deernet_jacs.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "l0VUDUV-ipaR"
      },
      "outputs": [],
      "source": [
        "import numpy as np \n",
        "import scipy\n",
        "from scipy import io\n",
        "import torch\n",
        "import matplotlib.pyplot as plt\n",
        "import os\n",
        "import tqdm\n",
        "os.chdir('/content/drive/MyDrive/Colab Notebooks/deernet_data/deernet_sigmoid/sigmoid_18nov')\n",
        "from torch.autograd.functional import jacobian"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install onnx2torch\n",
        "import onnx2torch"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "A9MmPa7YK-wK",
        "outputId": "de3c6489-afc0-432d-d74a-c57c900ee993"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Collecting onnx2torch\n",
            "  Downloading onnx2torch-1.5.4-py3-none-any.whl (111 kB)\n",
            "\u001b[K     |████████████████████████████████| 111 kB 28.5 MB/s \n",
            "\u001b[?25hCollecting onnx>=1.9.0\n",
            "  Downloading onnx-1.12.0-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (13.1 MB)\n",
            "\u001b[K     |████████████████████████████████| 13.1 MB 48.3 MB/s \n",
            "\u001b[?25hRequirement already satisfied: torchvision>=0.9.0 in /usr/local/lib/python3.7/dist-packages (from onnx2torch) (0.13.1+cu113)\n",
            "Requirement already satisfied: torch>=1.8.0 in /usr/local/lib/python3.7/dist-packages (from onnx2torch) (1.12.1+cu113)\n",
            "Requirement already satisfied: numpy>=1.16.4 in /usr/local/lib/python3.7/dist-packages (from onnx2torch) (1.21.6)\n",
            "Requirement already satisfied: typing-extensions>=3.6.2.1 in /usr/local/lib/python3.7/dist-packages (from onnx>=1.9.0->onnx2torch) (4.1.1)\n",
            "Requirement already satisfied: protobuf<=3.20.1,>=3.12.2 in /usr/local/lib/python3.7/dist-packages (from onnx>=1.9.0->onnx2torch) (3.19.6)\n",
            "Requirement already satisfied: pillow!=8.3.*,>=5.3.0 in /usr/local/lib/python3.7/dist-packages (from torchvision>=0.9.0->onnx2torch) (7.1.2)\n",
            "Requirement already satisfied: requests in /usr/local/lib/python3.7/dist-packages (from torchvision>=0.9.0->onnx2torch) (2.23.0)\n",
            "Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.7/dist-packages (from requests->torchvision>=0.9.0->onnx2torch) (1.24.3)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.7/dist-packages (from requests->torchvision>=0.9.0->onnx2torch) (2022.9.24)\n",
            "Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests->torchvision>=0.9.0->onnx2torch) (3.0.4)\n",
            "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests->torchvision>=0.9.0->onnx2torch) (2.10)\n",
            "Installing collected packages: onnx, onnx2torch\n",
            "Successfully installed onnx-1.12.0 onnx2torch-1.5.4\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "x = io.loadmat('x.mat')\n",
        "x = torch.tensor(x['x'].T)"
      ],
      "metadata": {
        "id": "0z5dBqJmt6qJ"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# load network \n",
        "net = onnx2torch.convert('deernet_3.onnx')\n",
        "def y(x): return net(x) \n",
        "\n",
        "# compute jacobian and base point\n",
        "J = jacobian(y,x)\n",
        "J = J[0,:,0,:]\n",
        "f_0 = y(x) \n",
        "\n",
        "# save as mat file for descrambling \n",
        "io.savemat('layer_3.mat', {'J': J.numpy(), 'f0': f_0.numpy().reshape(256,)})"
      ],
      "metadata": {
        "id": "g-AddNqnnT4h"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.imshow(J)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "id": "3b4axXSWFNwp",
        "outputId": "f81d729f-5109-46cc-eabc-b4d0cf47f5a8"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fa9cc0f6f10>"
            ]
          },
          "metadata": {},
          "execution_count": 9
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}