Mark 1.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Research Project - Brain Inspired Computing Fall 2020"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Goal with this research project is to develop a spiking binary classifier for detecting between residential and industrial labeled satellite images.\n",
    "\n",
    "Our model uses unsupervised learning, more specifically creates a decoder from the training data then uses the decoder to decode the testing data.\n",
    "\n",
    "Google Colab Link: https://colab.research.google.com/drive/1EJoDJLQKNqixMP4WTeH14mGiwUrnpR0t#scrollTo=74kKJEmJ43eS\n",
    "\n",
    "Link to the pitch talk which highlights our idea in more detail: https://docs.google.com/presentation/d/1TsljP_CYUNU8F0qEeb-vXzG6FSWaU5R1PI5Xx1kBa4s/edit?usp=sharing\n",
    "\n",
    "Link to term paper: https://docs.google.com/document/d/1MOwNo2fCIBoCrqcHnFHrbyNMeSEMaRIxoSYYLZOqx7Y/edit?usp=sharing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "In /usr/local/anaconda3/envs/deeplearning/lib/python3.7/site-packages/matplotlib/mpl-data/stylelib/_classic_test.mplstyle: \n",
      "The text.latex.preview rcparam was deprecated in Matplotlib 3.3 and will be removed two minor releases later.\n",
      "In /usr/local/anaconda3/envs/deeplearning/lib/python3.7/site-packages/matplotlib/mpl-data/stylelib/_classic_test.mplstyle: \n",
      "The mathtext.fallback_to_cm rcparam was deprecated in Matplotlib 3.3 and will be removed two minor releases later.\n",
      "In /usr/local/anaconda3/envs/deeplearning/lib/python3.7/site-packages/matplotlib/mpl-data/stylelib/_classic_test.mplstyle: Support for setting the 'mathtext.fallback_to_cm' rcParam is deprecated since 3.3 and will be removed two minor releases later; use 'mathtext.fallback : 'cm' instead.\n",
      "In /usr/local/anaconda3/envs/deeplearning/lib/python3.7/site-packages/matplotlib/mpl-data/stylelib/_classic_test.mplstyle: \n",
      "The validate_bool_maybe_none function was deprecated in Matplotlib 3.3 and will be removed two minor releases later.\n",
      "In /usr/local/anaconda3/envs/deeplearning/lib/python3.7/site-packages/matplotlib/mpl-data/stylelib/_classic_test.mplstyle: \n",
      "The savefig.jpeg_quality rcparam was deprecated in Matplotlib 3.3 and will be removed two minor releases later.\n",
      "In /usr/local/anaconda3/envs/deeplearning/lib/python3.7/site-packages/matplotlib/mpl-data/stylelib/_classic_test.mplstyle: \n",
      "The keymap.all_axes rcparam was deprecated in Matplotlib 3.3 and will be removed two minor releases later.\n",
      "In /usr/local/anaconda3/envs/deeplearning/lib/python3.7/site-packages/matplotlib/mpl-data/stylelib/_classic_test.mplstyle: \n",
      "The animation.avconv_path rcparam was deprecated in Matplotlib 3.3 and will be removed two minor releases later.\n",
      "In /usr/local/anaconda3/envs/deeplearning/lib/python3.7/site-packages/matplotlib/mpl-data/stylelib/_classic_test.mplstyle: \n",
      "The animation.avconv_args rcparam was deprecated in Matplotlib 3.3 and will be removed two minor releases later.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from keras_preprocessing.image import ImageDataGenerator\n",
    "import os\n",
    "import shutil\n",
    "import random\n",
    "np.random.seed(101)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Satellite Dataset \n",
    "\n",
    "Link to the EuroSat Dataset that we used for training and testing: https://github.com/phelber/EuroSAT\n",
    "\n",
    "Split up the satellite dataset into training, validating and testing.\n",
    "\n",
