deep_learning_with_keras_python.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<h1><center>Deep Learning with Keras</center></h1>\n",
    "\n",
    "<center>Owen Jones | BathML | 3rd June 2018</center>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's start at the very beginning..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "np.random.seed(2018)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "OK yes, that is not the most interesting beginning. But it means that the results in this notebook are now reproducible (yay!). Good. Moving on."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## What is Keras?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Keras is a high-level neural networks API, written in Python and capable of running on top of TensorFlow, CNTK, or Theano. -- **[keras.io](https://keras.io)**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In other words, Keras makes it super easy to build neural networks. And that is exactly what we're going to do."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\jones\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\h5py\\__init__.py:34: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n",
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import keras"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Cue the mandolin"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To get the hang of the Keras syntax, we're going to start off with a really simple network on a really simple dataset. You might have come across this one before...\n",
    "\n",
    "We're going to be using a little bit of `numpy`, because Keras works with data stored in `numpy` arrays."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[5.1, 3.5, 1.4, 0.2, 0. ],\n",
       "       [4.9, 3. , 1.4, 0.2, 0. ],\n",
       "       [4.7, 3.2, 1.3, 0.2, 0. ],\n",
       "       [4.6, 3.1, 1.5, 0.2, 0. ],\n",
       "       [5. , 3.6, 1.4, 0.2, 0. ],\n",
       "       [5.4, 3.9, 1.7, 0.4, 0. ],\n",
       "       [4.6, 3.4, 1.4, 0.3, 0. ],\n",
       "       [5. , 3.4, 1.5, 0.2, 0. ],\n",
       "       [4.4, 2.9, 1.4, 0.2, 0. ],\n",
       "       [4.9, 3.1, 1.5, 0.1, 0. ],\n",
       "       [5.4, 3.7, 1.5, 0.2, 0. ],\n",
       "       [4.8, 3.4, 1.6, 0.2, 0. ],\n",
       "       [4.8, 3. , 1.4, 0.1, 0. ],\n",
       "       [4.3, 3. , 1.1, 0.1, 0. ],\n",
       "       [5.8, 4. , 1.2, 0.2, 0. ],\n",
       "       [5.7, 4.4, 1.5, 0.4, 0. ],\n",
       "       [5.4, 3.9, 1.3, 0.4, 0. ],\n",
       "       [5.1, 3.5, 1.4, 0.3, 0. ],\n",
       "       [5.7, 3.8, 1.7, 0.3, 0. ],\n",
       "       [5.1, 3.8, 1.5, 0.3, 0. ],\n",
       "       [5.4, 3.4, 1.7, 0.2, 0. ],\n",
       "       [5.1, 3.7, 1.5, 0.4, 0. ],\n",
       "       [4.6, 3.6, 1. , 0.2, 0. ],\n",
       "       [5.1, 3.3, 1.7, 0.5, 0. ],\n",
       "       [4.8, 3.4, 1.9, 0.2, 0. ],\n",
       "       [5. , 3. , 1.6, 0.2, 0. ],\n",
       "       [5. , 3.4, 1.6, 0.4, 0. ],\n",
       "       [5.2, 3.5, 1.5, 0.2, 0. ],\n",
       "       [5.2, 3.4, 1.4, 0.2, 0. ],\n",
       "       [4.7, 3.2, 1.6, 0.2, 0. ],\n",
       "       [4.8, 3.1, 1.6, 0.2, 0. ],\n",
       "       [5.4, 3.4, 1.5, 0.4, 0. ],\n",
       "       [5.2, 4.1, 1.5, 0.1, 0. ],\n",
       "       [5.5, 4.2, 1.4, 0.2, 0. ],\n",
       "       [4.9, 3.1, 1.5, 0.1, 0. ],\n",
       "       [5. , 3.2, 1.2, 0.2, 0. ],\n",
       "       [5.5, 3.5, 1.3, 0.2, 0. ],\n",
       "       [4.9, 3.1, 1.5, 0.1, 0. ],\n",
       "       [4.4, 3. , 1.3, 0.2, 0. ],\n",
       "       [5.1, 3.4, 1.5, 0.2, 0. ],\n",
       "       [5. , 3.5, 1.3, 0.3, 0. ],\n",
       "       [4.5, 2.3, 1.3, 0.3, 0. ],\n",
       "       [4.4, 3.2, 1.3, 0.2, 0. ],\n",
       "       [5. , 3.5, 1.6, 0.6, 0. ],\n",
       "       [5.1, 3.8, 1.9, 0.4, 0. ],\n",
       "       [4.8, 3. , 1.4, 0.3, 0. ],\n",
       "       [5.1, 3.8, 1.6, 0.2, 0. ],\n",
       "       [4.6, 3.2, 1.4, 0.2, 0. ],\n",
       "       [5.3, 3.7, 1.5, 0.2, 0. ],\n",
       "       [5. , 3.3, 1.4, 0.2, 0. ],\n",
       "       [7. , 3.2, 4.7, 1.4, 1. ],\n",
       "       [6.4, 3.2, 4.5, 1.5, 1. ],\n",
       "       [6.9, 3.1, 4.9, 1.5, 1. ],\n",
       "       [5.5, 2.3, 4. , 1.3, 1. ],\n",
       "       [6.5, 2.8, 4.6, 1.5, 1. ],\n",
       "       [5.7, 2.8, 4.5, 1.3, 1. ],\n",
       "       [6.3, 3.3, 4.7, 1.6, 1. ],\n",
       "       [4.9, 2.4, 3.3, 1. , 1. ],\n",
       "       [6.6, 2.9, 4.6, 1.3, 1. ],\n",
       "       [5.2, 2.7, 3.9, 1.4, 1. ],\n",
       "       [5. , 2. , 3.5, 1. , 1. ],\n",
       "       [5.9, 3. , 4.2, 1.5, 1. ],\n",
       "       [6. , 2.2, 4. , 1. , 1. ],\n",
       "       [6.1, 2.9, 4.7, 1.4, 1. ],\n",
       "       [5.6, 2.9, 3.6, 1.3, 1. ],\n",
       "       [6.7, 3.1, 4.4, 1.4, 1. ],\n",
       "       [5.6, 3. , 4.5, 1.5, 1. ],\n",
       "       [5.8, 2.7, 4.1, 1. , 1. ],\n",
       "       [6.2, 2.2, 4.5, 1.5, 1. ],\n",
       "       [5.6, 2.5, 3.9, 1.1, 1. ],\n",
       "       [5.9, 3.2, 4.8, 1.8, 1. ],\n",
       "       [6.1, 2.8, 4. , 1.3, 1. ],\n",
       "       [6.3, 2.5, 4.9, 1.5, 1. ],\n",
       "       [6.1, 2.8, 4.7, 1.2, 1. ],\n",
       "       [6.4, 2.9, 4.3, 1.3, 1. ],\n",
       "       [6.6, 3. , 4.4, 1.4, 1. ],\n",
       "       [6.8, 2.8, 4.8, 1.4, 1. ],\n",
       "       [6.7, 3. , 5. , 1.7, 1. ],\n",
       "       [6. , 2.9, 4.5, 1.5, 1. ],\n",
       "       [5.7, 2.6, 3.5, 1. , 1. ],\n",
       "       [5.5, 2.4, 3.8, 1.1, 1. ],\n",
       "       [5.5, 2.4, 3.7, 1. , 1. ],\n",
       "       [5.8, 2.7, 3.9, 1.2, 1. ],\n",
       "       [6. , 2.7, 5.1, 1.6, 1. ],\n",
       "       [5.4, 3. , 4.5, 1.5, 1. ],\n",
       "       [6. , 3.4, 4.5, 1.6, 1. ],\n",
       "       [6.7, 3.1, 4.7, 1.5, 1. ],\n",
       "       [6.3, 2.3, 4.4, 1.3, 1. ],\n",
