4.Deconvolution.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# I. Preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import torch\n",
    "import torchvision\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "from torchvision.transforms import transforms\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from nn_interpretability.model.model_trainer import ModelTrainer\n",
    "from nn_interpretability.model.model_repository import ModelRepository\n",
    "from nn_interpretability.visualization.mnist_visualizer import MnistVisualizer\n",
    "from nn_interpretability.dataset.mnist_data_loader import MnistDataLoader\n",
    "from nn_interpretability.interpretation.deconv.deconv import Deconvolution\n",
    "from nn_interpretability.interpretation.deconv.deconv_partial_reconstruction import DeconvolutionPartialReconstruction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_name = 'model_cnn.pt'\n",
    "\n",
    "train = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "mnist_data_loader = MnistDataLoader()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "MnistVisualizer.show_dataset_examples(mnist_data_loader.trainloader)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = ModelRepository.get_general_mnist_cnn(model_name)\n",
    "\n",
    "if train:\n",
    "    criterion = nn.CrossEntropyLoss()\n",
    "    optimizer = optim.Adam(model.parameters(), lr=0.0005)\n",
    "\n",
    "    model.train()\n",
    "    ModelTrainer.train(model, criterion, optimizer, mnist_data_loader.trainloader)\n",
    "    \n",
    "    ModelRepository.save(model, model_name)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# II. Deconvolution\n",
    "\n",
    "Deconvolution is a decision-based interpretability method which aims to reconstruct the input based on the output. This is achieved through building a second network which mirrors the architecture of the model under test in a reversed order. Furthermore, the deconvolutional network reuses the weights of the model under test.\n",
    "Our implementations builds the deconvolutional network automatically upon the creation of the Deconvolution object. Currently, only the following layers are supported: `Conv2D`, `ReLU`, `Linear`, `MaxPOOL`. \n",
    "In the following, we are conduction two experiments: Full Input Reconstruction and Partial Input Reconstruction.\n",
    "\n",
    "## 1. Full Input Reconstruction\n",
    "In the Full Input Reconstruction. we are reusing all weights for all layers and we are aiming at the complete reconstruction of the input that has led to the model output."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "images = []\n",
    "img_shape = (28, 28)\n",
    "\n",
    "interpretor = Deconvolution(model, None, None)\n",
    "\n",
    "for i in range(10):\n",
    "    img = mnist_data_loader.generate_mean_image_for_class(i).expand((1, 1, 28, 28))\n",
    "    out = interpretor.interpret(img)\n",
    "    out = out.numpy().reshape(img_shape)\n",
    "    images.append(out)\n",
    "\n",
    "MnistVisualizer.display_images_for_each_class(images)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Partial Input Reconstruction\n",
    "\n",
    "In the Partial Input Reconstruction experiment, we are first choosing a convolutional layer of interest. In this layer, we choose a single filter of interest and set the remainder of them to zeros. The remainder of the network does not change. Based on this, we can see what parts of the input can be reconstructed if only this filter was available. In the following cells, we are executing this procedure for every filter in every convolutional layer in our model under test.\n",
    "\n",
    "### 2.1 Partial Input Reconstruction for the first CONV layer\n",
    "We are executing the Partial Input Reconstruction for the first convolutional layer. The first image is the reconstruction achieved by only using the first filter from the first layer. The second image is the partial reconstruction using only the second filter from the first layer and so on."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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"
    },
    {
     "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": [
    "interpretor = DeconvolutionPartialReconstruction(model, None, None, 1)\n",
    "\n",
    "img = mnist_data_loader.generate_mean_image_for_class(0).expand((1, 1, 28, 28))\n",
    "results = interpretor.interpret(img)\n",
    "\n",
    "MnistVisualizer.display_images(results[:5], 10, 10)\n",
    "MnistVisualizer.display_images(results[5:], 10, 10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2. Partial Input Reconstruction for the second input reconstruction\n",
    "\n",
    "We are executing the Partial Input Reconstruction for the second convolutional layer. The first image is the reconstruction achieved by only using the first filter from the second layer. The second image is the partial reconstruction using only the second filter from the second layer and so on."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 5 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 5 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "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": [
    "interpretor = DeconvolutionPartialReconstruction(model, None, None, 2)\n",
    "\n",
    "img = mnist_data_loader.generate_mean_image_for_class(0).expand((1, 1, 28, 28))\n",
    "results = interpretor.interpret(img)\n",
    "\n",
    "MnistVisualizer.display_images(results[:5], 10, 10)\n",
    "MnistVisualizer.display_images(results[5:10], 10, 10)\n",
    "MnistVisualizer.display_images(results[10:15], 10, 10)\n",
    "MnistVisualizer.display_images(results[15:], 10, 10)"
   ]
  }
 ],
 "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}