICBwcm9taXNlOiBmaWxlRGF0YVByb21pc2UsCiAgICAgIHJlc3BvbnNlOiB7CiAgICAgICAgYWN0aW9uOiAnY29udGludWUnLAogICAgICB9CiAgICB9OwoKICAgIC8vIFVzZSBhIGNodW5rZWQgc2VuZGluZyB0byBhdm9pZCBtZXNzYWdlIHNpemUgbGltaXRzLiBTZWUgYi82MjExNTY2MC4KICAgIGxldCBwb3NpdGlvbiA9IDA7CiAgICB3aGlsZSAocG9zaXRpb24gPCBmaWxlRGF0YS5ieXRlTGVuZ3RoKSB7CiAgICAgIGNvbnN0IGxlbmd0aCA9IE1hdGgubWluKGZpbGVEYXRhLmJ5dGVMZW5ndGggLSBwb3NpdGlvbiwgTUFYX1BBWUxPQURfU0laRSk7CiAgICAgIGNvbnN0IGNodW5rID0gbmV3IFVpbnQ4QXJyYXkoZmlsZURhdGEsIHBvc2l0aW9uLCBsZW5ndGgpOwogICAgICBwb3NpdGlvbiArPSBsZW5ndGg7CgogICAgICBjb25zdCBiYXNlNjQgPSBidG9hKFN0cmluZy5mcm9tQ2hhckNvZGUuYXBwbHkobnVsbCwgY2h1bmspKTsKICAgICAgeWllbGQgewogICAgICAgIHJlc3BvbnNlOiB7CiAgICAgICAgICBhY3Rpb246ICdhcHBlbmQnLAogICAgICAgICAgZmlsZTogZmlsZS5uYW1lLAogICAgICAgICAgZGF0YTogYmFzZTY0LAogICAgICAgIH0sCiAgICAgIH07CiAgICAgIHBlcmNlbnQudGV4dENvbnRlbnQgPQogICAgICAgICAgYCR7TWF0aC5yb3VuZCgocG9zaXRpb24gLyBmaWxlRGF0YS5ieXRlTGVuZ3RoKSAqIDEwMCl9JSBkb25lYDsKICAgIH0KICB9CgogIC8vIEFsbCBkb25lLgogIHlpZWxkIHsKICAgIHJlc3BvbnNlOiB7CiAgICAgIGFjdGlvbjogJ2NvbXBsZXRlJywKICAgIH0KICB9Owp9CgpzY29wZS5nb29nbGUgPSBzY29wZS5nb29nbGUgfHwge307CnNjb3BlLmdvb2dsZS5jb2xhYiA9IHNjb3BlLmdvb2dsZS5jb2xhYiB8fCB7fTsKc2NvcGUuZ29vZ2xlLmNvbGFiLl9maWxlcyA9IHsKICBfdXBsb2FkRmlsZXMsCiAgX3VwbG9hZEZpbGVzQ29udGludWUsCn07Cn0pKHNlbGYpOwo=",
              "ok": true,
              "headers": [
                [
                  "content-type",
                  "application/javascript"
                ]
              ],
              "status": 200,
              "status_text": ""
            }
          },
          "base_uri": "https://localhost:8080/",
          "height": 312
        },
        "outputId": "8268ff19-e2aa-4570-a7e8-0cb458cc4e9b"
      },
      "source": [
        "from google.colab import files\n",
        "fl = files.upload() #Upload the RAR File here\n",
        "!unrar e -y utils3.rar\n",
        "import zeroPadding\n",
        "import normalization\n",
        "import doPCA\n",
        "import modelStatsRecord\n",
        "import averageAccuracy\n",
        "import ssrn_SS_IN"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "     <input type=\"file\" id=\"files-f95e85b5-59b3-4910-82bb-2ef0227f1c61\" name=\"files[]\" multiple disabled />\n",
              "     <output id=\"result-f95e85b5-59b3-4910-82bb-2ef0227f1c61\">\n",
              "      Upload widget is only available when the cell has been executed in the\n",
              "      current browser session. Please rerun this cell to enable.\n",
              "      </output>\n",
              "      <script src=\"/nbextensions/google.colab/files.js\"></script> "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "UNRAR 5.50 freeware      Copyright (c) 1993-2017 Alexander Roshal\n",
            "\n",
            "\n",
            "Extracting from utils3.rar\n",
            "\n",
            "Extracting  averageAccuracy.py                                           \b\b\b\b  2%\b\b\b\b\b  OK \n",
            "Extracting  doPCA.py                                                     \b\b\b\b  4%\b\b\b\b\b  OK \n",
            "Extracting  modelStatsRecord.py                                          \b\b\b\b 11%\b\b\b\b\b  OK \n",
            "Extracting  New Text Document.txt                                        \b\b\b\b 12%\b\b\b\b\b  OK \n",
            "Extracting  normalization.py                                             \b\b\b\b 14%\b\b\b\b\b  OK \n",
            "Extracting  Rawdata_3D_ResNet.py                                         \b\b\b\b 42%\b\b\b\b\b  OK \n",
            "Extracting  Real3D_conv.py                                               \b\b\b\b 67%\b\b\b\b\b  OK \n",
            "Extracting  ssrn_SS_IN.py                                                \b\b\b\b 96%\b\b\b\b\b  OK \n",
            "Extracting  zeroPadding.py                                               \b\b\b\b 99%\b\b\b\b\b  OK \n",
            "All OK\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FsNPWyeNiOYr",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "## GLOBAL VARIABLES\n",
