classifier-pretrained-with-convolutional-autoencoder.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"convolutional-autoencoder-classifier.ipynb","provenance":[{"file_id":"1Hff4QhIpiC8EHM_ofp77zIWvFwpyhSpV","timestamp":1603708478260}],"collapsed_sections":[],"toc_visible":true,"authorship_tag":"ABX9TyNsRtUgIT+HwMew2t+Pw/5f"},"kernelspec":{"name":"python3","display_name":"Python 3"}},"cells":[{"cell_type":"markdown","metadata":{"id":"eZ5vLdtnHUmn"},"source":["#### Copyright 2020 Connaissance Publishing."]},{"cell_type":"code","metadata":{"id":"e0itxBa9HWgS"},"source":["# Licensed under the Apache License, Version 2.0 (the \"License\");\n","# you may not use this file except in compliance with the License.\n","# You may obtain a copy of the License at\n","#\n","# https://www.apache.org/licenses/LICENSE-2.0\n","#\n","# Unless required by applicable law or agreed to in writing, software\n","# distributed under the License is distributed on an \"AS IS\" BASIS,\n","# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n","# See the License for the specific language governing permissions and\n","# limitations under the License."],"execution_count":null,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"KC5U1MKio4A-"},"source":["# **Understanding Deep Learning: Application in Rare Event Prediction**\n","\n","Author: Chitta Ranjan, Ph.D.\n","\n","[ProcessMiner, Inc.](https://processminer.com)\n","\n","cranjan@processminer.com\n","\n","LinkedIn: https://www.linkedin.com/in/chitta-ranjan-b0851911/\n","\n","Medium: https://medium.com/@cran2367\n","\n","\n","---"]},{"cell_type":"markdown","metadata":{"id":"efUIQHrlo8-f"},"source":["# Convolutional Classifier pretrained with Autoencoder"]},{"cell_type":"markdown","metadata":{"id":"ElBWj90KHqFo"},"source":["# Connection"]},{"cell_type":"code","metadata":{"id":"EC_UZmPqHmB4","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1606005131723,"user_tz":300,"elapsed":26050,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"77164827-fbba-4fb9-afc8-82a374436da5"},"source":["from google.colab import drive\n","drive.mount('/content/gdrive', force_remount=True)\n","root_path = 'gdrive/My Drive/understanding-deep-learning/codebase'\n","import os\n","os.chdir(root_path)"],"execution_count":null,"outputs":[{"output_type":"stream","text":["Mounted at /content/gdrive\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"TVMhD2fRH34v"},"source":["os.getcwd()"],"execution_count":null,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"l6MvUvGgIAxy"},"source":["# Libraries"]},{"cell_type":"code","metadata":{"id":"_m8snzeAH9Vl"},"source":["%matplotlib inline\n","\n","import pandas as pd\n","import numpy as np\n","\n","%tensorflow_version 2.x\n","import tensorflow as tf\n","from tensorflow.keras import optimizers\n","from tensorflow.keras.models import Model\n","from tensorflow.keras.models import Sequential\n","from tensorflow.keras.layers import Input \n","from tensorflow.keras.layers import Dense\n","from tensorflow.keras.layers import Dropout\n","from tensorflow.keras.layers import Conv1D\n","from tensorflow.keras.layers import MaxPool1D\n","from tensorflow.keras.layers import Conv1DTranspose\n","from tensorflow.keras.layers import BatchNormalization\n","from tensorflow.keras.layers import Reshape\n","from tensorflow.keras.layers import TimeDistributed\n","from tensorflow.keras.layers import Flatten\n","from tensorflow.keras.callbacks import ModelCheckpoint\n","from tensorflow.keras.callbacks import TensorBoard\n","from tensorflow.python.keras import backend as K\n","\n","from sklearn.preprocessing import StandardScaler\n","from sklearn.model_selection import train_test_split\n","\n","import matplotlib.pyplot as plt\n","import seaborn as sns\n","\n","# user-defined libraries\n","import utilities.datapreprocessing as dp\n","import utilities.performancemetrics as pm\n","import utilities.simpleplots as sp\n","import utilities.reconstructionperformance as rp\n","\n","from numpy.random import seed\n","seed(1)\n","\n","SEED = 123  # used to help randomly select the data points\n","DATA_SPLIT_PCT = 0.2\n","\n","from pylab import rcParams\n","rcParams['figure.figsize'] = 6, 4\n","plt.rcParams.update({'font.size': 22})"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"Dm2QJeEXIljC","colab":{"base_uri":"https://localhost:8080/","height":37},"executionInfo":{"status":"ok","timestamp":1606005138231,"user_tz":300,"elapsed":237,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"ba16972b-1948-481c-c484-d08a3be8683c"},"source":["# Check the tensorflow version.\n","tf.__version__"],"execution_count":null,"outputs":[{"output_type":"execute_result","data":{"application/vnd.google.colaboratory.intrinsic+json":{"type":"string"},"text/plain":["'2.3.0'"]},"metadata":{"tags":[]},"execution_count":4}]},{"cell_type":"markdown","metadata":{"id":"wC4DQWDCNDns"},"source":["# Data Preparation"]},{"cell_type":"markdown","metadata":{"id":"vfAum_oRLWWU"},"source":["The data is taken from https://arxiv.org/abs/1809.10717. Please use this source for any citation."]