{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# A very, very short introduction to Keras"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. The Iris example"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's start with a example of neural network for classification.  We'll use the `iris` dataset, which is a well-known dataset in statistics (first used by Fisher in 1936 in his paper _The use of multiple measurements in taxonomic problems_). The data set consists of 50 samples from each of three species of Iris (Iris setosa, Iris virginica and Iris versicolor). Four features were measured from each sample: the length and the width of the sepals and petals, in centimeters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "import seaborn\n",
    "iris = seaborn.load_dataset(\"iris\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's look at some samples:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "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>sepal_length</th>\n",
       "      <th>sepal_width</th>\n",
       "      <th>petal_length</th>\n",
       "      <th>petal_width</th>\n",
       "      <th>species</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>5.5</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1.3</td>\n",
       "      <td>0.2</td>\n",
       "      <td>setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>144</th>\n",
       "      <td>6.7</td>\n",
       "      <td>3.3</td>\n",
       "      <td>5.7</td>\n",
       "      <td>2.5</td>\n",
       "      <td>virginica</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>146</th>\n",
       "      <td>6.3</td>\n",
       "      <td>2.5</td>\n",
       "      <td>5.0</td>\n",
       "      <td>1.9</td>\n",
       "      <td>virginica</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>6.9</td>\n",
       "      <td>3.1</td>\n",
       "      <td>4.9</td>\n",
       "      <td>1.5</td>\n",
       "      <td>versicolor</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>94</th>\n",
       "      <td>5.6</td>\n",
       "      <td>2.7</td>\n",
       "      <td>4.2</td>\n",
       "      <td>1.3</td>\n",
       "      <td>versicolor</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     sepal_length  sepal_width  petal_length  petal_width     species\n",
       "36            5.5          3.5           1.3          0.2      setosa\n",
       "144           6.7          3.3           5.7          2.5   virginica\n",
       "146           6.3          2.5           5.0          1.9   virginica\n",
       "52            6.9          3.1           4.9          1.5  versicolor\n",
       "94            5.6          2.7           4.2          1.3  versicolor"
      ]
     },
     "execution_count": 103,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "iris.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The are some obvious correlations between the features:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<seaborn.axisgrid.PairGrid at 0x7fa913f906d0>"
      ]
     },
     "execution_count": 104,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 20 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "seaborn.pairplot(iris)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's build a simple neural network that classifies each sample in these 3 classes:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow.keras as keras\n",
    "\n",
    "# We start by defining the inputs to the network (in this case, the four features)\n",
    "iris_inputs = keras.layers.Input((4,), dtype=np.float32)\n",
    "\n",
    "# And the outputs (three categories)\n",
    "# Note the softmax: the three values should correspond to a distribution (summing to 1)\n",
    "output_layer = keras.layers.Dense(3, activation=\"softmax\")\n",
    "iris_outputs = output_layer(iris_inputs)\n",
    "\n",