       "       [5.6, 3. , 4.1, 1.3, 1. ],\n",
       "       [5.5, 2.5, 4. , 1.3, 1. ],\n",
       "       [5.5, 2.6, 4.4, 1.2, 1. ],\n",
       "       [6.1, 3. , 4.6, 1.4, 1. ],\n",
       "       [5.8, 2.6, 4. , 1.2, 1. ],\n",
       "       [5. , 2.3, 3.3, 1. , 1. ],\n",
       "       [5.6, 2.7, 4.2, 1.3, 1. ],\n",
       "       [5.7, 3. , 4.2, 1.2, 1. ],\n",
       "       [5.7, 2.9, 4.2, 1.3, 1. ],\n",
       "       [6.2, 2.9, 4.3, 1.3, 1. ],\n",
       "       [5.1, 2.5, 3. , 1.1, 1. ],\n",
       "       [5.7, 2.8, 4.1, 1.3, 1. ],\n",
       "       [6.3, 3.3, 6. , 2.5, 2. ],\n",
       "       [5.8, 2.7, 5.1, 1.9, 2. ],\n",
       "       [7.1, 3. , 5.9, 2.1, 2. ],\n",
       "       [6.3, 2.9, 5.6, 1.8, 2. ],\n",
       "       [6.5, 3. , 5.8, 2.2, 2. ],\n",
       "       [7.6, 3. , 6.6, 2.1, 2. ],\n",
       "       [4.9, 2.5, 4.5, 1.7, 2. ],\n",
       "       [7.3, 2.9, 6.3, 1.8, 2. ],\n",
       "       [6.7, 2.5, 5.8, 1.8, 2. ],\n",
       "       [7.2, 3.6, 6.1, 2.5, 2. ],\n",
       "       [6.5, 3.2, 5.1, 2. , 2. ],\n",
       "       [6.4, 2.7, 5.3, 1.9, 2. ],\n",
       "       [6.8, 3. , 5.5, 2.1, 2. ],\n",
       "       [5.7, 2.5, 5. , 2. , 2. ],\n",
       "       [5.8, 2.8, 5.1, 2.4, 2. ],\n",
       "       [6.4, 3.2, 5.3, 2.3, 2. ],\n",
       "       [6.5, 3. , 5.5, 1.8, 2. ],\n",
       "       [7.7, 3.8, 6.7, 2.2, 2. ],\n",
       "       [7.7, 2.6, 6.9, 2.3, 2. ],\n",
       "       [6. , 2.2, 5. , 1.5, 2. ],\n",
       "       [6.9, 3.2, 5.7, 2.3, 2. ],\n",
       "       [5.6, 2.8, 4.9, 2. , 2. ],\n",
       "       [7.7, 2.8, 6.7, 2. , 2. ],\n",
       "       [6.3, 2.7, 4.9, 1.8, 2. ],\n",
       "       [6.7, 3.3, 5.7, 2.1, 2. ],\n",
       "       [7.2, 3.2, 6. , 1.8, 2. ],\n",
       "       [6.2, 2.8, 4.8, 1.8, 2. ],\n",
       "       [6.1, 3. , 4.9, 1.8, 2. ],\n",
       "       [6.4, 2.8, 5.6, 2.1, 2. ],\n",
       "       [7.2, 3. , 5.8, 1.6, 2. ],\n",
       "       [7.4, 2.8, 6.1, 1.9, 2. ],\n",
       "       [7.9, 3.8, 6.4, 2. , 2. ],\n",
       "       [6.4, 2.8, 5.6, 2.2, 2. ],\n",
       "       [6.3, 2.8, 5.1, 1.5, 2. ],\n",
       "       [6.1, 2.6, 5.6, 1.4, 2. ],\n",
       "       [7.7, 3. , 6.1, 2.3, 2. ],\n",
       "       [6.3, 3.4, 5.6, 2.4, 2. ],\n",
       "       [6.4, 3.1, 5.5, 1.8, 2. ],\n",
       "       [6. , 3. , 4.8, 1.8, 2. ],\n",
       "       [6.9, 3.1, 5.4, 2.1, 2. ],\n",
       "       [6.7, 3.1, 5.6, 2.4, 2. ],\n",
       "       [6.9, 3.1, 5.1, 2.3, 2. ],\n",
       "       [5.8, 2.7, 5.1, 1.9, 2. ],\n",
       "       [6.8, 3.2, 5.9, 2.3, 2. ],\n",
       "       [6.7, 3.3, 5.7, 2.5, 2. ],\n",
       "       [6.7, 3. , 5.2, 2.3, 2. ],\n",
       "       [6.3, 2.5, 5. , 1.9, 2. ],\n",
       "       [6.5, 3. , 5.2, 2. , 2. ],\n",
       "       [6.2, 3.4, 5.4, 2.3, 2. ],\n",
       "       [5.9, 3. , 5.1, 1.8, 2. ]])"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "iris = np.load(\"data/iris.npy\")\n",
    "iris"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The first four columns are numeric features (plant-related measurements... don't worry too much), and the fifth column is a label corresponding to the species, which is what we're going to use as our target.\n",
    "\n",
    "First we're just going to shuffle the rows, because at the moment they're in order (notice the label in the last column); in a minute we'll be splitting the data and we want a mixture of labels in each part."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.random.shuffle(iris)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we'll separate the labels..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "iris_labels = iris[:, 4]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "... because Keras needs the labels to be \"one-hot encoded\".\n",
    "\n",
    "**One-hot encoded label:** list out all the possible labels, and mark the one which is correct.\n",
    "\n",
    "Here, our label could be 0, 1 or 2.\n",
    "\n",
    "    Is it: 0? 1? 2?\n",
    "    ---------------\n",
    "    0 =>  [1, 0, 0]\n",
    "    1 =>  [0, 1, 0]\n",
    "    2 =>  [0, 0, 1]\n",
    "\n",
    "Keras can do this for us..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [0., 1., 0.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.],\n",
       "       [1., 0., 0.],\n",
       "       [0., 0., 1.]])"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from keras.utils import to_categorical\n",
    "iris_onehot = to_categorical(iris_labels)\n",
    "iris_onehot"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can get on with building a neural net!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## A simple network"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We're going to build a \"sequential\" model. The clue's in the name - we start with an empty model, and _sequentially_ add layers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from keras.models import Sequential\n",
    "\n",
    "model = Sequential()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There are plenty of layers to choose from, and we'll see some more exciting ones later on, but for now we'll stick with dense layers.\n",
    "\n",
    "For each dense layer, we specify:\n",
    "* The number of `units`, or how many neurons we want in the layer - the final layer will need to have 3 units, because we're classifying each input as one of 3 labels\n",
    "* The `activation` function we want to use - in a fully dense network, we tend to use sigmoid activation on the middle layers and softmax on the output layer\n",
    "\n",
    "And Keras takes care of everything else for us.\n",
    "\n",