},{"cell_type":"code","metadata":{"id":"ZM-RR7qcIr5K","colab":{"base_uri":"https://localhost:8080/","height":343},"executionInfo":{"status":"ok","timestamp":1606005147962,"user_tz":300,"elapsed":1100,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"ef3579b3-3e3b-4bc3-bd46-23bd863fa29c"},"source":["'''Download data here:\n","https://docs.google.com/forms/d/e/1FAIpQLSdPCkTFW1wTp6tAP1C1BhT7HWTQ60-GfCGvWs9Ca4WbR0FIdg/viewform\n","'''\n","df = pd.read_csv(\"data/processminer-sheet-break-rare-event-dataset.csv\")\n","df.head(n=5)  # visualize the data."],"execution_count":null,"outputs":[{"output_type":"execute_result","data":{"text/html":["<div>\n","<style scoped>\n","    .dataframe tbody tr th:only-of-type {\n","        vertical-align: middle;\n","    }\n","\n","    .dataframe tbody tr th {\n","        vertical-align: top;\n","    }\n","\n","    .dataframe thead th {\n","        text-align: right;\n","    }\n","</style>\n","<table border=\"1\" class=\"dataframe\">\n","  <thead>\n","    <tr style=\"text-align: right;\">\n","      <th></th>\n","      <th>DateTime</th>\n","      <th>SheetBreak</th>\n","      <th>RSashScanAvg</th>\n","      <th>CT#1 BLADE PSI</th>\n","      <th>P4 CT#2 BLADE PSI</th>\n","      <th>Bleached GWD Flow</th>\n","      <th>ShwerTemp</th>\n","      <th>BlndStckFloTPD</th>\n","      <th>C1 BW SPREAD CD</th>\n","      <th>RS BW SPREAD CD</th>\n","      <th>C1 BW SPREAD MD</th>\n","      <th>RS BW SPREAD MD</th>\n","      <th>C1 BW SCAN AVG</th>\n","      <th>RS BW SCAN AVG</th>\n","      <th>CoatBrkFlo</th>\n","      <th>Clay Flow</th>\n","      <th>CouchLoVac</th>\n","      <th>COUCH VAC</th>\n","      <th>4PrsTopLd</th>\n","      <th>4PrsBotLod</th>\n","      <th>CalndrDrw</th>\n","      <th>2DryrDrw</th>\n","      <th>3DryrDrw</th>\n","      <th>4DryrDraw</th>\n","      <th>1PrsTopDrw</th>\n","      <th>4PrsBotDrw</th>\n","      <th>FanPmpSpd</th>\n","      <th>FlBxHdrVac</th>\n","      <th>FlatBxVac</th>\n","      <th>Grade&amp;Bwt</th>\n","      <th>UnblGWDFlo</th>\n","      <th>Hdbox pH</th>\n","      <th>HdBxLiqLvl</th>\n","      <th>TotHead\"</th>\n","      <th>HorzSlcPos</th>\n","      <th>KraftFlow</th>\n","      <th>CouchLoad</th>\n","      <th>C1MoSprdCD</th>\n","      <th>RSMoSprdCD</th>\n","      <th>C1MoSprdMD</th>\n","      <th>RSMoSprdMD</th>\n","      <th>RL MoisAct</th>\n","      <th>PrScrRjFlo</th>\n","      <th>RwBrkFlo</th>\n","      <th>RcycFbrFlo</th>\n","      <th>RetnAidFlo</th>\n","      <th>RUSH DRAG</th>\n","      <th>Rush/Drag</th>\n","      <th>SilicaFlo</th>\n","      <th>HBxSlcTemp</th>\n","      <th>SodAlumFlo</th>\n","      <th>CouchSpd</th>\n","      <th>MachSpd</th>\n","      <th>1PrsTopSpd</th>\n","      <th>4PrsBotSpd</th>\n","      <th>WtNStarFlo</th>\n","      <th>BasWgtFlo</th>\n","      <th>TMP Flow</th>\n","      <th>HBxTotHead</th>\n","      <th>TrayCons</th>\n","      <th>UpprHdTmpRL</th>\n","      <th>VertSlcPos</th>\n","      <th>EventPress</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>5/1/99 0:00</td>\n","      <td>0</td>\n","      <td>0.376665</td>\n","      <td>-4.596435</td>\n","      <td>-4.095756</td>\n","      <td>13.497687</td>\n","      <td>-0.118830</td>\n","      <td>-20.669883</td>\n","      <td>0.000732</td>\n","      <td>-0.061114</td>\n","      <td>-0.059966</td>\n","      <td>-0.038189</td>\n","      <td>0.877951</td>\n","      <td>-0.052959</td>\n","      <td>-13.306135</td>\n","      <td>0.101068</td>\n","      <td>0.041800</td>\n","      <td>0.199901</td>\n","      <td>-2.327329</td>\n","      <td>-0.944167</td>\n","      <td>3.075199</td>\n","      <td>0.123154</td>\n","      <td>-0.104334</td>\n","      <td>-0.570710</td>\n","      <td>-9.784456</td>\n","      <td>0.355960</td>\n","      <td>15.842819</td>\n","      <td>-0.451974</td>\n","      <td>-0.105282</td>\n","      <td>96</td>\n","      <td>-134.277862</td>\n","      <td>0.058726</td>\n","      <td>-0.021645</td>\n","      <td>9.366755</td>\n","      <td>0.002151</td>\n","      <td>-69.187583</td>\n","      <td>4.232571</td>\n","      <td>-0.225267</td>\n","      <td>-0.196872</td>\n","      <td>-0.072449</td>\n","      <td>-0.103732</td>\n","      <td>-0.720746</td>\n","      <td>-5.412436</td>\n","      <td>76.679042</td>\n","      <td>-0.632727</td>\n","      <td>1351.632859</td>\n","      <td>-0.657095</td>\n","      <td>-0.434947</td>\n","      <td>-108.775969</td>\n","      <td>0.084856</td>\n","      <td>10.210182</td>\n","      <td>11.295155</td>\n","      <td>29.984624</td>\n","      <td>10.091721</td>\n","      <td>0.053279</td>\n","      <td>-4.936434</td>\n","      <td>-24.590146</td>\n","      <td>18.515436</td>\n","      <td>3.473400</td>\n","      <td>0.033444</td>\n","      <td>0.953219</td>\n","      <td>0.006076</td>\n","      <td>0</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>5/1/99 0:02</td>\n","      <td>0</td>\n","      <td>0.475720</td>\n","      <td>-4.542502</td>\n","      <td>-4.018359</td>\n","      <td>16.230658</td>\n","      <td>-0.128733</td>\n","      <td>-18.758079</td>\n","      <td>0.000732</td>\n","      <td>-0.061114</td>\n","      <td>-0.059966</td>\n","      <td>-0.038189</td>\n","      <td>0.873273</td>\n","      <td>-0.014244</td>\n","      <td>-13.306135</td>\n","      <td>0.101108</td>\n","      <td>0.041447</td>\n","      <td>0.304313</td>\n","      <td>-2.340627</td>\n","      <td>-0.939994</td>\n","      <td>3.075199</td>\n","      <td>0.123154</td>\n","      <td>-0.104334</td>\n","      <td>-0.574861</td>\n","      <td>-9.784456</td>\n","      <td>0.360160</td>\n","      <td>16.491684</td>\n","      <td>-0.450451</td>\n","      <td>-0.092430</td>\n","      <td>96</td>\n","      <td>-134.480194</td>\n","      <td>0.058759</td>\n","      <td>-0.004579</td>\n","      <td>9.350215</td>\n","      <td>0.002149</td>\n","      <td>-68.585197</td>\n","      <td>4.311490</td>\n","      <td>-0.225267</td>\n","      <td>-0.196872</td>\n","      <td>-0.059103</td>\n","      <td>-0.083895</td>\n","      <td>-0.720746</td>\n","      <td>-8.343222</td>\n","      <td>78.181598</td>\n","      <td>-0.632727</td>\n","      <td>1370.378949</td>\n","      <td>-0.875628</td>\n","      <td>-1.125819</td>\n","      <td>-108.848967</td>\n","      <td>0.085146</td>\n","      <td>12.534340</td>\n","      <td>11.290761</td>\n","      <td>29.984624</td>\n","      <td>10.095871</td>\n","      <td>0.062801</td>\n","      <td>-4.937179</td>\n","      <td>-32.413266</td>\n","      <td>22.760065</td>\n","      <td>2.682933</td>\n","      <td>0.033536</td>\n","      <td>1.090502</td>\n","      <td>0.006083</td>\n","      <td>0</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>5/1/99 0:04</td>\n","      <td>0</td>\n","      <td>0.363848</td>\n","      <td>-4.681394</td>\n","      <td>-4.353147</td>\n","      <td>14.127997</td>\n","      <td>-0.138636</td>\n","      <td>-17.836632</td>\n","      <td>0.010803</td>\n","      <td>-0.061114</td>\n","      <td>-0.030057</td>\n","      <td>-0.018352</td>\n","      <td>1.004910</td>\n","      <td>0.065150</td>\n","      <td>-9.619596</td>\n","      <td>0.101148</td>\n","      <td>0.041095</td>\n","      <td>0.252839</td>\n","      <td>-2.353925</td>\n","      <td>-0.935824</td>\n","      <td>3.075199</td>\n","      <td>0.123154</td>\n","      <td>-0.104334</td>\n","      <td>-0.579013</td>\n","      <td>-9.784456</td>\n","      <td>0.364356</td>\n","      <td>15.972885</td>\n","      <td>-0.448927</td>\n","      <td>-0.097144</td>\n","      <td>96</td>\n","      <td>-133.946587</td>\n","      <td>0.058791</td>\n","      <td>-0.084658</td>\n","      <td>9.037409</td>\n","      <td>0.002148</td>\n","      <td>-67.838187</td>\n","      <td>4.809914</td>\n","      <td>-0.225267</td>\n","      <td>-0.186801</td>\n","      <td>-0.048696</td>\n","      <td>-0.073823</td>\n","      <td>-0.720746</td>\n","      <td>-1.085166</td>\n","      <td>79.684154</td>\n","      <td>-0.632727</td>\n","      <td>1368.123089</td>\n","      <td>-0.037775</td>\n","      <td>-0.519541</td>\n","      <td>-109.086577</td>\n","      <td>0.085436</td>\n","      <td>18.582893</td>\n","      <td>11.286366</td>\n","      <td>29.984624</td>\n","      <td>10.100265</td>\n","      <td>0.072322</td>\n","      <td>-4.937924</td>\n","      <td>-34.183774</td>\n","      <td>27.004663</td>\n","      <td>3.537487</td>\n","      <td>0.033629</td>\n","      <td>1.840540</td>\n","      <td>0.006090</td>\n","      <td>0</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>5/1/99 0:06</td>\n","      <td>0</td>\n","      <td>0.301590</td>\n","      <td>-4.758934</td>\n","      <td>-4.023612</td>\n","      <td>13.161566</td>\n","      <td>-0.148142</td>\n","      <td>-18.517601</td>\n","      <td>0.002075</td>\n","      <td>-0.061114</td>\n","      <td>-0.019986</td>\n","      <td>-0.008280</td>\n","      <td>0.930037</td>\n","      <td>-0.067199</td>\n","      <td>-15.196531</td>\n","      <td>0.101188</td>\n","      <td>0.040742</td>\n","      <td>0.072873</td>\n","      <td>-2.367223</td>\n","      <td>-0.931651</td>\n","      <td>3.075199</td>\n","      <td>0.123154</td>\n","      <td>-0.104334</td>\n","      <td>-0.583165</td>\n","      <td>-9.784456</td>\n","      <td>0.368556</td>\n","      <td>15.608688</td>\n","      <td>-0.447404</td>\n","      <td>-0.160073</td>\n","      <td>96</td>\n","      <td>-134.002586</td>\n","      <td>0.058824</td>\n","      <td>-0.055118</td>\n","      <td>9.020625</td>\n","      <td>0.002146</td>\n","      <td>-67.091148</td>\n","      <td>5.308343</td>\n","      <td>-0.225267</td>\n","      <td>-0.186801</td>\n","      <td>-0.047017</td>\n","      <td>-0.063752</td>\n","      <td>-0.720746</td>\n","      <td>6.172891</td>\n","      <td>81.186702</td>\n","      <td>-0.632727</td>\n","      <td>1365.691449</td>\n","      <td>-0.987410</td>\n","      <td>0.674524</td>\n","      <td>-109.562774</td>\n","      <td>0.085726</td>\n","      <td>17.719032</td>\n","      <td>11.281972</td>\n","      <td>29.984624</td>\n","      <td>10.104660</td>\n","      <td>0.081600</td>\n","      <td>-4.938669</td>\n","      <td>-35.954281</td>\n","      <td>21.672449</td>\n","      <td>3.986095</td>\n","      <td>0.033721</td>\n","      <td>2.554880</td>\n","      <td>0.006097</td>\n","      <td>0</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>5/1/99 0:08</td>\n","      <td>0</td>\n","      <td>0.265578</td>\n","      <td>-4.749928</td>\n","      <td>-4.333150</td>\n","      <td>15.267340</td>\n","      <td>-0.155314</td>\n","      <td>-17.505913</td>\n","      <td>0.000732</td>\n","      <td>-0.061114</td>\n","      <td>-0.030057</td>\n","      <td>-0.008280</td>\n","      <td>0.828410</td>\n","      <td>-0.018472</td>\n","      <td>-14.609266</td>\n","      <td>0.101229</td>\n","      <td>0.040390</td>\n","      <td>0.171033</td>\n","      <td>-2.380521</td>\n","      <td>-0.927478</td>\n","      <td>3.075199</td>\n","      <td>0.123154</td>\n","      <td>-0.104334</td>\n","      <td>-0.587316</td>\n","      <td>-9.784456</td>\n","      <td>0.372756</td>\n","      <td>15.606125</td>\n","      <td>-0.445880</td>\n","      <td>-0.131630</td>\n","      <td>96</td>\n","      <td>-133.145706</td>\n","      <td>0.058856</td>\n","      <td>-0.153851</td>\n","      <td>9.344233</td>\n","      <td>0.002145</td>\n","      <td>-65.991813</td>\n","      <td>5.806771</td>\n","      <td>-0.225267</td>\n","      <td>-0.186801</td>\n","      <td>-0.057088</td>\n","      <td>-0.063752</td>\n","      <td>-0.720746</td>\n","      <td>-3.379599</td>\n","      <td>82.689258</td>\n","      <td>-0.632727</td>\n","      <td>1363.257859</td>\n","      <td>-0.238444</td>\n","      <td>-0.063044</td>\n","      <td>-110.038909</td>\n","      <td>0.086016</td>\n","      <td>16.855202</td>\n","      <td>11.277577</td>\n","      <td>29.984624</td>\n","      <td>10.109054</td>\n","      <td>0.091121</td>\n","      <td>-4.939414</td>\n","      <td>-37.724789</td>\n","      <td>21.907251</td>\n","      <td>3.601573</td>\n","      <td>0.033777</td>\n","      <td>1.410494</td>\n","      <td>0.006105</td>\n","      <td>0</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["      DateTime  SheetBreak  RSashScanAvg  ...  UpprHdTmpRL  VertSlcPos  EventPress\n","0  5/1/99 0:00           0      0.376665  ...     0.953219    0.006076           0\n","1  5/1/99 0:02           0      0.475720  ...     1.090502    0.006083           0\n","2  5/1/99 0:04           0      0.363848  ...     1.840540    0.006090           0\n","3  5/1/99 0:06           0      0.301590  ...     2.554880    0.006097           0\n","4  5/1/99 0:08           0      0.265578  ...     1.410494    0.006105           0\n","\n","[5 rows x 63 columns]"]},"metadata":{"tags":[]},"execution_count":5}]},{"cell_type":"code","metadata":{"id":"TpXyxgnkLZhB"},"source":["# Convert Categorical column to hot dummy columns\n","hotencoding1 = pd.get_dummies(df['Grade&Bwt'])\n","hotencoding1 = hotencoding1.add_prefix('grade_')\n","hotencoding2 = pd.get_dummies(df['EventPress'])\n","hotencoding2 = hotencoding2.add_prefix('eventpress_')\n","\n","df=df.drop(['Grade&Bwt', 'EventPress'], axis=1)\n","\n","df=pd.concat([df, hotencoding1, hotencoding2], axis=1)\n","\n","# Rename response column name for ease of understanding\n","df=df.rename(columns={'SheetBreak':'y'})\n","\n","# Shift the response column y by 2 rows to do a 4-min ahead prediction.\n","df = dp.curve_shift(df, shift_by=-2)\n","\n","# Sort by time and drop the time column.\n","df['DateTime'] = pd.to_datetime(df.DateTime)\n","df = df.sort_values(by='DateTime')\n","df = df.drop(['DateTime'], axis=1)"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"7GGllJp3vQVp"},"source":["input_X = df.loc[:, df.columns != 'y'].values  # converts df to numpy array\n","input_y = df['y'].values"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"TNcx7t7wjHHZ"},"source":["# Data Temporalize\n","lookback = 20\n","X, y = dp.temporalize(X=input_X, y=input_y, lookback=lookback)\n","\n","# Divide the data into train, valid, and test\n","X_train, X_test, y_train, y_test = train_test_split(np.array(X),\n","                                                    np.array(y),\n","                                                    test_size=DATA_SPLIT_PCT,\n","                                                    random_state=SEED)\n","X_train, X_valid, y_train, y_valid = train_test_split(X_train,\n","                                                      y_train,\n","                                                      test_size=DATA_SPLIT_PCT,\n","                                                      random_state=SEED)\n","\n","# Scaler using the training data.