        "dataset = ['PU']#,'IP','PU']#\n",
        "test_ratio = [0.8]#\n",
        "windowSize = [15]#\n",
        "mp = {'IP':16,'PU':9,'SA':16}"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "q23JMm5FiOpO",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "92f87f1c-5d71-4cbb-927a-21e243d4db43"
      },
      "source": [
        "for d in dataset:\n",
        "  for t in test_ratio:\n",
        "    for w in windowSize:\n",
        "      X , Y = loadData(d)\n",
        "      K = 30 if d == 'IP' else 15\n",
        "      X,pca = applyPCA(X,numComponents=K)\n",
        "      X, Y = createImageCubes(X, Y, windowSize=w)\n",
        "      X.shape, Y.shape\n",
        "      sz = X.shape[0]//10\n",
        "      Y = to_categorical(Y)\n",
        "      X = X.reshape((X.shape[0],w,w,K,1))\n",
        "      train_X, test_X, train_Y, test_Y = splitTrainTestSet(X, Y, t)\n",
        "      val_X = test_X[:sz]\n",
        "      test_X = test_X[sz:]\n",
        "      val_Y = test_Y[:sz]\n",
        "      test_Y = test_Y[sz:]\n",
        "      KAPPA_RES_SS4 = []\n",
        "      OA_RES_SS4 = []\n",
        "      AA_RES_SS4 = []\n",
        "      TRAINING_TIME_RES_SS4 = []\n",
        "      TESTING_TIME_RES_SS4 = []\n",
        "      ELEMENT_ACC_RES_SS4 = np.zeros((1, mp[d]))\n",
        "\n",
        "\n",
        "      for index_iter in range(1):\n",
        "          #try:\n",
        "            print(\"# %d Iteration\" % (index_iter + 1))\n",
        "\n",
        "            # save the best validated model \n",
        "\n",
        "            best_weights_RES_path_ss4 = data_path+ '/' + str(\n",
        "                index_iter + 1) + '_NEW_KERNEL32_B_32_hashing'+str(w)+'_'+str(int((1-t)*100))+'.hdf5'\n",
        "\n",
        "            \n",
        "            \n",
        "            #Bilinear\n",
        "            #\n",
        "            #\n",
        "            \n",
        "            model_res4_SS_BN = res4_model_ss()\n",
        "\n",
        "            earlyStopping6 = kcallbacks.EarlyStopping(monitor='val_loss', patience=200, verbose=1, mode='auto')\n",
        "            saveBestModel6 = kcallbacks.ModelCheckpoint(best_weights_RES_path_ss4, monitor='val_loss', verbose=1,\n",
        "                                                        save_best_only=True,\n",
        "                                                        mode='auto')\n",
        "\n",
        "            tic6 = time.clock()\n",
        "            print(train_X.shape, test_X.shape)\n",
        "            history_res4_SS_BN = model_res4_SS_BN.fit(\n",
        "                train_X, train_Y,\n",
        "                validation_data=(val_X, val_Y),\n",
        "                batch_size=256,\n",
        "                nb_epoch=50, shuffle=True, callbacks=[earlyStopping6,saveBestModel6])\n",
        "            toc6 = time.clock()\n",
        "\n",
        "            tic7 = time.clock()\n",
        "            loss_and_metrics_res4_SS_BN = model_res4_SS_BN.evaluate(\n",
        "                test_X, test_Y,\n",
        "                batch_size=32)\n",
        "            toc7 = time.clock()\n",
        "\n",
        "            print('3D RES_SS4 without BN Training Time: ', toc6 - tic6)\n",
        "            print('3D RES_SS4 without BN Test time:', toc7 - tic7)\n",
        "\n",
        "            print('3D RES_SS4 without BN Test score:', loss_and_metrics_res4_SS_BN[0])\n",
        "            print('3D RES_SS4 without BN Test accuracy:', loss_and_metrics_res4_SS_BN[1])\n",
        "\n",
        "            print(history_res4_SS_BN.history.keys())\n",
        "\n",
        "            pred_test_res4 = model_res4_SS_BN.predict(test_X).argmax(axis=1)\n",
        "            collections.Counter(pred_test_res4)\n",
        "            gt_test = test_Y\n",
        "            overall_acc_res4 = metrics.accuracy_score(pred_test_res4, gt_test.argmax(axis=1))\n",
        "            confusion_matrix_res4 = metrics.confusion_matrix(pred_test_res4, gt_test.argmax(axis=1))\n",
