Functional API-Keras.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.layers import Input, Dense, Flatten, Conv1D, AveragePooling1D, Concatenate\n",
    "from tensorflow.keras.models import Model\n",
    "from tensorflow import keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras.models import Sequential\n",
    "\n",
    "seq_model = Sequential([\n",
    "    Conv1D(16, 5, activation='relu', input_shape=(32, 1)),\n",
    "    AveragePooling1D(3),\n",
    "    Flatten(),\n",
    "    Dense(20, activation='sigmoid')\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv1d (Conv1D)              (None, 28, 16)            96        \n",
      "_________________________________________________________________\n",
      "average_pooling1d (AveragePo (None, 9, 16)             0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 144)               0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 20)                2900      \n",
      "=================================================================\n",
      "Total params: 2,996\n",
      "Trainable params: 2,996\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "seq_model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Building this same Sequential Model with Functional API\n",
    "\n",
    "* The `Model` class which we use is the same class which Sequential inherits from.\n",
    "* This allows us to build models with different topologies, models with multiple outputs and inputs and much more."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Explicit layer to specify the inputs\n",
    "inputs = Input(shape=(32, 1))  # We provide this layer with the shape of a single data example\n",
    "h = Conv1D(16, 5, activation='relu')(inputs)  # We use the layer object `Conv1D` as a function that calls on the inputs\n",
    "h = AveragePooling1D(3)(h)\n",
    "h = Flatten()(h)\n",
    "outputs = Dense(20, activation='sigmoid')(h)\n",
    "\n",
    "# We have now defined the flow of our model.\n",
    "functional_api_model = Model(inputs=inputs, outputs=outputs)\n",
    "# Model training is the same as training a Sequential Model\n",
    "# We will be having much more flexibility in training\n",
    "# which we will see shortly"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 32, 1)]           0         \n",
      "_________________________________________________________________\n",
      "conv1d_1 (Conv1D)            (None, 28, 16)            96        \n",
      "_________________________________________________________________\n",
      "average_pooling1d_1 (Average (None, 9, 16)             0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 144)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 20)                2900      \n",
      "=================================================================\n",
      "Total params: 2,996\n",
      "Trainable params: 2,996\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "functional_api_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Deleting the models(so that we can use these names later on in the notebook)\n",
    "del functional_api_model\n",
    "del seq_model\n",
    "del inputs\n",
    "del outputs\n",
    "del h"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Models with multiple inputs and outputs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "inputs = Input(shape=(32, 1), name='inputs')\n",
    "h = Conv1D(16, 5, activation='relu')(inputs)\n",
    "h = AveragePooling1D(3)(h)\n",
    "h = Flatten()(h)\n",
    "aux_inputs = Input(shape=(12, ), name='aux_inputs')\n",
    "h = Concatenate()([h, aux_inputs])  # h.shape = (None, 156) => 144 + 12(aux_inputs)\n",
    "outputs = Dense(20, activation='sigmoid', name='outputs')(h)\n",
    "aux_outputs = Dense(1, activation='linear', name='aux_outputs')(h)\n",
    "\n",
    "model = Model(inputs=[inputs, aux_inputs], outputs=[outputs, aux_outputs])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Compiling and Fitting the model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model.compile(loss=['binary_crossentropy', 'mse'], loss_weights=[1, 0.4], metrics=['accuracy'])\n",
    "# history = model.fit([X_train, X_aux], [y_train, y_aux], validation_split=0.2, epochs=20)\n",
    "\n",
    "# # Pass a list of loss functions which are of the same order as the outputs we have created.\n",
    "# # We need to combine the losses into a single loss to do gradient descent\n",
    "# # that is the final loss = 1*binary_crossentropy + 0.4*mse\n",
    "\n",
    "# # Another way to compile and train the model is\n",
    "# # But to use this style of compiling and fitting, \n",
    "# # It is mandatory to name our input and output layers\n",
    "# model.compile(\n",
    "#     loss={'outputs': 'binary_crossentropy', 'aux_inputs': 'mse'},\n",
    "#     loss_weights={'outputs': 1, 'aux_outputs': 0.4}, metrics=['accuracy']\n",
    "# )\n",
    "# history = model.fit(\n",
    "#     {'inputs': X_train, 'aux_inputs': X_aux},\n",
    "#     {'outputs': y_train, 'aux_outputs': y_aux},\n",
    "#     validation_split=0.2, epochs=20\n",
    "# )\n",
    "\n",
    "# # This way of compiling and training is much verbose and easier to understand"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# The UCI acute inflammation dataset\n",
    "\n",
    "* This dataset contains multiple outpts.\n",
    "#### Load the acute inflammations dataset\n",
    "\n",