    "~The split ratio is: 70/15/15 (training/validating/testing)~\n",
    "\n",
    "The split ratio is: 80/20 (training/testing)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total images (Industrial): 2500\n",
      "Training (Industrial):  1750\n",
      "Validation (Industrial):  375\n",
      "Testing (Industrial):  375\n",
      "Total images (Residential): 3000\n",
      "Training (Residential):  2100\n",
      "Validation (Residential):  450\n",
      "Testing (Residential):  450\n"
     ]
    }
   ],
   "source": [
    "# Creating Train / Val / Test folders \n",
    "root_dir = 'Data'\n",
    "# classes_dir = ['/Industrial', '/Residential']\n",
    "\n",
    "val_ratio = 0.0\n",
    "test_ratio = 0.2\n",
    "\n",
    "def getData(categories): \n",
    "    \"\"\"\n",
    "    categories name of the classes in the EuroSat Dataset (i.e. categories = [\"Industrial\", \"Residential\"]) \n",
    "    the case of the categories needs to match the folder name\n",
    "    \"\"\"\n",
    "    classes_dir = list(map(lambda category:  \"/\" + category, categories))\n",
    "    \n",
    "    for cls in classes_dir:\n",
    "        for name in [\"/train\", \"/val\", \"/test\"]:\n",
    "            path_name = root_dir + name + cls\n",
    "            if not os.path.isdir(path_name):\n",
    "                os.makedirs(path_name)\n",
    "\n",
    "        # Creating partitions of the data after shuffeling\n",
    "        src = root_dir + cls # Folder to copy images from\n",
    "\n",
    "        allFileNames = os.listdir(src)\n",
    "        np.random.shuffle(allFileNames)\n",
    "        train_FileNames, val_FileNames, test_FileNames = np.split(np.array(allFileNames),\n",
    "                                                                  [int(len(allFileNames)* (1 - (val_ratio + test_ratio))), \n",
    "                                                                   int(len(allFileNames)* (1 - test_ratio))])\n",
    "\n",
    "\n",
    "        train_FileNames = [src+'/'+ name for name in train_FileNames.tolist()]\n",
    "        val_FileNames = [src+'/' + name for name in val_FileNames.tolist()]\n",
    "        test_FileNames = [src+'/' + name for name in test_FileNames.tolist()]\n",
    "\n",
    "        print(f'Total images ({cls[1:]}):', len(allFileNames))\n",
    "        print(f'Training ({cls[1:]}): ', len(train_FileNames))\n",
    "        print(f'Validation ({cls[1:]}): ', len(val_FileNames))\n",
    "        print(f'Testing ({cls[1:]}): ', len(test_FileNames))\n",
    "\n",
    "        # Copy-pasting images\n",
    "        for name in train_FileNames:\n",
    "            shutil.copy(name, root_dir +'/train' + cls)\n",
    "\n",
    "        for name in val_FileNames:\n",
    "            shutil.copy(name, root_dir +'/val' + cls)\n",
    "\n",
    "        for name in test_FileNames:\n",
    "            shutil.copy(name, root_dir +'/test' + cls)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Images are 64 x 64 (3 Channel rgb) 64x larger than the digit images in assignment 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Picture of Industrial Satellite Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7fa6fd13bf50>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "pic = plt.imread(\"Data/train/Industrial/Industrial_1.jpg\")\n",
    "# pic.shape\n",
    "plt.imshow(pic)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Picture of Residential Satellite Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7fa6fd22cc10>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "pic = plt.imread(\"Data/train/Residential/Residential_1.jpg\")\n",