        "            each_acc_res4, average_acc_res4 = averageAccuracy.AA_andEachClassAccuracy(confusion_matrix_res4)\n",
        "            kappa = metrics.cohen_kappa_score(pred_test_res4, gt_test.argmax(axis=1))\n",
        "            KAPPA_RES_SS4.append(kappa)\n",
        "            OA_RES_SS4.append(overall_acc_res4)\n",
        "            AA_RES_SS4.append(average_acc_res4)\n",
        "            TRAINING_TIME_RES_SS4.append(toc6 - tic6)\n",
        "            TESTING_TIME_RES_SS4.append(toc7 - tic7)\n",
        "            ELEMENT_ACC_RES_SS4[index_iter, :] = each_acc_res4\n",
        "\n",
        "            print(\"3D RESNET_SS4 without BN training finished.\")\n",
        "            print(\"# %d Iteration\" % (index_iter + 1))\n",
        "\n",
        "            modelStatsRecord.outputStats(KAPPA_RES_SS4, OA_RES_SS4, AA_RES_SS4, ELEMENT_ACC_RES_SS4,\n",
        "                                     TRAINING_TIME_RES_SS4, TESTING_TIME_RES_SS4,\n",
        "                                     history_res4_SS_BN, loss_and_metrics_res4_SS_BN, mp[d],\n",
        "                                     data_path + '/IN_train_RESNET8_KERNEL64_20_B_32_Fuse_'+str(w)+'_'+str(int((1-t)*100))+'.txt',\n",
        "                                     data_path + '/IN_train_RESNET8_KERNEL64_20_B_32_PU_element_9_Fuse_'+str(w)+'_'+str(int((1-t)*100))+'.txt')"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "# 1 Iteration\n",
            "input shape: (None, 15, 15, 5, 32)\n",
            "(?, 1, 1, 32)\n",
            "input shape: (None, 13, 13, 1, 32)\n",
            "(?, 1, 1, 32)\n",
            "(None, 13, 13, 32)\n",
            "call\n",
            "output\n",
            "(None, 13, 13, 32)\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_3 (InputLayer)            (None, 15, 15, 15, 1 0                                            \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_17 (Conv3D)              (None, 15, 15, 5, 32 256         input_3[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_17 (BatchNo (None, 15, 15, 5, 32 128         conv3d_17[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_17 (Activation)      (None, 15, 15, 5, 32 0           batch_normalization_17[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_18 (Conv3D)              (None, 15, 15, 5, 32 9248        activation_17[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_18 (BatchNo (None, 15, 15, 5, 32 128         conv3d_18[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_18 (Activation)      (None, 15, 15, 5, 32 0           batch_normalization_18[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_19 (Conv3D)              (None, 15, 15, 5, 32 9248        activation_18[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling3d_5 (Glo (None, 32)           0           conv3d_19[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling3d_5 (GlobalM (None, 32)           0           conv3d_19[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_13 (Reshape)            (None, 1, 1, 32)     0           global_average_pooling3d_5[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "reshape_14 (Reshape)            (None, 1, 1, 32)     0           global_max_pooling3d_5[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "dense_19 (Dense)                (None, 1, 1, 2)      64          reshape_13[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "dense_21 (Dense)                (None, 1, 1, 2)      64          reshape_14[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "dense_20 (Dense)                (None, 1, 1, 32)     64          dense_19[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "dense_22 (Dense)                (None, 1, 1, 32)     64          dense_21[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "multiply_9 (Multiply)           (None, 15, 15, 5, 32 0           conv3d_19[0][0]                  \n",
            "                                                                 dense_20[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "multiply_10 (Multiply)          (None, 15, 15, 5, 32 0           conv3d_19[0][0]                  \n",