    "Well, almost. For the very first layer in any model, we need to tell Keras what \"shape\" our input will be. We ignore the first dimension (the number of observations, or \"rows\"), because it can change without consequence; but we have to specify the other dimensions in a tuple. Here, it's a size-1 tuple, specifying that we have 4 features (\"columns\").\n",
    "\n",
    "Oh, and to add layers we use... umm, `add()`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from keras.layers import Dense\n",
    "\n",
    "model.add(Dense(15, activation=\"sigmoid\", input_shape=(4,)))\n",
    "model.add(Dense(3, activation=\"softmax\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's see what we've created..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_1 (Dense)              (None, 15)                75        \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 3)                 48        \n",
      "=================================================================\n",
      "Total params: 123\n",
      "Trainable params: 123\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Looking good!\n",
    "\n",
    "There's one more thing to do before we can train our model though: now that we're happy with it, we have to lock it in by \"compiling\" it.\n",
    "\n",
    "At this point we also need to tell Keras which loss function (always categorical crossentropy for multiclass classification) and optimizer (stochastic gradient descent, or something more fancy) to use.\n",
    "\n",
    "We can also ask for a list of metrics which we would like to see reported during training. These have nothing to do with the training process. The training process tries to minimise the loss function. _Usually_ that results in an increase in accuracy. Once again - the metrics have NOTHING to do with the training process. It's just nice to see them improving as we train!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.compile(loss=\"categorical_crossentropy\", optimizer=\"sgd\", metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can fit our model! We pass in our training data (remember that the label is in the last column, and we don't want to include that here!) and our one-hot encoded labels, as well as:\n",
    "* The number of `epochs` to train for, or the total number of times that all our training data gets passed through the network\n",
    "* A \"batch size\" - the parameters in the network get updated after each `batch_size` observations have been passed through\n",
    "* A proportion of the data which we'll set aside and use to assess our model's generalised performance (because a model can usually make great predictions about the data that's been used to train it, but it might not do so well on data it hasn't seen before)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 120 samples, validate on 30 samples\n",
      "Epoch 1/50\n",
      "120/120 [==============================] - 0s 3ms/step - loss: 1.1625 - acc: 0.3083 - val_loss: 1.1106 - val_acc: 0.3333\n",
      "Epoch 2/50\n",
      "120/120 [==============================] - 0s 100us/step - loss: 1.1377 - acc: 0.3000 - val_loss: 1.1006 - val_acc: 0.3333\n",
      "Epoch 3/50\n",
      "120/120 [==============================] - 0s 109us/step - loss: 1.1167 - acc: 0.3167 - val_loss: 1.0926 - val_acc: 0.3000\n",
      "Epoch 4/50\n",
      "120/120 [==============================] - 0s 138us/step - loss: 1.1013 - acc: 0.3250 - val_loss: 1.0861 - val_acc: 0.3333\n",
      "Epoch 5/50\n",
      "120/120 [==============================] - 0s 230us/step - loss: 1.0848 - acc: 0.3750 - val_loss: 1.0803 - val_acc: 0.2667\n",
      "Epoch 6/50\n",
      "120/120 [==============================] - 0s 134us/step - loss: 1.0729 - acc: 0.4083 - val_loss: 1.0748 - val_acc: 0.2333\n",
      "Epoch 7/50\n",
      "120/120 [==============================] - 0s 175us/step - loss: 1.0629 - acc: 0.3583 - val_loss: 1.0697 - val_acc: 0.2333\n",
      "Epoch 8/50\n",
      "120/120 [==============================] - 0s 217us/step - loss: 1.0536 - acc: 0.3750 - val_loss: 1.0646 - val_acc: 0.2333\n",
      "Epoch 9/50\n",
      "120/120 [==============================] - 0s 137us/step - loss: 1.0455 - acc: 0.4000 - val_loss: 1.0600 - val_acc: 0.2333\n",
      "Epoch 10/50\n",
      "120/120 [==============================] - 0s 150us/step - loss: 1.0388 - acc: 0.3750 - val_loss: 1.0556 - val_acc: 0.2333\n",
      "Epoch 11/50\n",
      "120/120 [==============================] - 0s 134us/step - loss: 1.0319 - acc: 0.3583 - val_loss: 1.0511 - val_acc: 0.2333\n",
      "Epoch 12/50\n",
      "120/120 [==============================] - 0s 150us/step - loss: 1.0250 - acc: 0.3833 - val_loss: 1.0469 - val_acc: 0.2333\n",
      "Epoch 13/50\n",
      "120/120 [==============================] - 0s 163us/step - loss: 1.0190 - acc: 0.3750 - val_loss: 1.0429 - val_acc: 0.2333\n",
      "Epoch 14/50\n",
      "120/120 [==============================] - 0s 159us/step - loss: 1.0143 - acc: 0.3833 - val_loss: 1.0384 - val_acc: 0.2333\n",
      "Epoch 15/50\n",
      "120/120 [==============================] - 0s 175us/step - loss: 1.0093 - acc: 0.4167 - val_loss: 1.0343 - val_acc: 0.3000\n",
      "Epoch 16/50\n",
      "120/120 [==============================] - 0s 150us/step - loss: 1.0044 - acc: 0.5083 - val_loss: 1.0307 - val_acc: 0.4000\n",
      "Epoch 17/50\n",
      "120/120 [==============================] - 0s 125us/step - loss: 0.9997 - acc: 0.6583 - val_loss: 1.0271 - val_acc: 0.5333\n",
      "Epoch 18/50\n",
      "120/120 [==============================] - 0s 138us/step - loss: 0.9954 - acc: 0.7083 - val_loss: 1.0231 - val_acc: 0.5333\n",
      "Epoch 19/50\n",
      "120/120 [==============================] - 0s 142us/step - loss: 0.9900 - acc: 0.7167 - val_loss: 1.0193 - val_acc: 0.5333\n",
      "Epoch 20/50\n",