    "model = keras.models.Model(inputs=iris_inputs, outputs=iris_outputs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And we compile the model by defining a loss function and an optimisation algorithm (in this case Stochastic Gradient Descent):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss=\"sparse_categorical_crossentropy\", optimizer=\"SGD\", metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can look at a summary of the layers. Right now it is not very interesting, but it's very useful when debugging complex architecture with many layers:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_12\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_13 (InputLayer)        [(None, 4)]               0         \n",
      "_________________________________________________________________\n",
      "dense_21 (Dense)             (None, 3)                 15        \n",
      "=================================================================\n",
      "Total params: 15\n",
      "Trainable params: 15\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note the model has 15 parameters, which make sense, since each output node will be connected to the four features (hence 4 weights), plus one bias term. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, in order to train the model, we need to split the data into training, development and test sets:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [],
   "source": [
    "#We shuffle through the data\n",
    "iris = iris.sample(frac=1)\n",
    "\n",
    "# And separate the features from the targets\n",
    "X = iris[[\"sepal_length\", \"sepal_width\", \"petal_length\", \"petal_width\"]].values\n",
    "y= iris[\"species\"].map({\"setosa\":0, \"virginica\":1, \"versicolor\":2}).values\n",
    "\n",
    "# The first 90 samples will be used for training, and the rest will be used for dev (30) and test (30)\n",
    "X_train, y_train = X[:90], y[:90]\n",
    "X_dev, y_dev = X[90:120], y[90:120]\n",
    "X_test, y_test = X[120:], y[120:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 90 samples, validate on 30 samples\n",
      "Epoch 1/5\n",
      "90/90 [==============================] - 0s 4ms/sample - loss: 4.8510 - accuracy: 0.3444 - val_loss: 4.1978 - val_accuracy: 0.3333\n",
      "Epoch 2/5\n",
      "90/90 [==============================] - 0s 250us/sample - loss: 4.2829 - accuracy: 0.3444 - val_loss: 3.7667 - val_accuracy: 0.2667\n",
      "Epoch 3/5\n",
      "90/90 [==============================] - 0s 219us/sample - loss: 3.8378 - accuracy: 0.3111 - val_loss: 3.3693 - val_accuracy: 0.2667\n",
      "Epoch 4/5\n",
      "90/90 [==============================] - 0s 231us/sample - loss: 3.4149 - accuracy: 0.3000 - val_loss: 2.9808 - val_accuracy: 0.2333\n",
      "Epoch 5/5\n",
      "90/90 [==============================] - 0s 234us/sample - loss: 2.9980 - accuracy: 0.2889 - val_loss: 2.5881 - val_accuracy: 0.2000\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7fa922b2db10>"
      ]
     },
     "execution_count": 111,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X_train, y_train, validation_data=(X_dev, y_dev), epochs=5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Not too good... However, we are dealing with an extremely small dataset, so we need to increase the learning rate and run on more epochs in order to get good results:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 90 samples, validate on 30 samples\n",
      "Epoch 1/20\n",
      "90/90 [==============================] - 0s 5ms/sample - loss: 10.6058 - accuracy: 0.2333 - val_loss: 2.1102 - val_accuracy: 0.3000\n",
      "Epoch 2/20\n",
      "90/90 [==============================] - 0s 231us/sample - loss: 3.6810 - accuracy: 0.5000 - val_loss: 7.6421 - val_accuracy: 0.3667\n",
      "Epoch 3/20\n",
      "90/90 [==============================] - 0s 226us/sample - loss: 5.7194 - accuracy: 0.4667 - val_loss: 0.3304 - val_accuracy: 0.8333\n",
      "Epoch 4/20\n",
      "90/90 [==============================] - 0s 217us/sample - loss: 1.4719 - accuracy: 0.7222 - val_loss: 3.1420 - val_accuracy: 0.4667\n",
      "Epoch 5/20\n",
      "90/90 [==============================] - 0s 214us/sample - loss: 1.9114 - accuracy: 0.5667 - val_loss: 3.2710 - val_accuracy: 0.3333\n",
      "Epoch 6/20\n",