\n","scaler = StandardScaler().fit(dp.flatten(X_train))\n","\n","X_train_scaled = dp.scale(X_train, scaler)\n","X_valid_scaled = dp.scale(X_valid, scaler)\n","X_test_scaled = dp.scale(X_test, scaler)\n","\n","# Extract ''Normal'' samples\n","X_train_y0 = X_train[y_train==0]  # Normal state samples.\n","X_train_y1 = X_train[y_train==1]  # Anomalous samples.\n","\n","X_valid_y0 = X_valid[y_valid==0]  # Normal state samples.\n","X_valid_y1 = X_valid[y_valid==1]  # Anomalous samples.\n","\n","# Initialize a scaler using the training data.\n","scaler = StandardScaler().fit(dp.flatten(X_train_y0))\n","\n","X_train_y0_scaled = dp.scale(X_train_y0, scaler)\n","X_train_y1_scaled = dp.scale(X_train_y1, scaler)\n","X_train_scaled = dp.scale(X_train, scaler)\n","\n","X_valid_scaled = dp.scale(X_valid, scaler)\n","X_valid_y0_scaled = dp.scale(X_valid_y0, scaler)\n","\n","X_test_scaled = dp.scale(X_test, scaler)\n","\n","# Axes lengths\n","TIMESTEPS = X_train_scaled.shape[1]\n","N_FEATURES = X_train_scaled.shape[2]"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"oiGfTjSegb9_","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1606005155510,"user_tz":300,"elapsed":5495,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"1171650e-aa9f-4df6-84d6-f725f8c88bd8"},"source":["X_train_scaled.shape"],"execution_count":null,"outputs":[{"output_type":"execute_result","data":{"text/plain":["(12992, 20, 69)"]},"metadata":{"tags":[]},"execution_count":9}]},{"cell_type":"markdown","metadata":{"id":"xEuXNTXdlnND"},"source":["# Sparse Convolutional Autoencoder"]},{"cell_type":"code","metadata":{"id":"JRy3rgWpowiU","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1606005162058,"user_tz":300,"elapsed":667,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"529f4271-1e96-446f-a6b0-bf931147d8a6"},"source":["inputs = Input(shape=(TIMESTEPS, \n","                      N_FEATURES), \n","               name='encoder_input')\n","\n","def encoder(inp):\n","  '''\n","  Encoder.\n","\n","  Input\n","  inp   A tensor of input data.\n","\n","  Process\n","  Extract the essential features of the input as\n","  its encodings by filtering it through convolutional\n","  layer(s). Pooling can also be used to further\n","  summarize the features.\n","\n","  A linearly activated dense layer is added as the\n","  final layer in encoding to perform any affine\n","  transformation required. The dense layer is not\n","  for any feature extraction. Instead, it is only\n","  to make the encoding and decoding connections\n","  simpler for training.\n","\n","  Output\n","  encoding  A tensor of encodings.\n","  '''\n","\n","  # Multiple (conv, pool) blocks can be added here\n","  conv1 = Conv1D(filters=N_FEATURES,\n","                 kernel_size=4,\n","                 activation='relu',\n","                 padding='same',\n","                 name='encoder-conv1')(inp)\n","  pool1 = MaxPool1D(pool_size=4,\n","                    strides=1,\n","                    padding='same',\n","                    name='encoder-pool1')(conv1)\n","  \n","  # The last layer in encoding\n","  encoding = Dense(units=N_FEATURES,\n","                   activation='linear',\n","                   activity_regularizer=\n","                    tf.keras.regularizers.L1(l1=0.01),\n","                   name='encoder-dense1')(pool1)\n","  \n","  return encoding\n","\n","def decoder(encoding):\n","  '''\n","  Decoder.\n","\n","  Input\n","  encoding    The encoded data.\n","\n","  Process\n","  The decoding process requires a transposed \n","  convolutional layer, a.k.a. a deconvolution \n","  layer. Decoding must not be done with a \n","  regular convolutional layer. A regular conv\n","  layer is meant to extract a downsampled \n","  feature map. Decoding, on the other hand, \n","  is reconstruction of the original data \n","  from the downsampled feature map. A \n","  regular convolutional layer would try to \n","  extract further higher level features from\n","  the encodings instead of a reconstruction.\n","  \n","  For a similar reason, pooling must not be \n","  used in a decoder. A pooling operation is \n","  for summarizing a data into a few summary\n","  statistics which is useful in tasks such \n","  as classification. The purpose of \n","  decoding is the opposite, i.e., reconstruct \n","  the original data from the summarizations. \n","  Adding pooling in a decoder makes it lose \n","  the variations in the data and,\n","  hence, a poor reconstruction.