    "# pic.shape\n",
    "plt.imshow(pic)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We normalize the pixel values between the values of 0 to 1, and fetch our data dividing them into the images (x_..) and their one hot encoded feature category."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 5497 images belonging to 2 classes.\n",
      "Found 3577 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "datagen = ImageDataGenerator(rescale=1./255) # Keras\n",
    "\n",
    "x_train_iterator = datagen.flow_from_directory('Data/train/', class_mode='categorical',shuffle=True,seed=101,target_size=(64,64), color_mode=\"grayscale\", batch_size=5500)\n",
    "\n",
    "x_train, one_hot_train = x_train_iterator.next()\n",
    "\n",
    "# x_val_iterator = datagen.flow_from_directory('Data/val/', class_mode='categorical',shuffle=True,seed=101,target_size=(64,64), color_mode=\"grayscale\", batch_size=2272)\n",
    "\n",
    "# x_val, one_hot_val = x_val_iterator.next()\n",
    "\n",
    "x_test_iterator = datagen.flow_from_directory('Data/test/', class_mode='categorical',shuffle=True,seed=101,target_size=(64,64), color_mode=\"grayscale\", batch_size=2730)\n",
    "\n",
    "x_test, one_hot_test = x_test_iterator.next()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "\n",
    "### N.B.:\n",
    "$V_{t+1} = V_{t} + dt * \\frac{dv}{dt}$\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training\n",
    "\n",
    "N.B.:\n",
    "\n",
    "**Encoding pixel to current**\n",
    "\n",
    "$J_M(x) = \\alpha x + J^{bias}$\n",
    "- $J_M$ is all past current input since *x*\n",
    "- $\\alpha$ is both the gain and a unit conversion factor\n",
    "- $J^{bias}$ accounts for the steady background input to the cell\n",
    "\n",
    "$J_i(x) = \\alpha_i \\langle x,e_i\\rangle + J^{bias}_i$\n",
    "- $e_i$ - random values normalised to unit length (represents the neuron's encoding or \"preferred direction vector\"\n",
    "- $\\langle\\,\\cdot,\\cdot\\rangle$ denotes the dot product between two vectors\n",
    "\n",
    "---\n",
    "\n",
    "**Decoding (w.o. Noise)**\n",
    "\n",
    "\n",
    "\n",
    "$\\hat{x} = Da$\n",
    "\n",
    "$D^T \\approx (A A^T)^{-1}AX^T$\n",
    "\n",
    "```\n",
    "\"\"\"\n",
    "Decoder Python Code\n",
    "d : Dimensionality of the value represented by a neuron population\n",
    "n : Number of neurons in a neuron population\n",
    "N : Number of samples\n",
    "\"\"\"\n",
    "A = np.array(...) # n x N array\n",
    "X = np.array(...) # d x N array\n",
    "D = np.linal.lstsq(A.T, X.T)[0].T\n",
    "```\n",
    "More Info: [Moore-Penrose Inverse](https://en.wikipedia.org/wiki/Moore%E2%80%93Penrose_inverse)\n",
    "\n",
    "**Decoding (w. Noise)**\n",
    "> Biological neural systems are subject to significant amounts of noise from various sources. Any analysis of such systems must take the effects of noise into account.\n",
    "\n",
    "The idea to introduce noise into our model is by applying regularisation factors to the decoder.\n",
    "\n",
    "$D^T \\approx (A A^T + N \\sigma^2 I)^{-1} A X^T$, where $I$ is the $n$ $x$ $n$ identity matrix\n",
    "\n",
    "```\n",
    "\"\"\"\n",
    "Decoder Python Code\n",
    "\"\"\"\n",
    "A = np.array(...) # n x N array\n",
    "X = np.array(...) # d x N array\n",
    "D = np.linal.lstsq(A @ A.T + 0.5 * N * np.square(sigma) * np.eye(n), A @ X.T)[0].T\n",
    "```\n",
    "\n",
    "---\n",
    "\n",
    "**Computer Firing Rate from Current**\n",
    "\n",
    "\n",
    "- $a(x)$ is the firing rate at time x\n",
    "- $\\tau^{ref}$ refractory period time constant\n",