      "120/120 [==============================] - 0s 117us/step - loss: 0.9860 - acc: 0.7417 - val_loss: 1.0159 - val_acc: 0.5333\n",
      "Epoch 21/50\n",
      "120/120 [==============================] - 0s 125us/step - loss: 0.9820 - acc: 0.7250 - val_loss: 1.0122 - val_acc: 0.5667\n",
      "Epoch 22/50\n",
      "120/120 [==============================] - 0s 117us/step - loss: 0.9781 - acc: 0.7250 - val_loss: 1.0085 - val_acc: 0.5667\n",
      "Epoch 23/50\n",
      "120/120 [==============================] - 0s 100us/step - loss: 0.9744 - acc: 0.7167 - val_loss: 1.0045 - val_acc: 0.5333\n",
      "Epoch 24/50\n",
      "120/120 [==============================] - 0s 130us/step - loss: 0.9701 - acc: 0.7333 - val_loss: 1.0011 - val_acc: 0.5667\n",
      "Epoch 25/50\n",
      "120/120 [==============================] - 0s 100us/step - loss: 0.9666 - acc: 0.7250 - val_loss: 0.9970 - val_acc: 0.5333\n",
      "Epoch 26/50\n",
      "120/120 [==============================] - 0s 92us/step - loss: 0.9628 - acc: 0.7167 - val_loss: 0.9930 - val_acc: 0.5333\n",
      "Epoch 27/50\n",
      "120/120 [==============================] - 0s 92us/step - loss: 0.9582 - acc: 0.7417 - val_loss: 0.9897 - val_acc: 0.5667\n",
      "Epoch 28/50\n",
      "120/120 [==============================] - 0s 100us/step - loss: 0.9550 - acc: 0.7250 - val_loss: 0.9867 - val_acc: 0.5667\n",
      "Epoch 29/50\n",
      "120/120 [==============================] - 0s 100us/step - loss: 0.9507 - acc: 0.7333 - val_loss: 0.9831 - val_acc: 0.5667\n",
      "Epoch 30/50\n",
      "120/120 [==============================] - 0s 100us/step - loss: 0.9477 - acc: 0.7333 - val_loss: 0.9796 - val_acc: 0.5667\n",
      "Epoch 31/50\n",
      "120/120 [==============================] - 0s 96us/step - loss: 0.9443 - acc: 0.7167 - val_loss: 0.9755 - val_acc: 0.5667\n",
      "Epoch 32/50\n",
      "120/120 [==============================] - 0s 87us/step - loss: 0.9411 - acc: 0.7417 - val_loss: 0.9722 - val_acc: 0.5667\n",
      "Epoch 33/50\n",
      "120/120 [==============================] - 0s 109us/step - loss: 0.9369 - acc: 0.7417 - val_loss: 0.9689 - val_acc: 0.5667\n",
      "Epoch 34/50\n",
      "120/120 [==============================] - 0s 117us/step - loss: 0.9330 - acc: 0.7417 - val_loss: 0.9655 - val_acc: 0.5667\n",
      "Epoch 35/50\n",
      "120/120 [==============================] - 0s 117us/step - loss: 0.9310 - acc: 0.7333 - val_loss: 0.9616 - val_acc: 0.5667\n",
      "Epoch 36/50\n",
      "120/120 [==============================] - 0s 125us/step - loss: 0.9267 - acc: 0.7417 - val_loss: 0.9586 - val_acc: 0.5667\n",
      "Epoch 37/50\n",
      "120/120 [==============================] - 0s 117us/step - loss: 0.9233 - acc: 0.7333 - val_loss: 0.9545 - val_acc: 0.5667\n",
      "Epoch 38/50\n",
      "120/120 [==============================] - 0s 117us/step - loss: 0.9191 - acc: 0.7333 - val_loss: 0.9507 - val_acc: 0.5667\n",
      "Epoch 39/50\n",
      "120/120 [==============================] - 0s 109us/step - loss: 0.9166 - acc: 0.7417 - val_loss: 0.9471 - val_acc: 0.5333\n",
      "Epoch 40/50\n",
      "120/120 [==============================] - 0s 109us/step - loss: 0.9127 - acc: 0.7333 - val_loss: 0.9432 - val_acc: 0.5333\n",
      "Epoch 41/50\n",
      "120/120 [==============================] - 0s 109us/step - loss: 0.9099 - acc: 0.7667 - val_loss: 0.9403 - val_acc: 0.5667\n",
      "Epoch 42/50\n",
      "120/120 [==============================] - 0s 92us/step - loss: 0.9059 - acc: 0.7417 - val_loss: 0.9367 - val_acc: 0.5333\n",
      "Epoch 43/50\n",
      "120/120 [==============================] - 0s 100us/step - loss: 0.9022 - acc: 0.7417 - val_loss: 0.9332 - val_acc: 0.5333\n",
      "Epoch 44/50\n",
      "120/120 [==============================] - 0s 113us/step - loss: 0.8996 - acc: 0.7500 - val_loss: 0.9297 - val_acc: 0.5333\n",
      "Epoch 45/50\n",
      "120/120 [==============================] - 0s 117us/step - loss: 0.8964 - acc: 0.7500 - val_loss: 0.9266 - val_acc: 0.5333\n",
      "Epoch 46/50\n",
      "120/120 [==============================] - 0s 117us/step - loss: 0.8925 - acc: 0.7583 - val_loss: 0.9231 - val_acc: 0.5333\n",
      "Epoch 47/50\n",
      "120/120 [==============================] - 0s 100us/step - loss: 0.8897 - acc: 0.7583 - val_loss: 0.9200 - val_acc: 0.5333\n",
      "Epoch 48/50\n",
      "120/120 [==============================] - 0s 92us/step - loss: 0.8857 - acc: 0.7417 - val_loss: 0.9164 - val_acc: 0.5333\n",
      "Epoch 49/50\n",
      "120/120 [==============================] - 0s 100us/step - loss: 0.8829 - acc: 0.7500 - val_loss: 0.9132 - val_acc: 0.5333\n",
      "Epoch 50/50\n",
      "120/120 [==============================] - 0s 117us/step - loss: 0.8800 - acc: 0.7667 - val_loss: 0.9102 - val_acc: 0.5333\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x28b84570f28>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(iris[:, :4], iris_onehot, epochs=50, batch_size=20, validation_split=0.2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Notice that the loss keeps dropping, and the accuracy fluctuates but generally speaking increases for both the training and validation sets.\n",
    "\n",
    "So, that's the syntax, but let's be honest - it's a rubbish boring network and a rubbish boring dataset. Let's move on to something more interesting!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Look who's walking"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's load in the data we're going to use and check its dimensions, which is always a good thing to do.\n",
    "\n",
    "The original dataset is from [here](https://archive.ics.uci.edu/ml/datasets/Activity+Recognition+from+Single+Chest-Mounted+Accelerometer); see [walking_data_prep.ipynb](data/walking_data_prep.ipynb) for details on how this data was prepared for use here."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(6792, 260, 3)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "walking = np.load(\"data/walking_data.npy\")\n",
