Certified Explanation-Mnist.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "b413da50",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/bhui/miniconda3/envs/xy_py38/lib/python3.8/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "from models import *\n",
    "import torchvision\n",
    "from torchvision import transforms, datasets\n",
    "from torch.distributions import normal\n",
    "import matplotlib.pyplot as plt\n",
    "import captum\n",
    "from captum.attr import DeepLift, IntegratedGradients, Saliency, NoiseTunnel, GradientShap\n",
    "from captum.attr import visualization as viz\n",
    "import argparse\n",
    "from utils import get_args\n",
    "import matplotlib\n",
    "from models.model import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "0f4867ca",
   "metadata": {},
   "outputs": [],
   "source": [
    "def attribute_image_features(algorithm, input, **kwargs):\n",
    "    global_model.zero_grad()\n",
    "    tensor_attributions = algorithm.attribute(input,\n",
    "                                              target=labels[ind],\n",
    "                                              **kwargs\n",
    "                                              )\n",
    "    return tensor_attributions\n",
    "\n",
    "def add_gaussian_noise(noise_scale, input):\n",
    "    noise = torch.randn_like(input) * noise_scale\n",
    "    return input + noise\n",
    "\n",
    "def predict_with_randomized_smoothing(noise_scale, model, image, num_samples):\n",
    "    # Expand the single image to a batch\n",
    "    image_batch = image.repeat(num_samples, 1, 1, 1)\n",
    "    # Add Gaussian noise\n",
    "    noisy_images = add_gaussian_noise(noise_scale, image_batch)\n",
    "\n",
    "    # Predict using the model\n",
    "    predictions = model(noisy_images)\n",
    "\n",
    "    # Get the predicted classes for each noisy image\n",
    "    _, predicted_classes = predictions.max(1)\n",
    "    #Result\n",
    "    avg_noisy_image = noisy_images.mean(0, keepdim=True)\n",
    "    prediction = torch.bincount(predicted_classes).argmax().item()\n",
    "    # Return the most frequently predicted class\n",
    "    return avg_noisy_image, predictions, predicted_classes, prediction\n",
    "\n",
    "def show_image(input, noise_scale):\n",
    "    fig, axes = plt.subplots(1, len(noise_scale), figsize=(20, 10))\n",
    "    if not isinstance(axes, (list, np.ndarray)):\n",
    "        axes = [axes]\n",
    "    for idx, noise in enumerate(noise_scale):\n",
    "        avg_noisy_image, predictions, predicted_classes, prediction= predict_with_randomized_smoothing(noise, global_model, input, 100)\n",
    "        print(f'Noise: {noise}, Ground Truth: {ground_truth_prediction}, Noised Result: {prediction}')\n",
    "        img = avg_noisy_image.squeeze().cpu().detach().numpy()\n",
    "        img = (img - img.min()) / (img.max() - img.min())\n",
    "        ax = axes[idx]\n",
    "        ax.imshow(img, cmap='gray')\n",
    "        ax.axis('off')\n",
    "        ax.set_title(f'noise_scale:{noise}')\n",
    "    plt.tight_layout()\n",
    "    # plt.savefig('picture.jpg')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "35c33f6a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Noise: 0, Ground Truth: 4, Noised Result: 4\n",
      "Noise: 0.5, Ground Truth: 4, Noised Result: 4\n",
      "Noise: 1, Ground Truth: 4, Noised Result: 4\n",
      "Noise: 1.5, Ground Truth: 4, Noised Result: 4\n",
      "Noise: 2.0, Ground Truth: 4, Noised Result: 4\n",
      "Noise: 2.5, Ground Truth: 4, Noised Result: 4\n",
      "Noise: 3.0, Ground Truth: 4, Noised Result: 2\n",
      "Noise: 3.5, Ground Truth: 4, Noised Result: 7\n",
      "Noise: 4.0, Ground Truth: 4, Noised Result: 4\n",
      "Noise: 4.5, Ground Truth: 4, Noised Result: 7\n",
      "Noise: 5.0, Ground Truth: 4, Noised Result: 7\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 2000x1000 with 11 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "args = get_args.get_arg()\n",
    "data_dir = './data/mnist'\n",
    "sigma = 0.01\n",
    "apply_transform = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.5,), (0.5,))])\n",