    "- $\\tau^{RC} $ Membrane time constant (aka the characteristic time or relaxation time)\n",
    "- $J_{th}$ threshold current\n",
    "\n",
    "- $a(x) = \\frac{1}{\\tau^{ref} - \\tau^{RC}ln(1-\\frac{J_{th}}{J_M(x)})}$\n",
    "- assume $J_{th}$ = 1\n",
    "\n",
    "$$\n",
    "G(J) = \\begin{cases} \n",
    "      a(x) = \\frac{1}{\\tau^{ref} - \\tau^{RC}ln(1-\\frac{1}{J_M(x)})} & \\text{if }{J_M(x)} > 1 \\\\\n",
    "      0 & \\text{otherwise}\n",
    "\\end{cases}\n",
    "$$\n",
    "\n",
    "\n",
    "Resources: \n",
    "1. http://compneuro.uwaterloo.ca/courses/syde-750.html\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Constants"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_size = 4096 # 64 x 64 images\n",
    "hidden_size = 8 # number of neurons in hidden layer\n",
    "\n",
    "# first and second one are for calculating firing rate\n",
    "# third on used in the PSC Filter\n",
    "t_ref = 0.002 #2ms\n",
    "t_rc = 0.02 #20ms\n",
    "t_psc = 0.05 #50ms"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Post-synaptic Current Filter**\n",
    "\n",
    "use case is to filter the spike train from the LIF neuron\n",
    "\n",
    "$\n",
    "h(t) = \\begin{cases} \n",
    "      c^{-1} t^n e^{-\\frac{t}{\\tau}} & \\text{if }{t} \\ge 0 \\\\\n",
    "      0 & \\text{otherwise}\n",
    "\\end{cases}\n",
    "$\n",
    "*where n is a non-negative integer, and c normalizes the filter to area one to preserve energy*\n",
    "\n",
    "$c = \\int_{0}^\\infty t^n e^{-\\frac{t}{\\tau}} dt$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def PSC_filter(Tsim, dt):\n",
    "    \"\"\"\n",
    "    Temporal Decoder to decode the spike trains during LIF neuron simulation\n",
    "    \"\"\"\n",
    "    t = np.linspace(0,Tsim,int(np.round(Tsim/dt))) #evenly spaced time values\n",
    "    h = np.exp(-(t-Tsim/2)/t_psc) #t_psc is network param\n",
    "    h[0:len(h)//2] = 0 #set first half of h to zero\n",
    "    h = (1/dt)*h/np.sum(h) #divide h by timestep * sum of h\n",
    "    return h\n",
    "\n",
    "def sample_spherical(npoints): #sample a vector of dimension \"ndim\" from the unit sphere randomly\n",
    "    \"\"\"\n",
    "    Returns a ndim unit vector sample\n",
    "    \"\"\"\n",
    "    vec = np.random.randn(npoints)\n",
    "    vec /= np.linalg.norm(vec)\n",
    "    return vec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "Training \n",
    "\"\"\"\n",
    "\n",
    "alpha_vec = np.zeros(hidden_size)\n",
    "J_bias_vec = np.zeros(hidden_size)\n",
    "e_vec = np.zeros((hidden_size, input_size))\n",
    "num_input = len(x_train)\n",
    "\n",
    "def train():\n",
    "    A_train = np.zeros((num_input, hidden_size))\n",
    "    for i in range(hidden_size):\n",
    "\n",
    "        #Encoding\n",
    "        amax = np.random.uniform(100, 200)  # maximum firing rates randomly chosen (uniformly distributed between 100Hz and 200Hz at x=1)\n",
    "        xi = np.random.uniform(-0.95, 0.95) # x-intercepts uniformly distributed between -0.95 and 0.95\n",
    "\n",
    "        alpha = (1/(1-np.exp((t_ref - 1/amax)/t_rc)) - 1)/(1-xi) # slope=(y2-y1)/(x2-x1)\n",
    "        J_bias = 1 - xi * alpha # y-intercept (assume e in this case to be 1)\n",
    "\n",
    "        e = sample_spherical(input_size) # randomize weight direction (magnitude will be 1)\n",
    "\n",
    "        alpha_vec[i] = alpha #set gain factor\n",
    "        J_bias_vec[i] = J_bias # set residual state\n",
    "        e_vec[i] = e # set weight orientation\n",
    "\n",
    "        a_x = np.zeros(num_input)\n",
