Part3_RandomLabels.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "increasing-hands",
   "metadata": {},
   "outputs": [],
   "source": [
    "# This program attempts to fit a DNN to the MNIST dataset with random target labels\n",
    "# Results show this does not work well\n",
    "\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import torch\n",
    "import torchvision as tv\n",
    "from torchvision import transforms, datasets\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('seaborn-whitegrid')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "compatible-plenty",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Grab MNIST Dataset\n",
    "trainingSet = datasets.MNIST('', train=True, download=False, transform=transforms.Compose([transforms.ToTensor()]))\n",
    "testingSet = datasets.MNIST('', train=False, download=False, transform=transforms.Compose([transforms.ToTensor()]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "identified-officer",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Randomly shuffle labels\n",
    "randomLabelsTrain = torch.tensor(np.random.randint(0,10, (len(trainingSet)),))\n",
    "trainingSet.targets = randomLabelsTrain\n",
    "randomLabelsTest = torch.tensor(np.random.randint(0,10, (len(testingSet)),))\n",
    "testingSet.targets = randomLabelsTest\n",
    "\n",
    "# Prep data for training and testing\n",
    "train = torch.utils.data.DataLoader(trainingSet, batch_size=50, shuffle=True)\n",
    "test = torch.utils.data.DataLoader(testingSet, batch_size=50, shuffle=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "middle-tennis",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Random Labels NN for training - 2 hidden layers, 8175 parameters\n",
    "class randomNN(nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        self.fc1 = nn.Linear(784, 10)\n",
    "        self.fc2 = nn.Linear(10, 15)\n",
    "        self.fc3 = nn.Linear(15, 10)\n",
    "\n",
    "    def forward(self, val):\n",
    "        val = F.relu(self.fc1(val))\n",
    "        val = F.relu(self.fc2(val))\n",
    "        val = self.fc3(val)\n",
    "        return val"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "cardiac-andrew",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set up necessary auxilaries for neural net training\n",
    "randomNet = randomNN()\n",
    "costFunc = nn.CrossEntropyLoss()\n",
    "opt = optim.Adam(randomNet.parameters(), lr=0.001)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "african-feelings",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Train Random Label neural networks, Calculate accuracy for training and testing in each epoch\n",
    "EPOCHS = 400\n",
    "counter = 0\n",
    "counterList = []\n",
    "costList = []\n",
    "testLossList = []\n",
    "trainLossList = []\n",
    "for index in range(EPOCHS):\n",
    "    counterList.append(counter)\n",
    "    counter += 1\n",
    "    # Train model\n",
    "    for batch in train:\n",
    "        inputImages, groundTruth = batch\n",
    "        randomNet.zero_grad()\n",
    "        output = randomNet(inputImages.view(-1,784))\n",
    "        cost = costFunc(output, groundTruth)\n",
    "        cost.backward()\n",
    "        opt.step()\n",
    "    costList.append(cost)\n",
    "    \n",
    "    # Calculate cost on training data\n",
    "    trainTotal = 0\n",
    "    trainCounter = 0\n",
    "    for batch in train: \n",
    "        inputImages, groundTruth = batch\n",
    "        output = randomNet(inputImages.view(-1,784))\n",
    "        cost = costFunc(output, groundTruth)\n",
    "        trainTotal += cost\n",
    "        trainCounter += 1\n",
    "    trainTotal = trainTotal / trainCounter\n",
    "    trainLossList.append(trainTotal)\n",
    "    \n",
    "    # Calculate cost on test data\n",
    "    testTotal = 0\n",
    "    testCounter = 0\n",
    "    for batch in test: \n",
    "        inputImages, groundTruth = batch\n",
    "        output = randomNet(inputImages.view(-1,784))\n",
    "        cost = costFunc(output, groundTruth)\n",
    "        testTotal += cost\n",
    "        testCounter += 1\n",
    "    testTotal = testTotal / testCounter\n",
    "    testLossList.append(testTotal)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "above-strength",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visulaize Accuracy of nn for for MNIST dataset\n",
    "plt.plot(counterList, trainLossList, 'r', label='Train')\n",
    "plt.plot(counterList, testLossList, 'b', label='Test')\n",
    "plt.title(\"Loss with Random Labels\")\n",
    "plt.xlabel(\"EPOCHS\")\n",
    "plt.ylabel(\"Loss\")\n",
    "plt.legend(loc=\"center right\")\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}