            "                                                                 dense_22[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "maximum_5 (Maximum)             (None, 15, 15, 5, 32 0           multiply_9[0][0]                 \n",
            "                                                                 multiply_10[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "add_5 (Add)                     (None, 15, 15, 5, 32 0           maximum_5[0][0]                  \n",
            "                                                                 conv3d_19[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_19 (BatchNo (None, 15, 15, 5, 32 128         add_5[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "activation_19 (Activation)      (None, 15, 15, 5, 32 0           batch_normalization_19[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_20 (Conv3D)              (None, 15, 15, 1, 12 20608       activation_19[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_20 (BatchNo (None, 15, 15, 1, 12 512         conv3d_20[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_20 (Activation)      (None, 15, 15, 1, 12 0           batch_normalization_20[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "reshape_15 (Reshape)            (None, 15, 15, 128,  0           activation_20[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_21 (Conv3D)              (None, 13, 13, 1, 32 36896       reshape_15[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_21 (BatchNo (None, 13, 13, 1, 32 128         conv3d_21[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_21 (Activation)      (None, 13, 13, 1, 32 0           batch_normalization_21[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_24 (Conv3D)              (None, 13, 13, 1, 32 1056        activation_21[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling3d_6 (Glo (None, 32)           0           conv3d_24[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "global_max_pooling3d_6 (GlobalM (None, 32)           0           conv3d_24[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_16 (Reshape)            (None, 1, 1, 32)     0           global_average_pooling3d_6[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "reshape_17 (Reshape)            (None, 1, 1, 32)     0           global_max_pooling3d_6[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "dense_23 (Dense)                (None, 1, 1, 2)      64          reshape_16[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "dense_25 (Dense)                (None, 1, 1, 2)      64          reshape_17[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_22 (Conv3D)              (None, 13, 13, 1, 32 9248        activation_21[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dense_24 (Dense)                (None, 1, 1, 32)     64          dense_23[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "dense_26 (Dense)                (None, 1, 1, 32)     64          dense_25[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_22 (BatchNo (None, 13, 13, 1, 32 128         conv3d_22[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "multiply_11 (Multiply)          (None, 13, 13, 1, 32 0           conv3d_24[0][0]                  \n",
            "                                                                 dense_24[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "multiply_12 (Multiply)          (None, 13, 13, 1, 32 0           conv3d_24[0][0]                  \n",
            "                                                                 dense_26[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "activation_22 (Activation)      (None, 13, 13, 1, 32 0           batch_normalization_22[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "maximum_6 (Maximum)             (None, 13, 13, 1, 32 0           multiply_11[0][0]                \n",