    "walking.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "What? Three dimensions? Rows, columns and... huh??\n",
    "\n",
    "Don't panic. We have:\n",
    "* Observations - same as usual!\n",
    "* Timesteps - each column corresponds to one timestep, and together they form a time series. In this case we have 260 timesteps, forming a time series representing 5 seconds of measurements (because the samples were made at 52Hz).\n",
    "* Channels - at each timestep there are three values, one for the acceleration measurement in each of the x, y and z directions.\n",
    "\n",
    "![](images/lines2array.png)\n",
    "\n",
    "![](images/data_array.gif)\n",
    "\n",
    "As an important first step, let's work out how to visualise this data. We'll use `matplotlib` for our plotting."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We'll plot one observation at a time. One observation is one \"row\" of our dataset, but remember that every \"row\" is 3 channels deep. We'll plot each channel individually but on the same graph..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def plot_series(series):\n",
    "    # x-channel\n",
    "    plt.plot(series[:, 0], color=\"red\")\n",
    "    # y-channel\n",
    "    plt.plot(series[:, 1], color=\"green\")\n",
    "    # z-channel\n",
    "    plt.plot(series[:, 2], color=\"blue\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAD8CAYAAAB+UHOxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x28b887eaa20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_series(walking[408, :10, :])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So now we just pass in one observation from our data (one row, all timesteps and all channels)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXYAAAD8CAYAAABjAo9vAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nBwsM/x/IEqUcV6p1DQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x28b8a86a6a0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_series(walking[100, :, :])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Great! But now we come to an interesting question... can we, intelligent humans, tell between different people's data by eye?\n",
    "\n",
    "Let's plot a few series for some different people - say, 5 series for 3 people."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![](images/three_people.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It doesn't look like this is an easy problem for us to solve, even with our big human brains. Maybe the computer will be able to give us a run for our money...\n",
    "\n",
    "\n",
    "So we have our data; now we need our labels. Each observation comes from one of 15 people so we can quickly check that we have 15 labels."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14}"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "walking_labels = np.load(\"data/walking_labels.npy\")\n",
    "set(walking_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The time has come to split our data, but we'll have to use a slightly different tactic to the `iris` example earlier, because our labels are already separated.\n",
    "\n",
    "We'll split the row indices into three sets - a 60:20:20 split into training, cross-validation and test sets."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "m = walking.shape[0]\n",
    "indices = [x for x in range(m)]\n",
    "np.random.shuffle(indices)\n",
    "\n",
    "train_indices = indices[:int(m*0.6)]\n",
    "val_indices = indices[int(m*0.6):int(m*0.8)]\n",
    "test_indices = indices[int(m*0.8):]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can use these indices to partition both our data and our labels, remembering that we need to one-hot encode the labels!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X_train = walking[train_indices, :, :]\n",
    "X_val = walking[val_indices, :, :]\n",
    "X_test = walking[test_indices, :, :]\n",
    "\n",
    "y_train = to_categorical(walking_labels[train_indices])\n",
    "y_val = to_categorical(walking_labels[val_indices])\n",
    "y_test = to_categorical(walking_labels[test_indices])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## A more interesting network"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Of course, to start with, we need a model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model = Sequential()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And we're gonna need some more layers..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from keras.layers import Conv1D, MaxPooling1D, Flatten"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Our first layer is going to be a **convolutional layer**. Instead of looking at the whole set of features (timesteps) in one go, this layer looks at a moving \"window\" of features, so it's great for identifying patterns which are repeated, or pattern which appear often but in not necessarily in the same place every time.\n",
    "\n",
    "![](images/convolution.gif)\n",
    "\n",
    "The \"dimensionality\" of the convolutional window depends on how we want to move this window, which depends on our data. Here, we have time series data, which we can move \"along\", forwards and backwards, in _one_ dimension; so we use 1D convolution. If we had images instead, we could move the window in _two_ dimensions (left-right and up-down), so it would make more sense to use 2D convolution.\n",
    "\n",
    "We need to pass some arguments into this convolutional layer:\n",
    "\n",