\n","\n","  If the purpose is only reconstruction, a \n","  linear activation should be used in \n","  decoding. A nonlinear activation is useful\n","  for predictive features but not for \n","  reconstruction.\n","\n","  Batch normalization helps a decoder by \n","  preventing the reconstructions\n","  from exploding.\n","\n","  Output\n","  decoding    The decoded data.\n","\n","  '''\n","\n","  convT1 = Conv1DTranspose(filters=N_FEATURES,\n","                           kernel_size=4,\n","                           activation='linear',\n","                           padding='same')(encoding)\n","\n","  decoding = BatchNormalization()(convT1)\n","\n","  decoding = Dense(units=N_FEATURES,\n","                   activation='linear')(decoding)\n","  \n","  return decoding\n","\n","autoencoder = Model(inputs=inputs, \n","                    outputs=decoder(encoder(inputs)))\n","\n","autoencoder.summary()\n","autoencoder.compile(loss='mean_squared_error', optimizer = 'adam')"],"execution_count":null,"outputs":[{"output_type":"stream","text":["Model: \"functional_1\"\n","_________________________________________________________________\n","Layer (type)                 Output Shape              Param #   \n","=================================================================\n","encoder_input (InputLayer)   [(None, 20, 69)]          0         \n","_________________________________________________________________\n","encoder-conv1 (Conv1D)       (None, 20, 69)            19113     \n","_________________________________________________________________\n","encoder-pool1 (MaxPooling1D) (None, 20, 69)            0         \n","_________________________________________________________________\n","encoder-dense1 (Dense)       (None, 20, 69)            4830      \n","_________________________________________________________________\n","conv1d_transpose (Conv1DTran (None, 20, 69)            19113     \n","_________________________________________________________________\n","batch_normalization (BatchNo (None, 20, 69)            276       \n","_________________________________________________________________\n","dense (Dense)                (None, 20, 69)            4830      \n","=================================================================\n","Total params: 48,162\n","Trainable params: 48,024\n","Non-trainable params: 138\n","_________________________________________________________________\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"RWXO02XtgVHm"},"source":["history = autoencoder.fit(x=X_train_y0_scaled, \n","                          y=X_train_y0_scaled, \n","                          epochs=100, \n","                          batch_size=128, \n","                          validation_data=(X_valid_y0_scaled, \n","                                           X_valid_y0_scaled),\n","                          verbose=1).history"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"9xT4bX6pS3d8"},"source":["autoencoder.save_weights('cnn-sparse-autoencoder.h5')"],"execution_count":null,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"eu-lSiUMluM0"},"source":["# Classifier pretrained with Convolutional Autoencoder"]},{"cell_type":"code","metadata":{"id":"TbLr8G67oxuD"},"source":["def fully_connected(encoding):\n","  conv1 = Conv1D(filters=16, \n","                 kernel_size=4,\n","                 activation='relu', \n","                 padding='valid',\n","                 name='fc-conv1')(encoding)\n","  pool1 = MaxPool1D(pool_size=4, \n","                    padding='valid',\n","                    name='fc-pool1')(conv1)\n","  flat1 = Flatten()(pool1)\n","\n","  den = Dense(units=16, \n","              activation='relu')(flat1)\n","  \n","  output = Dense(units=1, \n","                activation='sigmoid', \n","                name='output')(den)\n","  \n","  return(output)"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"c_Kj2wQRUiX4"},"source":["encoding = encoder(inputs)\n","classifier = Model(inputs=inputs, \n","                   outputs=fully_connected(encoding=encoder(inputs)))"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"XfBu0QsTU-WE","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1606005770901,"user_tz":300,"elapsed":259,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"2db8b73e-89f6-4687-e577-938ad59db817"},"source":["classifier.layers"],"execution_count":null,"outputs":[{"output_type":"execute_result","data":{"text/plain":["[<tensorflow.python.keras.engine.input_layer.InputLayer at 0x7ff9e03435f8>,\n"," <tensorflow.python.keras.layers.convolutional.Conv1D at 0x7ff9dd304b38>,\n"," <tensorflow.python.keras.layers.pooling.MaxPooling1D at 0x7ff9d521fa20>,\n"," <tensorflow.python.keras.layers.core.Dense at 0x7ff9d519db38>,\n"," <tensorflow.python.keras.layers.convolutional.Conv1D at 