            "                                                                 multiply_12[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "conv3d_23 (Conv3D)              (None, 13, 13, 1, 32 9248        activation_22[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "add_6 (Add)                     (None, 13, 13, 1, 32 0           maximum_6[0][0]                  \n",
            "                                                                 conv3d_23[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_23 (BatchNo (None, 13, 13, 1, 32 128         add_6[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "activation_23 (Activation)      (None, 13, 13, 1, 32 0           batch_normalization_23[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_24 (BatchNo (None, 13, 13, 1, 32 128         activation_23[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "activation_24 (Activation)      (None, 13, 13, 1, 32 0           batch_normalization_24[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "reshape_18 (Reshape)            (None, 13, 13, 32)   0           activation_24[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "bo_f__pooling_3 (BoF_Pooling)   (None, 64)           2112        reshape_18[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "dropout_3 (Dropout)             (None, 64)           0           bo_f__pooling_3[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "dense_27 (Dense)                (None, 9)            585         dropout_3[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 100,425\n",
            "Trainable params: 99,721\n",
            "Non-trainable params: 704\n",
            "__________________________________________________________________________________________________\n",
            "(8555, 15, 15, 15, 1) (29944, 15, 15, 15, 1)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:53: UserWarning:\n",
            "\n",
            "The `nb_epoch` argument in `fit` has been renamed `epochs`.\n",
            "\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Train on 8555 samples, validate on 4277 samples\n",
            "Epoch 1/50\n",
            "8555/8555 [==============================] - 5s 601us/step - loss: 7.7154 - acc: 0.0781 - val_loss: 12.1824 - val_acc: 0.0872\n",
            "\n",
            "Epoch 00001: val_loss improved from inf to 12.18237, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 2/50\n",
            "8555/8555 [==============================] - 3s 308us/step - loss: 7.5840 - acc: 0.2354 - val_loss: 12.1764 - val_acc: 0.1136\n",
            "\n",
            "Epoch 00002: val_loss improved from 12.18237 to 12.17643, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 3/50\n",
            "8555/8555 [==============================] - 3s 307us/step - loss: 5.2934 - acc: 0.3963 - val_loss: 7.3475 - val_acc: 0.4386\n",
            "\n",
            "Epoch 00003: val_loss improved from 12.17643 to 7.34753, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 4/50\n",
            "8555/8555 [==============================] - 3s 307us/step - loss: 2.3370 - acc: 0.6096 - val_loss: 6.6623 - val_acc: 0.0311\n",
            "\n",
            "Epoch 00004: val_loss improved from 7.34753 to 6.66234, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 5/50\n",
            "8555/8555 [==============================] - 3s 308us/step - loss: 1.0326 - acc: 0.7510 - val_loss: 1.6694 - val_acc: 0.6079\n",
            "\n",
            "Epoch 00005: val_loss improved from 6.66234 to 1.66938, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 6/50\n",
            "8555/8555 [==============================] - 3s 307us/step - loss: 0.5971 - acc: 0.8489 - val_loss: 0.8053 - val_acc: 0.8057\n",
            "\n",
            "Epoch 00006: val_loss improved from 1.66938 to 0.80525, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 7/50\n",
            "8555/8555 [==============================] - 3s 307us/step - loss: 0.4349 - acc: 0.8860 - val_loss: 2.0213 - val_acc: 0.6037\n",
            "\n",
            "Epoch 00007: val_loss did not improve from 0.80525\n",
            "Epoch 8/50\n",