    "train_dataset = datasets.MNIST(data_dir, train=True, download=True, transform=apply_transform)\n",
    "test_dataset = datasets.MNIST(data_dir, train=False, download=True, transform=apply_transform)\n",
    "train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=128,\n",
    "                                          shuffle=True, num_workers=2)\n",
    "test_loader = torch.utils.data.DataLoader(test_dataset, batch_size=128,\n",
    "                                         shuffle=False, num_workers=2)\n",
    "device = 'cuda'\n",
    "global_model = MLPMnist()\n",
    "global_model.load_state_dict(torch.load(f'./checkpoints/mnist/{sigma}/best_model_mnist.pth'))\n",
    "global_model.to(device)\n",
    "dataiter = iter(test_loader)\n",
    "images, labels = next(dataiter)\n",
    "images, labels = images.to(device), labels.to(device)\n",
    "classes = ('0', '1', '2', '3',\n",
    "           '4', '5', '6', '7', '8', '9')\n",
    "ind = 4\n",
    "input = images[ind].unsqueeze(0)\n",
    "input.requires_grad = True\n",
    "global_model.eval()\n",
    "ground_truth_predictions = global_model(input)\n",
    "_, predicted_classes = ground_truth_predictions.max(1)\n",
    "ground_truth_prediction = torch.bincount(predicted_classes).argmax().item()\n",
    "noise_scale = [0, 0.5, 1, 1.5, 2.0, 2.5, 3.0 ,3.5, 4.0, 4.5, 5.0]\n",
    "show_image(input, noise_scale)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "9aa097e7-3dde-4e6e-baf6-bdc00bc58e13",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([28, 28])\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "input_image = input.squeeze().cpu().detach()  # Remove batch dimension and move to cpu\n",
    "print(input_image.size())\n",
    "# Plotting\n",
    "plt.imshow(input_image, cmap='gray' if input_image.ndim == 2 else None)\n",
    "plt.title(f'Ground Truth Prediction: {ground_truth_prediction}')\n",
    "plt.axis('off')  # Hide axis\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "1e1833dc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Original Image\n",
      "Predicted: 4  Probability: 0.9985949397087097\n"
     ]
    }
   ],
   "source": [
    "saliency = Saliency(global_model)\n",
    "ig = IntegratedGradients(global_model)\n",
    "dl = DeepLift(global_model)\n",
    "nt = NoiseTunnel(ig)\n",
    "print('-------------------')\n",
    "print('Original Image')\n",
    "print('Predicted:', classes[predicted_classes],\n",
    "      ' Probability:', torch.max(F.softmax(ground_truth_predictions, 1)).item())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "54964a4f-0d44-4117-9ecc-29cf376e4c8c",
   "metadata": {},
   "outputs": [],
   "source": [
    "def quantization_visualization(args, sigma, model, input, noise_scale, method=\"masked_image\", sign=\"positive\"):\n",
    "    change_rates = []\n",
    "    avg_noisy_image_baseline, predictions_baseline, predicted_classes_baseline, prediction_baseline = predict_with_randomized_smoothing(0, model, input, 100)\n",
    "    attr_ig_baseline, delta_baseline = attribute_image_features(ig, avg_noisy_image_baseline, baselines=avg_noisy_image_baseline * 0, n_steps=1000, return_convergence_delta=True)\n",
    "    num_escape = int(args.num_escape)\n",
    "    grad_baseline = attr_ig_baseline.reshape(attr_ig_baseline.shape[0], -1)\n",
    "    w0_baseline = grad_baseline.mean(dim=0)\n",
    "    des_baseline, _ = torch.sort(grad_baseline, dim=0, descending=True)\n",
    "    e11_baseline = des_baseline[num_escape:]\n",
    "    e1_baseline = e11_baseline.mean(dim=0)\n",
    "    e21_baseline = des_baseline[:-num_escape]\n",
    "    e2_baseline = e21_baseline.mean(dim=0)\n",
    "    e_baseline = torch.stack((e1_baseline, e2_baseline), dim=0)\n",
    "    e_baseline = torch.abs(torch.sub(w0_baseline, e_baseline))\n",
    "    e_baseline, _ = torch.max(e_baseline, dim=0)\n",
    "    milestone_baseline = sigma * sigma / 2\n",
    "    m_baseline = normal.Normal(0, sigma * sigma)\n",
    "    g_x_baseline = torch.zeros(w0_baseline.shape[0], 3).to(device)\n",
    "    g_x_baseline[:, 0] = 1 - m_baseline.cdf(torch.sub(milestone_baseline, w0_baseline))\n",