    "\n",
    "        # training\n",
    "        for j in range(num_input):\n",
    "            J_M = np.multiply(alpha, np.inner(e, x_train[j].reshape(input_size))) + J_bias\n",
    "            if J_M > 1:\n",
    "                a_x[j] = 1/(t_ref - t_rc*np.log(1- 1/J_M))\n",
    "\n",
    "        A_train[:,i] = a_x\n",
    "\n",
    "\n",
    "    #Add Noise to replicate biological phenomenal\n",
    "    mu, sigma = 0, 0.01*np.max(A_train) # mean and standard deviation\n",
    "    s = np.random.normal(mu, sigma, A_train.shape)\n",
    "    A_noisy = np.add(A_train, s) \n",
    "\n",
    "    # decoding\n",
    "    global decoder\n",
    "    decoder = np.linalg.lstsq(\n",
    "        A_train.T @ A_train + 0.5 * num_input * np.square(sigma) * np.eye(hidden_size), A_train.T @ one_hot_train\n",
    "    , rcond=None)[0]\n",
    "\n",
    "    x_hat = np.dot(A_noisy, decoder)\n",
    "#     MSE = np.mean(np.power(one_hot_train-x_hat, 2))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Testing\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**LIF Equation**\n",
    "\n",
    "$ \\frac{dv}{dt} = \\frac{1}{\\tau_{rc}}(J-V)$\n",
    "\n",
    "Using Euler's Method\n",
    "\n",
    "$V_{t+1} = V_{t} + dt * \\frac{dv}{dt}$\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "Simulating LIF Neuron\n",
    "\"\"\"\n",
    "\n",
    "def simulate_neuron(Tsim, dt, t_rc, t_ref, vrest, vth, J):\n",
    "    N = int(np.round(Tsim/dt))\n",
    "    Vprev = 0\n",
    "    Jprev = 0\n",
    "    spike_train = np.zeros(N)\n",
    "    mutex = 0  \n",
    "    \n",
    "    for i in range(N):\n",
    "        if mutex == 0: #mutex indicates the refractory period\n",
    "            dvdt = 1/(t_rc) * (J[i] - Vprev)\n",
    "            V = Vprev + dt * dvdt\n",
    "            if V < vrest:\n",
    "                V = vrest\n",
    "            elif V > vth:\n",
    "                spike_train[i] = 1\n",
    "                V = vrest\n",
    "                mutex = np.round(t_ref/dt)\n",
    "            Jprev = J[i]\n",
    "            Vprev = V\n",
    "        else:\n",
    "            mutex -= 1\n",
    "    return spike_train\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "Testing\n",
    "\"\"\"\n",
    "def test():\n",
    "    num_input = len(x_test)\n",
    "\n",
    "    Tsim = 0.1\n",
    "    dt = 0.001\n",
    "    v_rest = 0\n",
    "    v_th = 1\n",
    "    Tlen = int(np.round(Tsim/dt))\n",
    "    \n",
    "    A_test = np.zeros((num_input, Tlen, hidden_size))\n",
    "    h = PSC_filter(Tsim, dt)\n",
    "    t = np.linspace(0, Tsim, Tlen)\n",
    "    for i, (alpha, J_bias, e) in enumerate(zip(alpha_vec, J_bias_vec, e_vec)):\n",
    "        a_x = np.zeros((num_input, Tlen))\n",
    "        \n",
    "        # run test on input\n",
    "        for j in range(num_input):\n",
    "            J_M = np.multiply(alpha, np.inner(e, x_test[j].reshape(input_size))) + J_bias\n",
    "            Jin = J_M*np.ones(Tlen)\n",
    "\n",
    "            spike_train = simulate_neuron(Tsim, dt, t_rc, t_ref, v_rest, v_th, Jin) #Only Jin changes on different function call\n",
    "            a_x[j] = np.convolve(spike_train, h, 'same')\n",
    "            \n",
    "        A_test[:,:,i] = a_x\n",
    "\n",
    "    \n",
    "    accuracy_history = np.zeros(Tlen)\n",
    "    for t_i in range(Tlen):\n",
    "        x_hat_test = np.dot(A_test[:,t_i,:], decoder) #decode activity of the population\n",
    "        cnt = sum(1 for i in range(num_input) if np.argmax(x_hat_test[i]) != np.argmax(one_hot_test[i]))#count error rate of model    \n",