            "8555/8555 [==============================] - 3s 306us/step - loss: 0.3551 - acc: 0.9191 - val_loss: 0.2559 - val_acc: 0.9479\n",
            "\n",
            "Epoch 00008: val_loss improved from 0.80525 to 0.25586, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 9/50\n",
            "8555/8555 [==============================] - 3s 308us/step - loss: 0.2896 - acc: 0.9319 - val_loss: 0.1812 - val_acc: 0.9764\n",
            "\n",
            "Epoch 00009: val_loss improved from 0.25586 to 0.18118, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 10/50\n",
            "8555/8555 [==============================] - 3s 305us/step - loss: 0.2797 - acc: 0.9414 - val_loss: 0.1346 - val_acc: 0.9867\n",
            "\n",
            "Epoch 00010: val_loss improved from 0.18118 to 0.13464, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 11/50\n",
            "8555/8555 [==============================] - 3s 305us/step - loss: 0.2109 - acc: 0.9578 - val_loss: 0.1136 - val_acc: 0.9911\n",
            "\n",
            "Epoch 00011: val_loss improved from 0.13464 to 0.11362, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 12/50\n",
            "8555/8555 [==============================] - 3s 303us/step - loss: 0.1872 - acc: 0.9681 - val_loss: 0.0913 - val_acc: 0.9944\n",
            "\n",
            "Epoch 00012: val_loss improved from 0.11362 to 0.09132, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 13/50\n",
            "8555/8555 [==============================] - 3s 307us/step - loss: 0.1609 - acc: 0.9737 - val_loss: 0.1030 - val_acc: 0.9906\n",
            "\n",
            "Epoch 00013: val_loss did not improve from 0.09132\n",
            "Epoch 14/50\n",
            "8555/8555 [==============================] - 3s 308us/step - loss: 0.1745 - acc: 0.9708 - val_loss: 0.0925 - val_acc: 0.9928\n",
            "\n",
            "Epoch 00014: val_loss did not improve from 0.09132\n",
            "Epoch 15/50\n",
            "8555/8555 [==============================] - 3s 307us/step - loss: 0.1448 - acc: 0.9784 - val_loss: 0.0851 - val_acc: 0.9942\n",
            "\n",
            "Epoch 00015: val_loss improved from 0.09132 to 0.08507, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 16/50\n",
            "8555/8555 [==============================] - 3s 306us/step - loss: 0.1810 - acc: 0.9797 - val_loss: 0.0854 - val_acc: 0.9937\n",
            "\n",
            "Epoch 00016: val_loss did not improve from 0.08507\n",
            "Epoch 17/50\n",
            "8555/8555 [==============================] - 3s 304us/step - loss: 0.1313 - acc: 0.9846 - val_loss: 0.0789 - val_acc: 0.9965\n",
            "\n",
            "Epoch 00017: val_loss improved from 0.08507 to 0.07887, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 18/50\n",
            "8555/8555 [==============================] - 3s 305us/step - loss: 0.1479 - acc: 0.9812 - val_loss: 0.1110 - val_acc: 0.9832\n",
            "\n",
            "Epoch 00018: val_loss did not improve from 0.07887\n",
            "Epoch 19/50\n",
            "8555/8555 [==============================] - 3s 304us/step - loss: 0.1048 - acc: 0.9909 - val_loss: 0.0757 - val_acc: 0.9965\n",
            "\n",
            "Epoch 00019: val_loss improved from 0.07887 to 0.07567, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 20/50\n",
            "8555/8555 [==============================] - 3s 305us/step - loss: 0.1215 - acc: 0.9862 - val_loss: 0.0826 - val_acc: 0.9949\n",
            "\n",
            "Epoch 00020: val_loss did not improve from 0.07567\n",
            "Epoch 21/50\n",
            "8555/8555 [==============================] - 3s 306us/step - loss: 0.1250 - acc: 0.9863 - val_loss: 0.0726 - val_acc: 0.9967\n",
            "\n",
            "Epoch 00021: val_loss improved from 0.07567 to 0.07260, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 22/50\n",
            "8555/8555 [==============================] - 3s 305us/step - loss: 0.1012 - acc: 0.9904 - val_loss: 0.2129 - val_acc: 0.9425\n",
            "\n",
            "Epoch 00022: val_loss did not improve from 0.07260\n",