    "The `acute inflammations` was created by a medical expert as a data set to test the expert system, which will perform the presumptive diagnosis of two diseases of the urinary system. You can find out more about the dataset [here](https://archive.ics.uci.edu/ml/datasets/Acute+Inflammations).\n",
    "\n",
    "Attribute information:\n",
    "\n",
    "Inputs:\n",
    "- Temperature of patient : 35C-42C\n",
    "- Occurrence of nausea : yes/no\n",
    "- Lumbar pain : yes/no\n",
    "- Urine pushing (continuous need for urination) : yes/no\n",
    "- Micturition pains : yes/no\n",
    "- Burning of urethra, itch, swelling of urethra outlet : yes/no\n",
    "\n",
    "Outputs:\n",
    "- Decision 1: Inflammation of urinary bladder : yes/no - 1/0\n",
    "- Decision 2: Nephritis of renal pelvis origin : yes/no - 1/0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load the dataset\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "pd_df = pd.read_csv('diagnosis.csv')\n",
    "dataset = pd_df.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "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>Temperature</th>\n",
       "      <th>Nausea-Occurrence</th>\n",
       "      <th>Lumbar Pain</th>\n",
       "      <th>Urine Pushing</th>\n",
       "      <th>Micturition pains</th>\n",
       "      <th>Urethra Burning</th>\n",
       "      <th>Decision1-UrinaryBladderInflammation</th>\n",
       "      <th>Decision2-NephritisRenalPelvis</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>35.5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>35.9</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>35.9</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>36.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>36.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Temperature  Nausea-Occurrence  Lumbar Pain  Urine Pushing  \\\n",
       "0         35.5                  0            1              0   \n",
       "1         35.9                  0            0              1   \n",
       "2         35.9                  0            1              0   \n",
       "3         36.0                  0            0              1   \n",
       "4         36.0                  0            1              0   \n",
       "\n",
       "   Micturition pains  Urethra Burning  Decision1-UrinaryBladderInflammation  \\\n",
       "0                  0                0                                     0   \n",
       "1                  1                1                                     1   \n",
       "2                  0                0                                     0   \n",
       "3                  1                1                                     1   \n",
       "4                  0                0                                     0   \n",
       "\n",
       "   Decision2-NephritisRenalPelvis  \n",
       "0                               0  \n",
       "1                               0  \n",
       "2                               0  \n",
       "3                               0  \n",
       "4                               0  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd_df.columns = [\n",
    "    'Temperature', 'Nausea-Occurrence', 'Lumbar Pain', 'Urine Pushing', 'Micturition pains', 'Urethra Burning',\n",
    "    'Decision1-UrinaryBladderInflammation',\n",
    "    'Decision2-NephritisRenalPelvis'\n",
    "]\n",
    "pd_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "X_train, X_test, Y_train, Y_test = train_test_split(dataset[:,:6], dataset[:,6:], test_size=0.33)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "temp_train, nocc_train, lumbp_train, up_train, mict_train, bis_train = np.transpose(X_train)\n",
    "temp_test, nocc_test, lumbp_test, up_test, mict_test, bis_test = np.transpose(X_test)\n",
    "\n",
    "inflam_train, nephr_train = Y_train[:, 0], Y_train[:, 1]\n",
    "inflam_test, nephr_test = Y_test[:, 0], Y_test[:, 1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "shape_inputs = (1, )\n",
    "temperature = Input(shape=shape_inputs, name='temp')\n",
    "nausea_occurence = Input(shape=shape_inputs, name='nocc')\n",
    "lumbar_pain = Input(shape=shape_inputs, name='lumbp')\n",
    "urine_pushing = Input(shape=shape_inputs, name='up')\n",
    "micturition_pains = Input(shape=shape_inputs, name='mict')\n",
    "bis = Input(shape=shape_inputs, name='bis')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "list_inputs = [temperature, nausea_occurence, lumbar_pain, urine_pushing, \n",
    "               micturition_pains, bis]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "x = keras.layers.concatenate(list_inputs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use a logistic regression classifier for disease prediction\n",
    "inflammation_pred = Dense(1, activation='sigmoid', name='inflam')(x)\n",
    "nephritis_pred = Dense(1, activation='sigmoid', name='nephr')(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "list_outputs = [inflammation_pred, nephritis_pred]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create the model object\n",
    "\n",
    "model = tf.keras.Model(inputs=list_inputs, outputs=list_outputs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Display the multiple input/output model\n",
    "\n",
    "tf.keras.utils.plot_model(model, 'mulit_input_output_model.png', show_shapes=True)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}