    "    g_x_baseline[:, 1] = m_baseline.cdf(torch.sub(milestone_baseline, w0_baseline)) - m_baseline.cdf(torch.sub(-milestone_baseline, w0_baseline))\n",
    "    g_x_baseline[:, 2] = m_baseline.cdf(torch.sub(-milestone_baseline, w0_baseline))\n",
    "    g_x_p_baseline, _ = torch.sort(g_x_baseline, dim=1, descending=True)\n",
    "    g_x_p_baseline = g_x_p_baseline[:, :2]\n",
    "    radius_baseline = (sigma / 2 * (m_baseline.icdf(g_x_p_baseline[:, 0]) - m_baseline.icdf(g_x_p_baseline[:, 1])))\n",
    "    failed_baseline = torch.gt(e_baseline, radius_baseline)\n",
    "    g_x_baseline[failed_baseline == True, 0] = 0.0\n",
    "    g_x_baseline[failed_baseline == True, 1] = 1.0\n",
    "    g_x_baseline[failed_baseline == True, 2] = 0.0\n",
    "    v_x_baseline = torch.argmax(g_x_baseline, dim=1)\n",
    "    v_x_baseline = F.one_hot(v_x_baseline, num_classes=3)\n",
    "    v_x_baseline = v_x_baseline.float()\n",
    "    v_x_baseline[:, 0] = torch.mul(v_x_baseline[:, 0], args.gamma)\n",
    "    v_x_baseline[:, 1] = torch.mul(v_x_baseline[:, 1], 0.0)\n",
    "    v_x_baseline[:, 2] = torch.mul(v_x_baseline[:, 2], -args.gamma)\n",
    "    v_x_baseline = torch.sum(v_x_baseline, dim=1).float()\n",
    "    v_x_baseline = v_x_baseline.reshape(grad_baseline.shape).to(device)\n",
    "    grad_baseline = v_x_baseline\n",
    "    tensor_28x28_baseline = grad_baseline.squeeze().cpu().detach().reshape(28, 28)\n",
    "    attr_ig_baseline = np.transpose(tensor_28x28_baseline.detach().numpy(), (1, 0))\n",
    "    count_ig1 = attr_ig_baseline\n",
    "\n",
    "    for i in noise_scale:\n",
    "        avg_noisy_image, predictions, predicted_classes, prediction= predict_with_randomized_smoothing(i, model, input, 100)\n",
    "        attr_ig, delta = attribute_image_features(ig, avg_noisy_image, baselines=avg_noisy_image * 0, n_steps=1000, return_convergence_delta=True)\n",
    "        grad = attr_ig.reshape(attr_ig.shape[0], -1)\n",
    "        w0 = grad.mean(dim=0)\n",
    "        des, _ = torch.sort(grad, dim=0, descending=True)\n",
    "        e11 = des[num_escape:]\n",
    "        e1 = e11.mean(dim=0)\n",
    "        e21 = des[:-num_escape]\n",
    "        e2 = e21.mean(dim=0)\n",
    "        e = torch.stack((e1, e2), dim=0)\n",
    "        e = torch.abs(torch.sub(w0, e))\n",
    "        e, _ = torch.max(e, dim=0)\n",
    "        milestone = sigma * sigma / 2\n",
    "        m = normal.Normal(0, sigma * sigma)\n",
    "        g_x = torch.zeros(w0.shape[0], 3).to(device)\n",
    "        g_x[:, 0] = 1 - m.cdf(torch.sub(milestone, w0))\n",
    "        g_x[:, 1] = m.cdf(torch.sub(milestone, w0)) - m.cdf(torch.sub(-milestone, w0))\n",
    "        g_x[:, 2] = m.cdf(torch.sub(-milestone, w0))\n",
    "        g_x_p, _ = torch.sort(g_x, dim=1, descending=True)\n",
    "        g_x_p = g_x_p[:, :2]\n",
    "        radius = (sigma / 2 * (m.icdf(g_x_p[:, 0]) - m.icdf(g_x_p[:, 1])))\n",
    "        failed = torch.gt(e, radius)\n",
    "        g_x[failed == True, 0] = 0.0\n",
    "        g_x[failed == True, 1] = 1.0\n",
    "        g_x[failed == True, 2] = 0.0\n",
    "        v_x = torch.argmax(g_x, dim=1)\n",
    "        v_x = F.one_hot(v_x, num_classes=3)\n",
    "        v_x = v_x.float()\n",
    "        v_x[:, 0] = torch.mul(v_x[:, 0], args.gamma)\n",
    "        v_x[:, 1] = torch.mul(v_x[:, 1], 0.0)\n",
    "        v_x[:, 2] = torch.mul(v_x[:, 2], -args.gamma)\n",
    "        v_x = torch.sum(v_x, dim=1).float()\n",
    "        v_x = v_x.reshape(grad.shape).to(device)\n",
    "        grad = v_x\n",
    "        tensor_28x28 = grad.squeeze().cpu().detach().reshape(28, 28)\n",
    "        attr_ig = np.transpose(tensor_28x28.detach().numpy(), (1, 0))\n",
    "        count_ig2 = attr_ig\n",
    "        # print('-------------------')\n",
    "        # print('Smoothing Image')\n",
    "        # print('Predicted:', classes[prediction],\n",
    "        #       ' Probability:', torch.max(F.softmax(predictions, 1)).item())\n",
    "        # attr_ig = attr_ig[:, :, np.newaxis]\n",
    "        # avg_noisy_np = avg_noisy_image.squeeze().unsqueeze(-1).cpu().detach().numpy()\n",