0x7ff9d51a2f28>,\n"," <tensorflow.python.keras.layers.pooling.MaxPooling1D at 0x7ff9d51b3fd0>,\n"," <tensorflow.python.keras.layers.core.Flatten at 0x7ff9d51b9630>,\n"," <tensorflow.python.keras.layers.core.Dense at 0x7ff9d51c1630>,\n"," <tensorflow.python.keras.layers.core.Dense at 0x7ff9d51c1ba8>]"]},"metadata":{"tags":[]},"execution_count":16}]},{"cell_type":"code","metadata":{"id":"GpRTQTZxVCdM","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1606005771694,"user_tz":300,"elapsed":222,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"f47e69bf-a8ff-437c-8f79-9dcf329ac37e"},"source":["autoencoder.layers"],"execution_count":null,"outputs":[{"output_type":"execute_result","data":{"text/plain":["[<tensorflow.python.keras.engine.input_layer.InputLayer at 0x7ff9e03435f8>,\n"," <tensorflow.python.keras.layers.convolutional.Conv1D at 0x7ff9e0e7d5c0>,\n"," <tensorflow.python.keras.layers.pooling.MaxPooling1D at 0x7ff9dcd93f60>,\n"," <tensorflow.python.keras.layers.core.Dense at 0x7ff9dcdaa278>,\n"," <tensorflow.python.keras.layers.convolutional.Conv1DTranspose at 0x7ff9dcdf59b0>,\n"," <tensorflow.python.keras.layers.normalization_v2.BatchNormalization at 0x7ff9dd49d630>,\n"," <tensorflow.python.keras.layers.core.Dense at 0x7ff9dd4a6b38>]"]},"metadata":{"tags":[]},"execution_count":17}]},{"cell_type":"markdown","metadata":{"id":"F5hFHbQ4ZPik"},"source":["## Classifier layer initialized with encoder\n","\n","In the following, the classifier layers are initialized with the weights learned in the encoder. While training the classifier, one can choose to fix the encoding weights. For this, we set the classifier layers (corresponding to the encoding layers) `trainable` parameter as `False`. Otherwise, we leave them as default `True`."]},{"cell_type":"code","metadata":{"id":"35BF9IkQZMlX"},"source":["retrain_encoding = False "],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"MtVa1gLiVGib","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1606005825652,"user_tz":300,"elapsed":358,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"a4f4c73b-5a56-407e-8218-76ad3535155d"},"source":["for classifier_layer in classifier.layers:\n","  for autoencoder_layer in autoencoder.layers:\n","    if classifier_layer.name == autoencoder_layer.name:\n","      # Set the weights of classifier same as the \n","      # corresponding autoencoder (encoder) layer\n","      classifier_layer.set_weights(autoencoder_layer.get_weights())\n","\n","      if retrain_encoding == False:\n","        classifier_layer.trainable = False\n","      print(classifier_layer.name + \n","            ' in classifier set to ' + \n","            autoencoder_layer.name + \n","            ' in the encoder' + \n","            'is trainable: ' +\n","            str(classifier_layer.trainable))"],"execution_count":null,"outputs":[{"output_type":"stream","text":["encoder_input in classifier set to encoder_input in the encoderis trainable: False\n","encoder-conv1 in classifier set to encoder-conv1 in the encoderis trainable: False\n","encoder-pool1 in classifier set to encoder-pool1 in the encoderis trainable: False\n","encoder-dense1 in classifier set to encoder-dense1 in the encoderis trainable: False\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"-KPJILI-WCmj","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1606005836060,"user_tz":300,"elapsed":252,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"36be12fa-29e3-4301-9cb8-72ebe1ef9a3b"},"source":["autoencoder.get_weights()[0][1]"],"execution_count":null,"outputs":[{"output_type":"execute_result","data":{"text/plain":["array([[-0.01426586, -0.02352436,  0.00128444, ..., -0.04561961,\n","        -0.00442693, -0.00299101],\n","       [ 0.02399521,  0.02866152,  0.00441821, ..., -0.00193002,\n","        -0.02574696,  0.05855628],\n","       [-0.0171952 , -0.0390075 , -0.01850044, ..., -0.0132401 ,\n","         0.01372061, -0.03303441],\n","       ...,\n","       [ 0.03177129,  0.03707249,  0.01379102, ..., -0.06285746,\n","        -0.04104598,  0.03818714],\n","       [ 0.03416493, -0.0173989 , -0.00705815, ...,  0.08893196,\n","         0.00943434, -0.02383002],\n","       [ 0.10410752,  0.01243877, -0.09863754, ..., -0.05653463,\n","         0.10176801,  0.01476166]], dtype=float32)"]},"metadata":{"tags":[]},"execution_count":22}]},{"cell_type":"code","metadata":{"id":"1s7g5DCFWGsT","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1606005836867,"user_tz":300,"elapsed":292,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"1e02a749-9e85-49a9-e072-eb37bbb2d386"},"source":["classifier.get_weights()[0][1]"],"execution_count":null,"outputs":[{"output_type":"execute_result","data":{"text/plain":["array([[-0.01426586, -0.02352436,  0.00128444, ..., -0.04561961,\n","        -0.00442693, -0.00299101],\n","       [ 0.02399521,  0.02866152,  