    "* `filters`: Like `units` in a dense layer - the number of \"features\" we want to learn, or number of patterns to try to identify.\n",
    "* `kernel_size`: The \"window\" we were just talking about is officially called a **kernel**. We use at a rolling window capturing `kernel_size` timesteps at once.\n",
    "* `strides`: How many time steps to \"roll forward\" each time we move the window. The larger we set it, the fewer snapshots of our series the kernel will see, so the fewer output neurons will be created.\n",
    "* `activation`: Just like in the dense layer earlier, except convolutional layers typically use the REctified Linear Unit activation function because it works well and is fast to train\n",
    "* `input_shape`: Remember, the first layer always needs an input shape so it knows what data it is expecting! This time, we're feeding in observations each of shape ({260 timesteps}, {3 directional acceleration channels})\n",
    "\n",
    "If we want to get fancy, we can call these arguments **hyperparameters** of our network. The **parameters** are hidden away inside the neurons in each layer, and they get updated during training. The hyperparameters are set by us, and control the shape and behaviour of the network."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.add(Conv1D(filters=30, kernel_size=40, strides=2, activation=\"relu\", input_shape=(260, 3)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, we'll add a \"subsampling\" layer. This type of layer groups neurons up based on their position, and then combines them, thereby reducing the number of neurons that pass forward into the next layer.\n",
    "\n",
    "This has two main effects:\n",
    "* With fewer neurons, the number of _parameters_ in the following layers of the network is reduced; so there's less updating to be done after each batch, and the network will be less computationally-intensive (i.e. faster!) to train.\n",
    "* We reduce the chances of learning to recognise super-specific features (and then relying on them later on for our classifications), because we're combining bits and pieces from different neurons. This means our network should generalise better to making predictions on data outside our training set.\n",
    "\n",
    "We're using \"max pooling\" as our subsampling tactic, which just takes the strongest neuron from each \"pool\" (i.e. the one with the highest activation)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.add(MaxPooling1D(pool_size=2))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "But... _why_ have we just added a convolutional layer and then a pooling layer? Why didn't we add another convolutional layer first? Why did we pair neurons up, instead of combining 3 at a time?\n",
    "\n",
    "These are all valid questions.\n",
    "\n",
    "The fun, slightly artistic part of building a deep learning model is deciding which layers to add, how big to make them, which hyperparameters to adjust, how to adjust them, when to adjust them, ...\n",
    "\n",
    "There are a few common rules-of-thumb though. The conv-pool combo we've just seen is very common - convolutional layers add a load of new parameters to the model, and then pooling layers (or other subsampling layers, such as \"dropout\") strip the new neurons down and help prevent overfitting.\n",
    "\n",
    "We're going to add another conv-pool pair of layers to our model. The neurons in the first convolutional layer will learn to respond to patterns in our time series data. The neurons in the _second_ convolutional layer will learn to respond to patterns in the (pooled) output of the _first_ layer - so it's responding to _meta-patterns_ in the original data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.add(Conv1D(filters=30, kernel_size=10, activation=\"relu\"))\n",
    "model.add(MaxPooling1D(pool_size=2))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's pause for a moment and check the shape of the data which is coming out of the last layer of the network right now."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(None, 23, 30)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.output_shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Notice anything?\n",
    "\n",
    "Yeah, it's still in 3D: some number of rows (which Keras represents with `None`, because we could feed in any number of observations!), some \"features\" (loosely related to our original timesteps, but twisted beyond recognition by the conv/pool layers), and some \"filters\" (the number which we set in the latest conv layer).\n",
    "\n",
    "But if we think about what we want to predict... that's in 2D! Some number of rows, where each one is a one-hot label.\n",
    "\n",
    "So we somehow need to _flatten_ our network by taking all those stacked-up parameters and laying them out next to each other in a big long line.\n",
    "\n",
    "This is Keras though, so it's easy to do that..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.add(Flatten())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And now if we check the shape again:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(None, 690)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.output_shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We have the same size of output as before, just reshaped!\n",
    "\n",
    "Now we can finish off by feeding this into a couple of dense layers. The first one will learn relationships between the now-flattened convolutional neurons, and the second produces the (one-hot) prediction."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "model.add(Dense(100, activation=\"sigmoid\"))\n",
    "model.add(Dense(15, activation=\"softmax\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a step back and admire our handiwork."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv1d_1 (Conv1D)            (None, 111, 30)           3630      \n",