    "        # _ = viz.visualize_image_attr(attr_ig, avg_noisy_np, method=method, sign=sign,\n",
    "        #                       show_colorbar=True, title=\"Overlayed Integrated Gradients\")\n",
    "        explanation_change = count_ig2 - count_ig1\n",
    "        num_zeros = np.sum(explanation_change == 0)\n",
    "        explanation_change_rate = num_zeros / explanation_change.size\n",
    "        change_rates.append(explanation_change_rate)\n",
    "    return change_rates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "0de068b7-a219-4de9-9f19-12a41ae5a925",
   "metadata": {},
   "outputs": [],
   "source": [
    "noise_level = np.linspace(0, 10, 1001)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "70f492be-f53f-4168-9f46-f190149a3d7f",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_001 = pd.read_excel('./noise_level_change_rates001.xlsx')\n",
    "df_001['Moving Average'] = df_001['Retention Rates'].rolling(window=30).mean()\n",
    "df_001 = df_02.dropna(subset=['Noise Level', 'Retention Rates', 'Moving Average'])\n",
    "\n",
    "df_01 = pd.read_excel('./noise_level_change_rates01.xlsx')\n",
    "df_01['Moving Average'] = df_01['Retention Rates'].rolling(window=30).mean()\n",
    "df_01 = df_01.dropna(subset=['Noise Level', 'Retention Rates', 'Moving Average'])\n",
    "\n",
    "df_05 = pd.read_excel('./noise_level_change_rates05.xlsx')\n",
    "df_05['Moving Average'] = df_05['Retention Rates'].rolling(window=30).mean()\n",
    "df_05 = df_05.dropna(subset=['Noise Level', 'Retention Rates', 'Moving Average'])\n",
    "\n",
    "df_10 = pd.read_excel('./noise_level_change_rates1.xlsx')\n",
    "df_10['Moving Average'] = df_10['Retention Rates'].rolling(window=30).mean()\n",
    "df_10 = df_10.dropna(subset=['Noise Level', 'Retention Rates', 'Moving Average'])\n",
    "\n",
    "df_15 = pd.read_excel('./noise_level_change_rates15.xlsx')\n",
    "df_15['Moving Average'] = df_15['Retention Rates'].rolling(window=30).mean()\n",
    "df_15 = df_15.dropna(subset=['Noise Level', 'Retention Rates', 'Moving Average'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "ccaa9147-4cb9-4e55-b9a7-421d1cea416f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10, 6))\n",
    "plt.plot(df_001['Noise Level'].to_numpy(), df_001['Moving Average'].to_numpy(), label='Sigma=0.01', color='yellow')\n",
    "plt.plot(df_01['Noise Level'].to_numpy(), df_01['Moving Average'].to_numpy(), label='Sigma=0.1', color='red')\n",
    "plt.plot(df_05['Noise Level'].to_numpy(), df_05['Moving Average'].to_numpy(), label='Sigma=0.5', color='blue')\n",
    "plt.plot(df_10['Noise Level'].to_numpy(), df_10['Moving Average'].to_numpy(), label='Sigma=1.0', color='green')\n",
    "plt.plot(df_15['Noise Level'].to_numpy(), df_15['Moving Average'].to_numpy(), label='Sigma=1.5', color='black')\n",
    "plt.xlabel('Noise Level')\n",
    "plt.ylabel('Retention Rates')\n",
    "plt.title('Noise Level vs Retention Rates with Moving Average')\n",
    "plt.legend()\n",
    "plt.savefig('retention_rate.jpg')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "40e402b1-5602-4bae-a446-4626fbaa680e",
   "metadata": {},
   "outputs": [],
   "source": [
    "change_rates = quantization_visualization(args=args, sigma=0.1, model=global_model, input=input, noise_scale=noise_level,\n",
    "                           method='masked_image', sign='positive')\n",
    "df = pd.DataFrame({\n",
    "    'Noise Level': noise_level,\n",
    "    'Retention Rates': change_rates\n",
    "})\n",
    "excel_path = './noise_level_change_rates01.xlsx'\n",
    "df.to_excel(excel_path, index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d07dc972-0754-4262-9ae2-60a9b1f89c9d",
   "metadata": {},
   "outputs": [],
   "source": [
    "change_rates = quantization_visualization(args=args, sigma=0.2, model=global_model, input=input, noise_scale=noise_level,\n",
    "                           method='masked_image', sign='positive')\n",
    "df = pd.DataFrame({\n",
    "    'Noise Level': noise_level,\n",
    "    'Change Rates': change_rates\n",
    "})\n",
    "excel_path = './noise_level_change_rates02.xlsx'\n",
