GANs_in_Slanted_Land.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Generative Adversarial Networks in Slanted Land\n",
    "\n",
    "In this notebook, we build a very simple pair of GANs.\n",
    "- The dataset is a dataset of 2x2 black and white images, where the images we want to generate (the faces) are backwards diagonals (\\\\)\n",
    "- The neural networks only have one layer.\n",
    "\n",
    "* Below is the YouTube video corresponding to this notebook.\n",
    "* Made some changes to make we get Different Examples on Generation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'https://www.youtube.com/watch?v=8L11aMN5KY8'"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\"https://www.youtube.com/watch?v=8L11aMN5KY8\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### * The main objective is trying to disgtinguish between Faces and noice.\n",
    "### * Faces represent the actual data and noise is something made by the generator."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "\n",
    "<img src=\"data_v_noise.PNG\" />"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Imports\n",
    "\n",
    "import numpy as np\n",
    "from numpy import random\n",
    "from matplotlib import pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Plotting function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Drawing function\n",
    "\n",
    "def view_samples(samples, m, n):\n",
    "    fig, axes = plt.subplots(figsize=(10, 10), nrows=m, ncols=n, sharey=True, sharex=True)\n",
    "    for ax, img in zip(axes.flatten(), samples):\n",
    "        ax.xaxis.set_visible(False)\n",
    "        ax.yaxis.set_visible(False)\n",
    "        im = ax.imshow(1-img.reshape((2,2)), cmap='Greys_r')  \n",
    "    return fig, axes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Examples of faces\n",
    "\n",
    "In the video, we've defined the faces as backwards diagonals. Here are some examples."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 5 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Examples of faces\n",
    "faces = [np.array([1,0,0,1]),\n",
    "         np.array([0.9,0.1,0.2,0.8]),\n",
    "         np.array([0.9,0.2,0.1,0.8]),\n",
    "         np.array([0.8,0.1,0.2,0.9]),\n",
    "         np.array([0.8,0.2,0.1,0.9]),]\n",
    "    \n",
    "_ = view_samples(faces, 1, 5,)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Examples of noise\n",
    "The following are randomly generated images, which don't correspond to faces."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "# Examples of noisy images\n",
    "noise = [np.random.randn(2,2) for i in range(20)]\n",
    "def generate_random_image():\n",
    "    return [np.random.random(), np.random.random(), np.random.random(), np.random.random()]\n",
    "\n",
    "_ = view_samples(noise, 4,5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Building the neural networks"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### * Data is coming from the generator in the following picture.\n",
    "### * Discriminator should output 0, showing it is fake."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "\n",
    "<img src=\"loss_function.png\"/>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# The sigmoid activation function\n",
    "def sigmoid(x):\n",
    "    return np.exp(x)/(1.0+np.exp(x))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## The Discriminator\n",
    "\n",
    "<img src=\"discriminator_math.png\" />"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Discriminator():\n",
    "    def __init__(self):\n",
    "        #self.weights = np.array([0.0 for i in range(4)])\n",
    "        #self.bias = 0.0\n",
    "        self.weights = np.array([np.random.normal() for i in range(4)])\n",
    "        self.bias = np.random.normal()\n",
    "    \n",
    "    def forward(self, x):\n",
    "        # Forward pass\n",
    "        return sigmoid(np.dot(x, self.weights) + self.bias)\n",
    "    \n",
    "    def error_from_image(self, image):\n",
    "        prediction = self.forward(image)\n",
    "        # We want the prediction to be 1, so the error is -log(prediction)\n",
    "        return -np.log(prediction)\n",
    "    \n",
    "    def derivatives_from_image(self, image):\n",
    "        prediction = self.forward(image)\n",
    "        derivatives_weights = -image * (1-prediction)\n",
    "        derivative_bias = -(1-prediction)\n",
    "        return derivatives_weights, derivative_bias\n",
    "    \n",
    "    def update_from_image(self, x):\n",
    "        ders = self.derivatives_from_image(x)\n",
    "        self.weights -= learning_rate * ders[0]\n",
    "        self.bias -= learning_rate * ders[1]\n",
    "\n",
    "    def error_from_noise(self, noise):\n",
    "        prediction = self.forward(noise)\n",
    "        # We want the prediction to be 0, so the error is -log(1-prediction)\n",
    "        return -np.log(1-prediction)\n",
    "    \n",
    "    def derivatives_from_noise(self, noise):\n",
    "        prediction = self.forward(noise)\n",
    "        derivatives_weights = noise * prediction\n",
    "        derivative_bias = prediction\n",
    "        return derivatives_weights, derivative_bias\n",
    "    \n",
    "    def update_from_noise(self, noise):\n",
    "        ders = self.derivatives_from_noise(noise)\n",
    "        self.weights -= learning_rate * ders[0]\n",
    "        self.bias -= learning_rate * ders[1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## The Generator\n",
    "<img src=\"generator_math.png\" />"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Generator():\n",
    "    def __init__(self):\n",
    "        self.weights = np.array([np.random.normal() for i in range(4)])\n",
    "        self.biases = np.array([np.random.normal() for i in range(4)])\n",
    "\n",
    "    def forward(self, z):\n",
    "        # Forward pass\n",
    "        return sigmoid(z * self.weights + self.biases)\n",
    "\n",
    "    def error(self, z, discriminator):\n",