            "Epoch 23/50\n",
            "8555/8555 [==============================] - 3s 304us/step - loss: 0.1003 - acc: 0.9906 - val_loss: 0.0942 - val_acc: 0.9883\n",
            "\n",
            "Epoch 00023: val_loss did not improve from 0.07260\n",
            "Epoch 24/50\n",
            "8555/8555 [==============================] - 3s 304us/step - loss: 0.1134 - acc: 0.9876 - val_loss: 0.0740 - val_acc: 0.9970\n",
            "\n",
            "Epoch 00024: val_loss did not improve from 0.07260\n",
            "Epoch 25/50\n",
            "8555/8555 [==============================] - 3s 306us/step - loss: 0.0904 - acc: 0.9918 - val_loss: 0.0722 - val_acc: 0.9974\n",
            "\n",
            "Epoch 00025: val_loss improved from 0.07260 to 0.07218, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 26/50\n",
            "8555/8555 [==============================] - 3s 308us/step - loss: 0.1056 - acc: 0.9889 - val_loss: 0.3290 - val_acc: 0.9422\n",
            "\n",
            "Epoch 00026: val_loss did not improve from 0.07218\n",
            "Epoch 27/50\n",
            "8555/8555 [==============================] - 3s 306us/step - loss: 0.0934 - acc: 0.9921 - val_loss: 0.0653 - val_acc: 0.9986\n",
            "\n",
            "Epoch 00027: val_loss improved from 0.07218 to 0.06526, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 28/50\n",
            "8555/8555 [==============================] - 3s 306us/step - loss: 0.0981 - acc: 0.9912 - val_loss: 0.0685 - val_acc: 0.9972\n",
            "\n",
            "Epoch 00028: val_loss did not improve from 0.06526\n",
            "Epoch 29/50\n",
            "8555/8555 [==============================] - 3s 307us/step - loss: 0.0845 - acc: 0.9933 - val_loss: 0.0711 - val_acc: 0.9970\n",
            "\n",
            "Epoch 00029: val_loss did not improve from 0.06526\n",
            "Epoch 30/50\n",
            "8555/8555 [==============================] - 3s 310us/step - loss: 0.0855 - acc: 0.9932 - val_loss: 0.0661 - val_acc: 0.9979\n",
            "\n",
            "Epoch 00030: val_loss did not improve from 0.06526\n",
            "Epoch 31/50\n",
            "8555/8555 [==============================] - 3s 312us/step - loss: 0.0870 - acc: 0.9922 - val_loss: 0.4311 - val_acc: 0.9528\n",
            "\n",
            "Epoch 00031: val_loss did not improve from 0.06526\n",
            "Epoch 32/50\n",
            "8555/8555 [==============================] - 3s 307us/step - loss: 0.0939 - acc: 0.9919 - val_loss: 0.0637 - val_acc: 0.9979\n",
            "\n",
            "Epoch 00032: val_loss improved from 0.06526 to 0.06368, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 33/50\n",
            "8555/8555 [==============================] - 3s 305us/step - loss: 0.0871 - acc: 0.9942 - val_loss: 0.0638 - val_acc: 0.9988\n",
            "\n",
            "Epoch 00033: val_loss did not improve from 0.06368\n",
            "Epoch 34/50\n",
            "8555/8555 [==============================] - 3s 304us/step - loss: 0.0762 - acc: 0.9952 - val_loss: 0.0726 - val_acc: 0.9958\n",
            "\n",
            "Epoch 00034: val_loss did not improve from 0.06368\n",
            "Epoch 35/50\n",
            "8555/8555 [==============================] - 3s 303us/step - loss: 0.1041 - acc: 0.9906 - val_loss: 0.0643 - val_acc: 0.9979\n",
            "\n",
            "Epoch 00035: val_loss did not improve from 0.06368\n",
            "Epoch 36/50\n",
            "8555/8555 [==============================] - 3s 303us/step - loss: 0.0883 - acc: 0.9917 - val_loss: 0.1083 - val_acc: 0.9876\n",
            "\n",
            "Epoch 00036: val_loss did not improve from 0.06368\n",
            "Epoch 37/50\n",
            "8555/8555 [==============================] - 3s 305us/step - loss: 0.0808 - acc: 0.9936 - val_loss: 0.0749 - val_acc: 0.9956\n",
            "\n",
            "Epoch 00037: val_loss did not improve from 0.06368\n",
            "Epoch 38/50\n",
            "8555/8555 [==============================] - 3s 306us/step - loss: 0.0883 - acc: 0.9929 - val_loss: 0.0986 - val_acc: 0.9935\n",