    "df.to_excel(excel_path, index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "55776b15-5be3-4c21-b515-cca65568ec28",
   "metadata": {},
   "outputs": [],
   "source": [
    "change_rates = quantization_visualization(args=args, sigma=0.5, model=global_model, input=input, noise_scale=noise_level,\n",
    "                           method='masked_image', sign='positive')\n",
    "df = pd.DataFrame({\n",
    "    'Noise Level': noise_level,\n",
    "    'Change Rates': change_rates\n",
    "})\n",
    "excel_path = './noise_level_change_rates05.xlsx'\n",
    "df.to_excel(excel_path, index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f39ed21d-25d8-48e1-bea3-66f5a5612903",
   "metadata": {},
   "outputs": [],
   "source": [
    "change_rates = quantization_visualization(args=args, sigma=1.0, model=global_model, input=input, noise_scale=noise_level,\n",
    "                           method='masked_image', sign='positive')\n",
    "df = pd.DataFrame({\n",
    "    'Noise Level': noise_level,\n",
    "    'Change Rates': change_rates\n",
    "})\n",
    "excel_path = './noise_level_change_rates1.xlsx'\n",
    "df.to_excel(excel_path, index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "1677352d-a8e7-49d5-9a1d-760fcfccb354",
   "metadata": {},
   "outputs": [],
   "source": [
    "change_rates = quantization_visualization(args=args, sigma=1.5, model=global_model, input=input, noise_scale=noise_level,\n",
    "                           method='masked_image', sign='positive')\n",
    "df = pd.DataFrame({\n",
    "    'Noise Level': noise_level,\n",
    "    'Change Rates': change_rates\n",
    "})\n",
    "excel_path = './noise_level_change_rates15.xlsx'\n",
    "df.to_excel(excel_path, index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "465ebf91-0a0b-40e5-acc3-da7e9da11cdc",
   "metadata": {},
   "outputs": [],
   "source": [
    "def visualization(args, sigma, model, input, noise_scale, method=\"masked_image\", sign=\"positive\"):\n",
    "    num_escape = int(args.num_escape)\n",
    "    avg_noisy_image, predictions, predicted_classes, prediction= predict_with_randomized_smoothing(noise_scale, model, input, 100)\n",
    "    attr_ig, delta = attribute_image_features(ig, avg_noisy_image, baselines=avg_noisy_image * 0, n_steps=1000, return_convergence_delta=True)\n",
    "    grad = attr_ig.reshape(attr_ig.shape[0], -1)\n",
    "    w0 = grad.mean(dim=0)\n",
    "    des, _ = torch.sort(grad, dim=0, descending=True)\n",
    "    e11 = des[num_escape:]\n",
    "    e1 = e11.mean(dim=0)\n",
    "    e21 = des[:-num_escape]\n",
    "    e2 = e21.mean(dim=0)\n",
    "    e = torch.stack((e1, e2), dim=0)\n",
    "    e = torch.abs(torch.sub(w0, e))\n",
    "    e, _ = torch.max(e, dim=0)\n",
    "    milestone = sigma * sigma / 2\n",
    "    m = normal.Normal(0, sigma * sigma)\n",
    "    g_x = torch.zeros(w0.shape[0], 3).to(device)\n",
    "    g_x[:, 0] = 1 - m.cdf(torch.sub(milestone, w0))\n",
    "    g_x[:, 1] = m.cdf(torch.sub(milestone, w0)) - m.cdf(torch.sub(-milestone, w0))\n",
    "    g_x[:, 2] = m.cdf(torch.sub(-milestone, w0))\n",
    "    g_x_p, _ = torch.sort(g_x, dim=1, descending=True)\n",
    "    g_x_p = g_x_p[:, :2]\n",
    "    radius = (sigma / 2 * (m.icdf(g_x_p[:, 0]) - m.icdf(g_x_p[:, 1])))\n",
    "    failed = torch.gt(e, radius)\n",
    "    g_x[failed == True, 0] = 0.0\n",
    "    g_x[failed == True, 1] = 1.0\n",
    "    g_x[failed == True, 2] = 0.0\n",
    "    v_x = torch.argmax(g_x, dim=1)\n",
    "    v_x = F.one_hot(v_x, num_classes=3)\n",
    "    v_x = v_x.float()\n",
    "    v_x[:, 0] = torch.mul(v_x[:, 0], args.gamma)\n",
    "    v_x[:, 1] = torch.mul(v_x[:, 1], 0.0)\n",
    "    v_x[:, 2] = torch.mul(v_x[:, 2], -args.gamma)\n",
    "    v_x = torch.sum(v_x, dim=1).float()\n",
    "    v_x = v_x.reshape(grad.shape).to(device)\n",
    "    grad = v_x\n",
    "    tensor_28x28 = grad.squeeze().cpu().detach().reshape(28, 28)\n",
    "    attr_ig = np.transpose(tensor_28x28.detach().numpy(), (1, 0))\n",
    "    count_ig2 = attr_ig\n",
    "    print('-------------------')\n",
    "    print('Smoothing Image')\n",