      "_________________________________________________________________\n",
      "max_pooling1d_1 (MaxPooling1 (None, 55, 30)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_2 (Conv1D)            (None, 46, 30)            9030      \n",
      "_________________________________________________________________\n",
      "max_pooling1d_2 (MaxPooling1 (None, 23, 30)            0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 690)               0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 100)               69100     \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 15)                1515      \n",
      "=================================================================\n",
      "Total params: 83,275\n",
      "Trainable params: 83,275\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Wow. Give yourself a pat on the back. That looks pretty darned good."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training and results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Remember that before we can train our network, we have to compile it! This time is just like before, except we're going to use the faster fancier \"ADAptive Movement estimation\" optimizer instead of gradient descent."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model.compile(loss=\"categorical_crossentropy\", optimizer=\"adam\", metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And now we fit the model on our training data - again, almost identical to last time, except we'll explicitly pass in our validation data rather than taking a split from the training data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 4075 samples, validate on 1358 samples\n",
      "Epoch 1/10\n",
      "4075/4075 [==============================] - 3s 653us/step - loss: 2.4759 - acc: 0.2388 - val_loss: 1.8864 - val_acc: 0.4440\n",
      "Epoch 2/10\n",
      "4075/4075 [==============================] - 2s 516us/step - loss: 1.3639 - acc: 0.6483 - val_loss: 0.9994 - val_acc: 0.7452\n",
      "Epoch 3/10\n",
      "4075/4075 [==============================] - 2s 509us/step - loss: 0.7748 - acc: 0.8184 - val_loss: 0.6575 - val_acc: 0.8262\n",
      "Epoch 4/10\n",
      "4075/4075 [==============================] - 2s 529us/step - loss: 0.5504 - acc: 0.8643 - val_loss: 0.5141 - val_acc: 0.8770\n",
      "Epoch 5/10\n",
      "4075/4075 [==============================] - 2s 513us/step - loss: 0.4225 - acc: 0.8994 - val_loss: 0.4399 - val_acc: 0.8895\n",
      "Epoch 6/10\n",
      "4075/4075 [==============================] - 2s 546us/step - loss: 0.3442 - acc: 0.9190 - val_loss: 0.3836 - val_acc: 0.9102\n",
      "Epoch 7/10\n",
      "4075/4075 [==============================] - 2s 562us/step - loss: 0.2815 - acc: 0.9399 - val_loss: 0.3348 - val_acc: 0.9168\n",
      "Epoch 8/10\n",
      "4075/4075 [==============================] - 2s 582us/step - loss: 0.2329 - acc: 0.9521 - val_loss: 0.3034 - val_acc: 0.9197\n",
      "Epoch 9/10\n",
      "4075/4075 [==============================] - 2s 554us/step - loss: 0.1942 - acc: 0.9620 - val_loss: 0.2898 - val_acc: 0.9190\n",
      "Epoch 10/10\n",
      "4075/4075 [==============================] - 2s 545us/step - loss: 0.1602 - acc: 0.9755 - val_loss: 0.2587 - val_acc: 0.9315\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x28b887ea2e8>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X_train, y_train, epochs=10, batch_size=100, validation_data=(X_val, y_val))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Not too shabby! Look at that lovely increase in validation-set accuracy. (But remember that it's the loss function that directs the training!)\n",
    "\n",
    "At this point we could try to improve that cross-validation accuracy score by changing the network structure, messing around with other layers, adjusting hyperparameters... and then once we're happy, we can assess the final performance of our model!\n",
    "\n",
    "We'll use a couple of tools from `scikit-learn` to help us out."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import classification_report, confusion_matrix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First we can use our model to make predictions on the test set (which we haven't touched yet, because we've been saving it for exactly this purpose!)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "y_pred = model.predict_classes(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then we can see how well we've done! But remember that our test labels are one-hot encoded. We can use `np.argmax()` to convert back to integer labels: we just need to look across the 1-axis (the columns, as opposed to the rows, which are the 0-axis) and pick the index of the highest value in each row (i.e. the position of the 1 among all the 0s).\n",
    "\n",
    "Generally speaking, we want the metrics in the classification report to be close to 1, and we want high values on the diagonal of the confusion matrix (indicating that our predicted labels, in the columns, match the actual values, in the rows)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.89      0.99      0.93        94\n",
      "          1       0.93      0.88      0.91        94\n",
      "          2       0.99      0.88      0.93        75\n",
      "          3       0.99      0.94      0.96       113\n",
      "          4       0.96      0.91      0.94        58\n",
      "          5       0.95      0.98      0.97       129\n",
      "          6       0.88      0.92      0.90        65\n",
      "          7       0.85      0.78      0.82        79\n",
      "          8       0.94      0.97      0.95       104\n",