            "\n",
            "Epoch 00038: val_loss did not improve from 0.06368\n",
            "Epoch 39/50\n",
            "8555/8555 [==============================] - 3s 304us/step - loss: 0.0839 - acc: 0.9937 - val_loss: 0.0753 - val_acc: 0.9949\n",
            "\n",
            "Epoch 00039: val_loss did not improve from 0.06368\n",
            "Epoch 40/50\n",
            "8555/8555 [==============================] - 3s 305us/step - loss: 0.0863 - acc: 0.9938 - val_loss: 0.0577 - val_acc: 0.9986\n",
            "\n",
            "Epoch 00040: val_loss improved from 0.06368 to 0.05767, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 41/50\n",
            "8555/8555 [==============================] - 3s 305us/step - loss: 0.0668 - acc: 0.9964 - val_loss: 0.0649 - val_acc: 0.9953\n",
            "\n",
            "Epoch 00041: val_loss did not improve from 0.05767\n",
            "Epoch 42/50\n",
            "8555/8555 [==============================] - 3s 302us/step - loss: 0.1018 - acc: 0.9899 - val_loss: 0.0745 - val_acc: 0.9942\n",
            "\n",
            "Epoch 00042: val_loss did not improve from 0.05767\n",
            "Epoch 43/50\n",
            "8555/8555 [==============================] - 3s 302us/step - loss: 0.0765 - acc: 0.9943 - val_loss: 0.0590 - val_acc: 0.9984\n",
            "\n",
            "Epoch 00043: val_loss did not improve from 0.05767\n",
            "Epoch 44/50\n",
            "8555/8555 [==============================] - 3s 303us/step - loss: 0.0647 - acc: 0.9960 - val_loss: 0.0611 - val_acc: 0.9974\n",
            "\n",
            "Epoch 00044: val_loss did not improve from 0.05767\n",
            "Epoch 45/50\n",
            "8555/8555 [==============================] - 3s 305us/step - loss: 0.0678 - acc: 0.9961 - val_loss: 0.0569 - val_acc: 0.9981\n",
            "\n",
            "Epoch 00045: val_loss improved from 0.05767 to 0.05687, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "Epoch 46/50\n",
            "8555/8555 [==============================] - 3s 305us/step - loss: 0.0620 - acc: 0.9970 - val_loss: 0.0738 - val_acc: 0.9918\n",
            "\n",
            "Epoch 00046: val_loss did not improve from 0.05687\n",
            "Epoch 47/50\n",
            "8555/8555 [==============================] - 3s 304us/step - loss: 0.0743 - acc: 0.9940 - val_loss: 0.0597 - val_acc: 0.9970\n",
            "\n",
            "Epoch 00047: val_loss did not improve from 0.05687\n",
            "Epoch 48/50\n",
            "8555/8555 [==============================] - 3s 308us/step - loss: 0.0775 - acc: 0.9939 - val_loss: 0.0617 - val_acc: 0.9967\n",
            "\n",
            "Epoch 00048: val_loss did not improve from 0.05687\n",
            "Epoch 49/50\n",
            "8555/8555 [==============================] - 3s 308us/step - loss: 0.0593 - acc: 0.9973 - val_loss: 0.1524 - val_acc: 0.9811\n",
            "\n",
            "Epoch 00049: val_loss did not improve from 0.05687\n",
            "Epoch 50/50\n",
            "8555/8555 [==============================] - 3s 303us/step - loss: 0.0637 - acc: 0.9947 - val_loss: 0.0545 - val_acc: 0.9984\n",
            "\n",
            "Epoch 00050: val_loss improved from 0.05687 to 0.05453, saving model to /content/gdrive/My Drive/FuseNet/1_NEW_KERNEL32_B_32_hashing15_19.hdf5\n",
            "29944/29944 [==============================] - 5s 155us/step\n",
            "3D RES_SS4 without BN Training Time:  144.873248\n",
            "3D RES_SS4 without BN Test time: 5.308132999999998\n",
            "3D RES_SS4 without BN Test score: 0.0560191359282368\n",
            "3D RES_SS4 without BN Test accuracy: 0.9984637990916377\n",
            "dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])\n",
            "3D RESNET_SS4 without BN training finished.\n",
            "# 1 Iteration\n",
            "Test score: 0.0560191359282368\n",
            "Test accuracy: 0.9984637990916377\n",
            "dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}