    "    print('Predicted:', classes[prediction],\n",
    "          ' Probability:', torch.max(F.softmax(predictions, 1)).item())\n",
    "    attr_ig = attr_ig[:, :, np.newaxis]\n",
    "    avg_noisy_np = avg_noisy_image.squeeze().unsqueeze(-1).cpu().detach().numpy()\n",
    "    print(attr_ig.size)\n",
    "    print(avg_noisy_np.size)\n",
    "    _ = viz.visualize_image_attr(attr_ig, avg_noisy_np, method=method, sign=sign,\n",
    "                          show_colorbar=True, title=\"Overlayed Integrated Gradients\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "08a329ce-ba93-4b61-a773-a58716310db8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: 4  Probability: 0.9986069798469543\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: 4  Probability: 0.9987125396728516\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: 4  Probability: 0.9993587136268616\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: 4  Probability: 0.9999774694442749\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: 4  Probability: 0.9998441934585571\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: 4  Probability: 0.9999781847000122\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: 4  Probability: 0.9999426603317261\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: 4  Probability: 0.9999938011169434\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualization(args=args, sigma=0.5, model=global_model, input=input, noise_scale=0.001,\n",
    "                           method='masked_image', sign='positive')\n",
    "visualization(args=args, sigma=0.5, model=global_model, input=input, noise_scale=0.01,\n",
    "                           method='masked_image', sign='positive')\n",
    "visualization(args=args, sigma=0.5, model=global_model, input=input, noise_scale=0.1,\n",
    "                           method='masked_image', sign='positive')\n",
    "visualization(args=args, sigma=0.5, model=global_model, input=input, noise_scale=0.5,\n",
    "                           method='masked_image', sign='positive')\n",
    "visualization(args=args, sigma=0.5, model=global_model, input=input, noise_scale=0.9,\n",
    "                           method='masked_image', sign='positive')\n",
    "visualization(args=args, sigma=0.5, model=global_model, input=input, noise_scale=1.0,\n",
    "                           method='masked_image', sign='positive')\n",
    "visualization(args=args, sigma=0.5, model=global_model, input=input, noise_scale=1.5,\n",
    "                           method='masked_image', sign='positive')\n",
    "visualization(args=args, sigma=0.5, model=global_model, input=input, noise_scale=2.0,\n",
    "                           method='masked_image', sign='positive')\n",
    "#Noise0.5_Confi1.0_Mnist4.png"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "518077c2-cd1a-43a5-90b3-ea517a2b20bd",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: 4  Probability: 0.998688280582428\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualization(args=args, sigma=0.01, model=global_model, input=input, noise_scale=0.01,\n",
    "                           method='masked_image', sign='positive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "ce20f1ca-dc74-4eee-8295-4007a7080bf1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: 4  Probability: 0.9987032413482666\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "visualization(args=args, sigma=0.5, model=global_model, input=input, noise_scale=0.01,\n",
    "                           method='masked_image', sign='positive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "dbcdd0d6-b909-4132-a91b-c9b79c732376",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: 4  Probability: 0.9986959099769592\n"
     ]
    },
    {
     "ename": "AssertionError",
     "evalue": "Cannot normalize by scale factor = 0",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mAssertionError\u001b[0m                            Traceback (most recent call last)",