      "90/90 [==============================] - 0s 219us/sample - loss: 1.8094 - accuracy: 0.5111 - val_loss: 1.3553 - val_accuracy: 0.6667\n",
      "Epoch 7/20\n",
      "90/90 [==============================] - 0s 224us/sample - loss: 0.9768 - accuracy: 0.7444 - val_loss: 0.7533 - val_accuracy: 0.7000\n",
      "Epoch 8/20\n",
      "90/90 [==============================] - 0s 240us/sample - loss: 0.8446 - accuracy: 0.7444 - val_loss: 0.8558 - val_accuracy: 0.6667\n",
      "Epoch 9/20\n",
      "90/90 [==============================] - 0s 215us/sample - loss: 0.9971 - accuracy: 0.6556 - val_loss: 0.2984 - val_accuracy: 0.8333\n",
      "Epoch 10/20\n",
      "90/90 [==============================] - 0s 213us/sample - loss: 0.6309 - accuracy: 0.7778 - val_loss: 0.1619 - val_accuracy: 0.9333\n",
      "Epoch 11/20\n",
      "90/90 [==============================] - 0s 199us/sample - loss: 0.4631 - accuracy: 0.7778 - val_loss: 0.1946 - val_accuracy: 0.8667\n",
      "Epoch 12/20\n",
      "90/90 [==============================] - 0s 192us/sample - loss: 0.2369 - accuracy: 0.9000 - val_loss: 0.1921 - val_accuracy: 0.8667\n",
      "Epoch 13/20\n",
      "90/90 [==============================] - 0s 215us/sample - loss: 0.3118 - accuracy: 0.8556 - val_loss: 0.2723 - val_accuracy: 0.8333\n",
      "Epoch 14/20\n",
      "90/90 [==============================] - 0s 215us/sample - loss: 0.1819 - accuracy: 0.9222 - val_loss: 0.1800 - val_accuracy: 0.8667\n",
      "Epoch 15/20\n",
      "90/90 [==============================] - 0s 208us/sample - loss: 0.1699 - accuracy: 0.9444 - val_loss: 0.2678 - val_accuracy: 0.8333\n",
      "Epoch 16/20\n",
      "90/90 [==============================] - 0s 212us/sample - loss: 0.1323 - accuracy: 0.9444 - val_loss: 0.1423 - val_accuracy: 0.8667\n",
      "Epoch 17/20\n",
      "90/90 [==============================] - 0s 212us/sample - loss: 0.1511 - accuracy: 0.9222 - val_loss: 0.1471 - val_accuracy: 0.9333\n",
      "Epoch 18/20\n",
      "90/90 [==============================] - 0s 207us/sample - loss: 0.1771 - accuracy: 0.9222 - val_loss: 0.0976 - val_accuracy: 0.9667\n",
      "Epoch 19/20\n",
      "90/90 [==============================] - 0s 212us/sample - loss: 0.1897 - accuracy: 0.9333 - val_loss: 0.0765 - val_accuracy: 1.0000\n",
      "Epoch 20/20\n",
      "90/90 [==============================] - 0s 205us/sample - loss: 0.1390 - accuracy: 0.9333 - val_loss: 0.1970 - val_accuracy: 0.8667\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7fa9202fe910>"
      ]
     },
     "execution_count": 112,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.compile(loss=\"sparse_categorical_crossentropy\", optimizer=keras.optimizers.Adam(lr=1), \n",
    "              metrics=[\"accuracy\"])\n",
    "model.fit(X_train, y_train, epochs=20, validation_data=(X_dev, y_dev))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now compute predictions on the test set:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[4.4, 3.2, 1.3, 0.2],\n",
       "       [5. , 2. , 3.5, 1. ],\n",
       "       [5.7, 2.8, 4.5, 1.3],\n",
       "       [5.7, 4.4, 1.5, 0.4],\n",
       "       [7.7, 2.8, 6.7, 2. ]])"
      ]
     },
     "execution_count": 113,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[9.9999857e-01, 2.4542074e-13, 1.4871696e-06],\n",
       "       [4.0861592e-10, 7.2378762e-02, 9.2762125e-01],\n",
       "       [1.2890141e-12, 3.6160052e-01, 6.3839954e-01],\n",
       "       [1.0000000e+00, 7.5171858e-17, 8.4541867e-09],\n",
       "       [5.0513398e-26, 9.9818558e-01, 1.8143895e-03]], dtype=float32)"
      ]
     },
     "execution_count": 115,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predictions[:5]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And we can compute the accuracy:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test set: 0.9666666666666667\n"
     ]
    }
   ],
   "source": [
    "import sklearn.metrics\n",
    "\n",
    "best_predictions = predictions.argmax(axis=1)\n",
