add dp notebook

holgerroth · Feb 25, 2025 · b8647ac · b8647ac
1 parent 751ef81
commit b8647ac
Show file tree

Hide file tree

Showing 18 changed files with 1,014 additions and 1,145 deletions.
diff --git a/examples/getting_started/pt/nvflare_pt_getting_started.ipynb b/examples/getting_started/pt/nvflare_pt_getting_started.ipynb
@@ -258,7 +258,11 @@
   {
    "cell_type": "markdown",
    "id": "1b70da5d-ba8b-4e65-b47f-44bb9bddae4d",
-   "metadata": {},
+   "metadata": {
+    "jupyter": {
+     "source_hidden": true
+    }
+   },
    "source": [
     "#### 2. Define a FedJob\n",
     "The `FedJob` is used to define how controllers and executors are placed within a federated job using the `to(object, target)` routine.\n",
@@ -271,7 +275,11 @@
    "cell_type": "code",
    "execution_count": null,
    "id": "13771bfb-901f-485a-9a23-84db1ccd5fe4",
-   "metadata": {},
+   "metadata": {
+    "jupyter": {
+     "source_hidden": true
+    }
+   },
    "outputs": [],
    "source": [
     "from src.net import Net\n",
@@ -368,7 +376,11 @@
   {
    "cell_type": "markdown",
    "id": "9ac3f0a8-06bb-4bea-89d3-4a5fc5b76c63",
-   "metadata": {},
+   "metadata": {
+    "jupyter": {
+     "source_hidden": true
+    }
+   },
    "source": [
     "#### 6. Run FL Simulation\n",
     "Finally, we can run our FedJob in simulation using NVFlare's [simulator](https://nvflare.readthedocs.io/en/main/user_guide/nvflare_cli/fl_simulator.html) under the hood. We can also specify which GPU should be used to run this client, which is helpful for simulated environments. The results will be saved in the specified `workdir`."
@@ -438,7 +450,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.8.19"
+   "version": "3.11.7"
   }
  },
  "nbformat": 4,

diff --git a/examples/tutorials/self-paced-training/.gitignore b/examples/tutorials/self-paced-training/.gitignore
@@ -0,0 +1,3 @@
+.virtual_documents
+
+job_configs/
diff --git a/..._and_privacy/chapter-5_Privacy_In_Federated_Learning/05.0_introduction/introduction.ipynb b/..._and_privacy/chapter-5_Privacy_In_Federated_Learning/05.0_introduction/introduction.ipynb
@@ -11,9 +11,9 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "nvflare_example",
+   "display_name": "Python 3 (ipykernel)",
    "language": "python",
-   "name": "nvflare_example"
+   "name": "python3"
   },
   "language_info": {
    "codemirror_mode": {
@@ -25,7 +25,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.2"
+   "version": "3.11.7"
   }
  },
  "nbformat": 4,