      "Cell \u001b[0;32mIn[12], line 1\u001b[0m\n\u001b[0;32m----> 1\u001b[0m \u001b[43mvisualization\u001b[49m\u001b[43m(\u001b[49m\u001b[43margs\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43msigma\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;241;43m1.0\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mmodel\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mglobal_model\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mnoise_scale\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;241;43m0.01\u001b[39;49m\u001b[43m,\u001b[49m\n\u001b[1;32m      2\u001b[0m \u001b[43m                           \u001b[49m\u001b[43mmethod\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[38;5;124;43mmasked_image\u001b[39;49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43msign\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[38;5;124;43mpositive\u001b[39;49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[43m)\u001b[49m\n",
      "Cell \u001b[0;32mIn[8], line 46\u001b[0m, in \u001b[0;36mvisualization\u001b[0;34m(args, sigma, model, input, noise_scale, method, sign)\u001b[0m\n\u001b[1;32m     44\u001b[0m attr_ig \u001b[38;5;241m=\u001b[39m attr_ig[:, :, np\u001b[38;5;241m.\u001b[39mnewaxis]\n\u001b[1;32m     45\u001b[0m avg_noisy_np \u001b[38;5;241m=\u001b[39m avg_noisy_image\u001b[38;5;241m.\u001b[39msqueeze()\u001b[38;5;241m.\u001b[39munsqueeze(\u001b[38;5;241m-\u001b[39m\u001b[38;5;241m1\u001b[39m)\u001b[38;5;241m.\u001b[39mcpu()\u001b[38;5;241m.\u001b[39mdetach()\u001b[38;5;241m.\u001b[39mnumpy()\n\u001b[0;32m---> 46\u001b[0m _ \u001b[38;5;241m=\u001b[39m \u001b[43mviz\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mvisualize_image_attr\u001b[49m\u001b[43m(\u001b[49m\u001b[43mattr_ig\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mavg_noisy_np\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mmethod\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43mmethod\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43msign\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[43msign\u001b[49m\u001b[43m,\u001b[49m\n\u001b[1;32m     47\u001b[0m \u001b[43m                      \u001b[49m\u001b[43mshow_colorbar\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;28;43;01mTrue\u001b[39;49;00m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtitle\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[38;5;124;43mOverlayed Integrated Gradients\u001b[39;49m\u001b[38;5;124;43m\"\u001b[39;49m\u001b[43m)\u001b[49m\n",
      "File \u001b[0;32m~/miniconda3/envs/xy_py38/lib/python3.8/site-packages/captum/attr/_utils/visualization.py:263\u001b[0m, in \u001b[0;36mvisualize_image_attr\u001b[0;34m(attr, original_image, method, sign, plt_fig_axis, outlier_perc, cmap, alpha_overlay, show_colorbar, title, fig_size, use_pyplot)\u001b[0m\n\u001b[1;32m    260\u001b[0m     plt_axis\u001b[38;5;241m.\u001b[39mimshow(original_image)\n\u001b[1;32m    261\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m    262\u001b[0m     \u001b[38;5;66;03m# Choose appropriate signed attributions and normalize.\u001b[39;00m\n\u001b[0;32m--> 263\u001b[0m     norm_attr \u001b[38;5;241m=\u001b[39m \u001b[43m_normalize_attr\u001b[49m\u001b[43m(\u001b[49m\u001b[43mattr\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43msign\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43moutlier_perc\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mreduction_axis\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;241;43m2\u001b[39;49m\u001b[43m)\u001b[49m\n\u001b[1;32m    265\u001b[0m     \u001b[38;5;66;03m# Set default colormap and bounds based on sign.\u001b[39;00m\n\u001b[1;32m    266\u001b[0m     \u001b[38;5;28;01mif\u001b[39;00m VisualizeSign[sign] \u001b[38;5;241m==\u001b[39m VisualizeSign\u001b[38;5;241m.\u001b[39mall:\n",