    "        accuracy_rate = 100 - 100*cnt/num_input #compute error rate of model\n",
    "        accuracy_history[t_i] = accuracy_rate\n",
    "\n",
    "    \"\"\"\n",
    "    Accuracy over time\n",
    "    \"\"\" \n",
    "    plt.figure(figsize=(20,10))\n",
    "    plt.plot(t, accuracy_history, '.-')\n",
    "    plt.title(f'Accuracy of model in function of time')\n",
    "    plt.xlabel('Time [s]')\n",
    "    plt.ylabel('Accuracy on test set [%]')\n",
    "    plt.grid('on')\n",
    "    plt.show()\n",
    "    \n",
    "    print(f\"Testing Error: {np.round(100 - accuracy_history[-1])}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "def run(categories):\n",
    "    getData(categories)\n",
    "    train()\n",
    "    test()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Neurons | Testing Error:\n",
    "\n",
    "- 4 - 33%\n",
    "- 8 - 29%\n",
    "- 16 - 27%\n",
    "- 32 - 26%\n",
    "- 64 - 21%\n",
    "- 128 - 17%"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Final Output\n",
    "\n",
    "**Training to Classify Images as Residential or Industrial**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total images (Industrial): 2500\n",
      "Training (Industrial):  2000\n",
      "Validation (Industrial):  0\n",
      "Testing (Industrial):  500\n",
      "Total images (Residential): 3000\n",
      "Training (Residential):  2400\n",
      "Validation (Residential):  0\n",
      "Testing (Residential):  600\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Testing Error: 31.0\n"
     ]
    }
   ],
   "source": [
    "run([\"Industrial\", \"Residential\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "References:\n",
    "\n",
    "- Ali Safa. (2020, July 19). Digit Classification Using a One-Hidden-Layer Spiking Neural Network (Version V1.3). Zenodo. http://doi.org/10.5281/zenodo.3951585\n",
    "- Eliasmith, Chris, and Charles H. Anderson. Neural Engineering: Computation, Representation, and Dynamics in Neurobiological Systems. Mitt Press, 2003. \n",
    "- Eric Hunsberger, Chris Eliasmith: “Spiking Deep Networks with LIF Neurons”, 2015; <a href='http://arxiv.org/abs/1510.08829'>arXiv:1510.08829</a>.\n",
    "- Runchun Wang, Chetan Singh Thakur, Tara Julia Hamilton, Jonathan Tapson, Andre van Schaik: “A neuromorphic hardware architecture using the Neural Engineering Framework for pattern recognition”, 2015; <a href='http://arxiv.org/abs/1507.05695'>arXiv:1507.05695</a>.\n",
    "- J. Dethier, V. Gilja, P. Nuyujukian, S. A. Elassaad, K. V. Shenoy, & K. Boahen (2011). Spiking neural network decoder for brain-machine interfaces. In 2011 5th International IEEE/EMBS Conference on Neural Engineering (pp. 396-399).\n",
    "- Terrence C. Stewart. A technical overview of the neural engineering framework. Technical Report, Centre for Theoretical Neuroscience, 2012.\n",
    "- Bekolay T, Bergstra J, Hunsberger E, DeWolf T, Stewart TC, Rasmussen D, Choo X, Voelker AR and Eliasmith C (2014) Nengo: a Python tool for building large-scale functional brain models. Front. Neuroinform. 7:48. doi: 10.3389/fninf.2013.00048\n",
    "- Eurosat: A novel dataset and deep learning benchmark for land use and land cover classification. Patrick Helber, Benjamin Bischke, Andreas Dengel, Damian Borth. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2019.\n",
    "- Introducing EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classification. Patrick Helber, Benjamin Bischke, Andreas Dengel. 2018 IEEE International Geoscience and Remote Sensing Symposium, 2018.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}