0.00441821, ..., -0.00193002,\n","        -0.02574696,  0.05855628],\n","       [-0.0171952 , -0.0390075 , -0.01850044, ..., -0.0132401 ,\n","         0.01372061, -0.03303441],\n","       ...,\n","       [ 0.03177129,  0.03707249,  0.01379102, ..., -0.06285746,\n","        -0.04104598,  0.03818714],\n","       [ 0.03416493, -0.0173989 , -0.00705815, ...,  0.08893196,\n","         0.00943434, -0.02383002],\n","       [ 0.10410752,  0.01243877, -0.09863754, ..., -0.05653463,\n","         0.10176801,  0.01476166]], dtype=float32)"]},"metadata":{"tags":[]},"execution_count":23}]},{"cell_type":"code","metadata":{"id":"ZiitoHiDW6_k","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1606005853687,"user_tz":300,"elapsed":253,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"aaca5473-98ab-4f46-9a0c-663ecbc826d9"},"source":["classifier.summary()\n","classifier.compile(optimizer='adam',\n","              loss='binary_crossentropy',\n","              metrics=[\n","                  'accuracy',\n","                  tf.keras.metrics.Recall(),\n","                  pm.F1Score(),\n","                  pm.FalsePositiveRate()\n","              ])"],"execution_count":null,"outputs":[{"output_type":"stream","text":["Model: \"functional_3\"\n","_________________________________________________________________\n","Layer (type)                 Output Shape              Param #   \n","=================================================================\n","encoder_input (InputLayer)   [(None, 20, 69)]          0         \n","_________________________________________________________________\n","encoder-conv1 (Conv1D)       (None, 20, 69)            19113     \n","_________________________________________________________________\n","encoder-pool1 (MaxPooling1D) (None, 20, 69)            0         \n","_________________________________________________________________\n","encoder-dense1 (Dense)       (None, 20, 69)            4830      \n","_________________________________________________________________\n","fc-conv1 (Conv1D)            (None, 17, 16)            4432      \n","_________________________________________________________________\n","fc-pool1 (MaxPooling1D)      (None, 4, 16)             0         \n","_________________________________________________________________\n","flatten (Flatten)            (None, 64)                0         \n","_________________________________________________________________\n","dense_1 (Dense)              (None, 16)                1040      \n","_________________________________________________________________\n","output (Dense)               (None, 1)                 17        \n","=================================================================\n","Total params: 29,432\n","Trainable params: 5,489\n","Non-trainable params: 23,943\n","_________________________________________________________________\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"7XRwUKzsXKMS"},"source":["history = classifier.fit(x=X_train_scaled,\n","                    y=y_train,\n","                    batch_size=128,\n","                    epochs=100,\n","                    validation_data=(X_valid_scaled, y_valid),\n","                    verbose=1).history"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"zYO8iYthXNAR","colab":{"base_uri":"https://localhost:8080/","height":292},"executionInfo":{"status":"ok","timestamp":1603707710917,"user_tz":240,"elapsed":422,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"7183f41c-0049-427c-e151-c2b41bef214e"},"source":["plt, fig = sp.plot_metric(history, metric='loss')\n","fig.savefig('sparse_cnn_encoder_classifier_loss.pdf', bbox_inches='tight')"],"execution_count":null,"outputs":[{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"code","metadata":{"id":"XxyG3p3GX82B","colab":{"base_uri":"https://localhost:8080/","height":297},"executionInfo":{"status":"ok","timestamp":1603707713214,"user_tz":240,"elapsed":384,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"ef595b39-ebc1-4ac3-c3f2-d9dec68eb264"},"source":["plt, fig = sp.plot_metric(history, metric='f1_score', ylim=[0., 1.])\n","fig.savefig('sparse_cnn_encoder_classifier_f1_score.pdf', bbox_inches='tight')"],"execution_count":null,"outputs":[{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"code","metadata":{"id":"chADIcj7YF8q","colab":{"base_uri":"https://localhost:8080/","height":294},"executionInfo":{"status":"ok","timestamp":1603707716555,"user_tz":240,"elapsed":561,"user":{"displayName":"Chitta Ranjan","photoUrl":"","userId":"03244148267376825421"}},"outputId":"0ccd2c15-edae-4d4e-fb72-8de75d0595e6"},"source":["plt, fig = sp.plot_model_recall_fpr(history)\n","fig.savefig('sparse_cnn_encoder_classifier_recall_fpr.pdf', bbox_inches='tight')"],"execution_count":null,"outputs":[{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]}]}