      "File \u001b[0;32m~/miniconda3/envs/xy_py38/lib/python3.8/site-packages/captum/attr/_utils/visualization.py:97\u001b[0m, in \u001b[0;36m_normalize_attr\u001b[0;34m(attr, sign, outlier_perc, reduction_axis)\u001b[0m\n\u001b[1;32m     95\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[1;32m     96\u001b[0m     \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mAssertionError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mVisualize Sign type is not valid.\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n\u001b[0;32m---> 97\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43m_normalize_scale\u001b[49m\u001b[43m(\u001b[49m\u001b[43mattr_combined\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mthreshold\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[0;32m~/miniconda3/envs/xy_py38/lib/python3.8/site-packages/captum/attr/_utils/visualization.py:49\u001b[0m, in \u001b[0;36m_normalize_scale\u001b[0;34m(attr, scale_factor)\u001b[0m\n\u001b[1;32m     48\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21m_normalize_scale\u001b[39m(attr: ndarray, scale_factor: \u001b[38;5;28mfloat\u001b[39m):\n\u001b[0;32m---> 49\u001b[0m     \u001b[38;5;28;01massert\u001b[39;00m scale_factor \u001b[38;5;241m!=\u001b[39m \u001b[38;5;241m0\u001b[39m, \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mCannot normalize by scale factor = 0\u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m     50\u001b[0m     \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mabs\u001b[39m(scale_factor) \u001b[38;5;241m<\u001b[39m \u001b[38;5;241m1e-5\u001b[39m:\n\u001b[1;32m     51\u001b[0m         warnings\u001b[38;5;241m.\u001b[39mwarn(\n\u001b[1;32m     52\u001b[0m             \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mAttempting to normalize by value approximately 0, visualized results\u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m     53\u001b[0m             \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mmay be misleading. This likely means that attribution values are all\u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m     54\u001b[0m             \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mclose to 0.\u001b[39m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m     55\u001b[0m         )\n",
      "\u001b[0;31mAssertionError\u001b[0m: Cannot normalize by scale factor = 0"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "ad962869",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: Shirt  Probability: 0.9690296649932861\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "quantization_visualization(args=args, sigma=sigma, model=global_model, input=input, noise_scale=0.5,\n",
    "                           method='original_image', sign='positive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "3af0c7a9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: Shirt  Probability: 0.9950793981552124\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "quantization_visualization(args=args, sigma=sigma, model=global_model, input=input, noise_scale=0.9,\n",
    "                           method='masked_image', sign='positive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "a6fffa5c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: Shirt  Probability: 0.9999667406082153\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "quantization_visualization(args=args, sigma=sigma, model=global_model, input=input, noise_scale=1.0,\n",
    "                           method='masked_image', sign='positive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "86d8d35e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: Sandal  Probability: 1.0\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "quantization_visualization(args=args, sigma=sigma, model=global_model, input=input, noise_scale=1.5,\n",
    "                           method='masked_image', sign='positive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "aa15128d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------------\n",
      "Smoothing Image\n",
      "Predicted: Sandal  Probability: 1.0\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 600x600 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "quantization_visualization(args=args, sigma=sigma, model=global_model, input=input, noise_scale=2.0,\n",
    "                           method='masked_image', sign='positive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f8c28a28",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Py38",
   "language": "python",
   "name": "xy_py38"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.18"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}