    "print(\"Accuracy on test set:\", sklearn.metrics.accuracy_score(y_test, best_predictions))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Much better!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The current model is a simple logistic regression.  One can also construct a more complex model with one hidden layer (this is just for illustrative purpose - statistically speaking, it doesn't make much sense given the size and dimension of our dataset)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_13\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_14 (InputLayer)        [(None, 4)]               0         \n",
      "_________________________________________________________________\n",
      "dense_22 (Dense)             (None, 10)                50        \n",
      "_________________________________________________________________\n",
      "dense_23 (Dense)             (None, 3)                 33        \n",
      "=================================================================\n",
      "Total params: 83\n",
      "Trainable params: 83\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "\n",
    "# We start by defining the inputs to the network (in this case, the four features)\n",
    "iris_inputs = keras.layers.Input((4,))\n",
    "\n",
    "# And the outputs (three categories)\n",
    "hidden_layer = keras.layers.Dense(10, activation=\"relu\")\n",
    "\n",
    "# Note the softmax: the three values should correspond to a distribution (summing to 1)\n",
    "output_layer = keras.layers.Dense(3, activation=\"softmax\")\n",
    "\n",
    "iris_outputs = output_layer(hidden_layer(iris_inputs))\n",
    "\n",
    "model = keras.models.Model(inputs=iris_inputs, outputs=iris_outputs)\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 90 samples, validate on 30 samples\n",
      "Epoch 1/10\n",
      "90/90 [==============================] - 0s 6ms/sample - loss: 1.3990 - accuracy: 0.3222 - val_loss: 1.0873 - val_accuracy: 0.3000\n",
      "Epoch 2/10\n",
      "90/90 [==============================] - 0s 218us/sample - loss: 1.1480 - accuracy: 0.3667 - val_loss: 1.0590 - val_accuracy: 0.3667\n",
      "Epoch 3/10\n",
      "90/90 [==============================] - 0s 205us/sample - loss: 1.0134 - accuracy: 0.4000 - val_loss: 0.8857 - val_accuracy: 0.7000\n",
      "Epoch 4/10\n",
      "90/90 [==============================] - 0s 213us/sample - loss: 0.8575 - accuracy: 0.6778 - val_loss: 0.7987 - val_accuracy: 0.7000\n",
      "Epoch 5/10\n",
      "90/90 [==============================] - 0s 217us/sample - loss: 0.7574 - accuracy: 0.6556 - val_loss: 0.6920 - val_accuracy: 0.5667\n",
      "Epoch 6/10\n",
      "90/90 [==============================] - 0s 210us/sample - loss: 0.6791 - accuracy: 0.5778 - val_loss: 0.5776 - val_accuracy: 0.7000\n",
      "Epoch 7/10\n",
      "90/90 [==============================] - 0s 213us/sample - loss: 0.5777 - accuracy: 0.7444 - val_loss: 0.5346 - val_accuracy: 0.9667\n",
      "Epoch 8/10\n",
      "90/90 [==============================] - 0s 238us/sample - loss: 0.5336 - accuracy: 0.9667 - val_loss: 0.4930 - val_accuracy: 0.9333\n",
      "Epoch 9/10\n",
      "90/90 [==============================] - 0s 222us/sample - loss: 0.4886 - accuracy: 0.9778 - val_loss: 0.4559 - val_accuracy: 0.9667\n",
      "Epoch 10/10\n",
      "90/90 [==============================] - 0s 203us/sample - loss: 0.4551 - accuracy: 0.9556 - val_loss: 0.4359 - val_accuracy: 0.9000\n"
     ]
    }
   ],
   "source": [
    "model.compile(loss=\"sparse_categorical_crossentropy\", optimizer=keras.optimizers.Adam(lr=0.1), metrics=[\"accuracy\"])\n",
    "history = model.fit(X_train, y_train, epochs=10, validation_data=(X_dev, y_dev))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can inspect the history of the training steps:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.plot(history.history['accuracy'])\n",
    "plt.plot(history.history['val_accuracy'])\n",
    "plt.title('Model accuracy')\n",
    "plt.ylabel('accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'val'], loc='upper left')\n",
    "plt.show()"
   ]
  }
 ],
 "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": "3.7.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}