    "        x = self.forward(z)\n",
    "        # We want the prediction to be 0, so the error is -log(1-prediction)\n",
    "        y = discriminator.forward(x)\n",
    "        return -np.log(y)\n",
    "\n",
    "    def derivatives(self, z, discriminator):\n",
    "        discriminator_weights = discriminator.weights\n",
    "        discriminator_bias = discriminator.bias\n",
    "        x = self.forward(z)\n",
    "        y = discriminator.forward(x)\n",
    "        factor = -(1-y) * discriminator_weights * x *(1-x)\n",
    "        derivatives_weights = factor * z\n",
    "        derivative_bias = factor\n",
    "        return derivatives_weights, derivative_bias\n",
    "\n",
    "    def update(self, z, discriminator):\n",
    "        error_before = self.error(z, discriminator)\n",
    "        ders = self.derivatives(z, discriminator)\n",
    "        self.weights -= learning_rate * ders[0]\n",
    "        self.biases -= learning_rate * ders[1]\n",
    "        error_after = self.error(z, discriminator)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set random seed\n",
    "np.random.seed(42)\n",
    "\n",
    "# Hyperparameters\n",
    "learning_rate = 0.01\n",
    "epochs = 1000\n",
    "\n",
    "# The GAN\n",
    "D = Discriminator()\n",
    "G = Generator()\n",
    "\n",
    "# For the error plot\n",
    "errors_discriminator = []\n",
    "errors_generator = []\n",
    "\n",
    "for epoch in range(epochs):\n",
    "    \n",
    "    for face in faces:\n",
    "        \n",
    "        # Update the discriminator weights from the real face\n",
    "        D.update_from_image(face)\n",
    "    \n",
    "        # Pick a random number to generate a fake face\n",
    "        z = random.rand()\n",
    "\n",
    "        # Calculate the discriminator error\n",
    "        errors_discriminator.append(sum(D.error_from_image(face) + D.error_from_noise(z)))\n",
    "        \n",
    "        # Calculate the generator error\n",
    "        errors_generator.append(G.error(z, D))\n",
    "        \n",
    "        # Build a fake face\n",
    "        noise = G.forward(z)\n",
    "        \n",
    "        # Update the discriminator weights from the fake face\n",
    "        D.update_from_noise(noise)\n",
    "    \n",
    "        # Update the generator weights from the fake face\n",
    "        G.update(z, D)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "scrolled": true
   },
   "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"
    },
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Discriminator error function')"
      ]
     },
     "execution_count": 22,
     "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": [
    "import matplotlib.pyplot as plt\n",
    "plt.plot(errors_generator)\n",
    "plt.title(\"Generator error function\")\n",
    "plt.legend(\"gen\")\n",
    "plt.show()\n",
    "plt.plot(errors_discriminator)\n",
    "plt.legend('disc')\n",
    "plt.title(\"Discriminator error function\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Generating images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.26319725020893436\n",
      "0.6369174270026706\n",
      "0.17843617321887062\n",
      "0.5597758799567953\n",
      "0.4798341412402477\n",
      "0.14491979119084397\n",
      "[0.92263984 0.02325962 0.07499688 0.96302314]\n",
      "[0.95483948 0.01689299 0.0600202  0.94994249]\n",
      "[0.91284755 0.0250017  0.07884213 0.96549648]\n",
      "[0.94946038 0.0180488  0.06286253 0.95295984]\n",
      "[0.9432493  0.01932843 0.06594016 0.95590385]\n",
      "[0.9086749  0.02572509 0.08041179 0.96642983]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "generated_images = []\n",
    "for i in range(6):\n",
    "    z = np.random.random()\n",
    "    print(z)\n",
    "    generated_image = G.forward(z)\n",
    "    generated_images.append(generated_image)\n",
    "_ = view_samples(generated_images, 1, 6)\n",
    "for i in generated_images:\n",
    "    print(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9605805719354462\n",
      "0.9135219227993473\n",
      "0.114654219434136\n",
      "0.7336067346855246\n",
      "0.5470683230338624\n",
      "0.9862245693116257\n",
      "[0.97200088 0.01278735 0.04935908 0.93517687]\n",
      "[0.96997058 0.01331658 0.05078946 0.93755175]\n",
      "[0.90475346 0.0263958  0.08185365 0.96725168]\n",
      "[0.96081363 0.0155466  0.05662749 0.94590136]\n",
      "[0.94851797 0.0182465  0.06334263 0.95344001]\n",
      "[0.9730505  0.01250775 0.04859577 0.93384758]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 6 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "generated_images = []\n",
    "for i in range(6):\n",
    "    z = np.random.random()\n",
    "    print(z)\n",
    "    generated_image = G.forward(z)\n",
    "    generated_images.append(generated_image)\n",
    "_ = view_samples(generated_images, 1, 6)\n",
    "for i in generated_images:\n",
    "    print(i)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Studying the weights and biases of the Generator and Discriminator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generator weights [ 1.5320363  -0.87316017 -0.63905187 -0.84704047]\n",
      "Generator biases [ 2.07553934 -3.50768858 -2.34415395  3.48272382]\n"
     ]
    }
   ],
   "source": [
    "print(\"Generator weights\", G.weights)\n",
    "print(\"Generator biases\", G.biases)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Discriminator weights [ 0.81155239 -1.16922343  0.24367562  0.63312302]\n",
      "Discriminator bias -1.3137841715192473\n"
     ]
    }
   ],
   "source": [
    "print(\"Discriminator weights\", D.weights)\n",
    "print(\"Discriminator bias\", D.bias)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Generator weights [ 0.62694977 -0.97987077 -2.07543555  1.52127964]\n",
    "Generator biases [ 1.28972108 -2.58494766 -2.3415147   1.69107561]"
   ]
  },
  {
   "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": "3.8.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}