      "          9       0.84      0.86      0.85        81\n",
      "         10       1.00      1.00      1.00       101\n",
      "         11       0.89      0.81      0.85        84\n",
      "         12       0.98      0.94      0.96       109\n",
      "         13       0.85      0.96      0.90        80\n",
      "         14       0.91      0.95      0.93        93\n",
      "\n",
      "avg / total       0.93      0.93      0.93      1359\n",
      "\n",
      "[[ 93   0   0   0   0   0   0   0   0   0   0   0   0   1   0]\n",
      " [  0  83   0   0   0   0   0   2   2   0   0   0   0   7   0]\n",
      " [  0   0  66   0   1   1   0   0   0   2   0   0   0   0   5]\n",
      " [  4   0   0 106   0   3   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0  53   0   2   0   2   0   0   0   0   0   1]\n",
      " [  0   0   0   1   0 127   0   0   0   1   0   0   0   0   0]\n",
      " [  3   0   0   0   0   0  60   0   0   2   0   0   0   0   0]\n",
      " [  0   3   1   0   0   0   2  62   1   1   0   7   1   1   0]\n",
      " [  1   0   0   0   1   0   1   0 101   0   0   0   0   0   0]\n",
      " [  4   0   0   0   0   1   0   0   0  70   0   1   1   1   3]\n",
      " [  0   0   0   0   0   0   0   0   0   0 101   0   0   0   0]\n",
      " [  0   1   0   0   0   0   1   8   1   1   0  68   0   4   0]\n",
      " [  0   0   0   0   0   0   1   0   0   5   0   0 103   0   0]\n",
      " [  0   2   0   0   0   0   0   1   0   0   0   0   0  77   0]\n",
      " [  0   0   0   0   0   2   1   0   1   1   0   0   0   0  88]]\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(np.argmax(y_test, axis=1), y_pred))\n",
    "print(confusion_matrix(np.argmax(y_test, axis=1), y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Opening the black box"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We hope that during training, the neurons in the network will learn to respond to certain patterns in the data which they receive as input. A neuron will \"respond\" when its weights line up nicely with the input data - so by treating these weights as series in their own right, we can see what the input to each neuron which would produce the biggest response would look like.\n",
    "\n",
    "Extending our `plot_series()` function from earlier, we can use it to grab the weights from a particular neuron in a model, and then plot them as a time series."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def plot_filter(model, layer, k):\n",
    "    weights = model.layers[layer].get_weights()[0][:, :, k]\n",
    "    plot_series(weights)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For example, taking the 0th layer in our current model (the first convolutional layer), and looking at the fifth neuron in it:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAD8CAYAAABzTgP2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x28b8d346048>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_filter(model, 0, 4)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "There does seem to be some sort of pattern in there, but it's a little bit hard to tell.\n",
    "\n",
    "We can confirm our suspicions by producing an \"autocorrelation plot\". To calculate the autocorrelation, we take two copies of the series and line the end of one up with the start of the next. Then we slide one series over the top of the other. The closer the two series are to each other at a given point, the higher the correlation, with a perfect score of 1 at the point where the two copies are directly on top of each other. If there are repeating patterns in the series, we should see smooth curves in the autocorrelation plot, with peaks where the repeated patterns line up nicely with the original (like echos!)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def plot_filter_corr(model, layer, k):\n",
    "    weights = model.layers[layer].get_weights()[0][:, :, k]\n",
    "    corrs = np.apply_along_axis(lambda y: np.correlate(y, y, mode=\"full\"), 0, weights)\n",
    "    plot_series(corrs[corrs.shape[0]//2:, :])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAD8CAYAAABzTgP2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMARBEAQtRGEIgiAIWvgT+ys7y1miT38AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x28b8f594518>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_filter_corr(model, 0, 4)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Nice! Let's see a few filters alongside their autocorrelation plots."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "![](images/corrs.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "This suggests that the neurons in the first convolutional layer have learned to recognise repeated patterns in the original time series. That's great! It means the neurons are most probably learning to recognise features that we would expect them to learn to recognise. And that's about as good as we can get when it comes to interpreting the features in black-box models like neural networks!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Conv-clusion"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "That's all we have time for here I'm afraid, but hopefully this has shown off some of the wonderful things which Keras (and deep learning in general) is capable of! Please do go away and look at this again, and make changes, and try things out, and break things, and figure out why they broke, and experiment some more, and keep learning!\n",
    "\n",
    "If you have questions about this workshop, Keras, neural networks or ML/data science in general - or if you just want a chat - you can get in touch with me at olj23@bath.ac.uk or on Twitter at @owenjonesuob. Thanks for your attention!"
   ]
  }
 ],
 "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}