diff --git a/...er-5_Privacy_In_Federated_Learning/05.1_privacy_filter/figs/nvflare_brats18.png b/...er-5_Privacy_In_Federated_Learning/05.1_privacy_filter/figs/nvflare_brats18.png
diff --git a/...ivacy/chapter-5_Privacy_In_Federated_Learning/05.1_privacy_filter/privacy_filtering.ipynb b/...ivacy/chapter-5_Privacy_In_Federated_Learning/05.1_privacy_filter/privacy_filtering.ipynb
@@ -5,43 +5,23 @@
    "id": "1398ef0a-f189-4d04-a8a9-276a17ab0f8b",
    "metadata": {},
    "source": [
-    "# Federated Learning with Differential Privacy for BraTS18 Segmentation\n",
+    "# Federated Learning with Differential Privacy\n",
     "\n",
-    "Please make sure you set up virtual environment and follows [example root readme](../../README.md)\n",
+    "Please make sure you set up a virtual environment and follow [example root readme](../../README.md) before starting this notebook.\n",
+    "Then, install the requirements.\n",
     "\n",
-    "## Introduction to MONAI, BraTS and Differential Privacy"
+    "<div class=\"alert alert-block alert-info\"> <b>NOTE</b> Some of the cells below generate long text output.  We're using <pre>%%capture --no-display --no-stderr cell_output</pre> to suppress this output.  Comment or delete this line in the cells below to restore full output.</div>"
    ]
   },
   {
-   "cell_type": "markdown",
-   "id": "af5f3c69-aeba-4cea-89c8-d54fd6520ab1",
-   "metadata": {},
-   "source": [
-    "### MONAI\n",
-    "This example shows how to use [NVIDIA FLARE](https://nvflare.readthedocs.io/en/main/index.html) on medical image applications.\n",
-    "It uses [MONAI](https://github.com/Project-MONAI/MONAI),\n",
-    "which is a PyTorch-based, open-source framework for deep learning in healthcare imaging, part of the PyTorch Ecosystem."
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "8a3b7e0b-9dbd-4d21-8b59-a3d08cf2b2bb",
+   "cell_type": "code",
+   "execution_count": 2,
+   "id": "5002e45c-f58e-4f68-bb5a-9626e084947f",
    "metadata": {},
+   "outputs": [],
    "source": [
-    "### BraTS\n",
-    "The application shown in this example is volumetric (3D) segmentation of brain tumor subregions from multimodal MRIs based on BraTS 2018 data.\n",
-    "It uses a deep network model published by [Myronenko 2018](https://arxiv.org/abs/1810.11654) [1].\n",
-    "\n",
-    "The model is trained to segment 3 nested subregions of primary brain tumors (gliomas): the \"enhancing tumor\" (ET), the \"tumor core\" (TC), the \"whole tumor\" (WT) based on 4 aligned input MRI scans (T1c, T1, T2, FLAIR). \n",
-    "\n",
-    "<img src=\"https://developer.download.nvidia.com/assets/Clara/Images/clara_pt_brain_mri_segmentation_workflow.png\" alt=\"clara_pt_brain_mri_segmentation_workflow\" width=\"600\"/>\\n\n",
-    "\n",
-    "- The ET is described by areas that show hyper intensity in T1c when compared to T1, but also when compared to \"healthy\" white matter in T1c. \n",
-    "- The TC describes the bulk of the tumor, which is what is typically resected. The TC entails the ET, as well as the necrotic (fluid-filled) and the non-enhancing (solid) parts of the tumor. \n",
-    "- The WT describes the complete extent of the disease, as it entails the TC and the peritumoral edema (ED), which is typically depicted by hyper-intense signal in FLAIR.\n",
-    "\n",
-    "To run this example, please make sure you have downloaded BraTS 2018 data, which can be obtained from [Multimodal Brain Tumor Segmentation Challenge (BraTS) 2018](https://www.med.upenn.edu/cbica/brats2018.html) [2-6]. Please download the data to [./dataset_brats18/dataset](./dataset_brats18/dataset). It should result in a sub-folder `./dataset_brats18/dataset/training`.\n",
-    "In this example, we split BraTS18 dataset into [4 subsets](./dataset_brats18/datalist) for 4 clients. Each client requires at least a 12 GB GPU to run. "
+    "%%capture --no-display --no-stderr cell_output\n",
+    "!pip install -r requirements.txt"
    ]
   },
   {
@@ -50,29 +30,8 @@
    "metadata": {},
    "source": [
     "### Differential Privacy (DP)\n",
-    "[Differential Privacy (DP)](https://arxiv.org/abs/1910.00962) [7] is method for ensuring that Federated Learning (FL) preserves privacy by obfuscating the model updates sent from clients to the central server.\n",
-    "This example shows the usage of a MONAI-based trainer for medical image applications with NVFlare, as well as the usage of DP filters in your FL training. DP is added as a filter in `config_fed_client.json`. Here, we use the \"Sparse Vector Technique\", i.e. the [SVTPrivacy](https://nvflare.readthedocs.io/en/main/apidocs/nvflare.app_common.filters.svt_privacy.html) protocol, as utilized in [Li et al. 2019](https://arxiv.org/abs/1910.00962) [7] (see [Lyu et al. 2016](https://arxiv.org/abs/1603.01699) [8] for more information)."
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "33306153-a8c5-4c2b-9eea-28c3e0d705a6",
-   "metadata": {},
-   "source": [
-    "## Prepare local configs\n",
-    "First, we add the image and datalist directory roots to `config_train.json` files for generating the absolute path to the dataset by replacing the `DATASET_ROOT` and  `DATALIST_ROOT` placeholders. In the current folder structure, it will be `${PWD}/dataset_brats18/dataset` for `DATASET_ROOT` and  `${PWD}/dataset_brats18/datalist` for `DATALIST_ROOT` but you can update the below `sed` commands if the data is located somewhere else."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 7,
-   "id": "be8a7f47-2a93-4992-85c0-d597f4ecf3d5",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "%%bash\n",
-    "sed -i \"s|DATASET_ROOT|${PWD}/dataset_brats18/dataset|g\" config_train.json\n",
-    "sed -i \"s|DATALIST_ROOT|${PWD}/dataset_brats18/datalist|g\" config_train.json"
+    "[Differential Privacy (DP)](https://arxiv.org/abs/1910.00962) [7] is a method for ensuring that Federated Learning (FL) preserves privacy by obfuscating the model updates sent from clients to the central server. \n",
+    "This example shows the usage of a CIFAR-10 training code with NVFlare, as well as the usage of DP filters in your FL training. DP is added as a filter in `config_fed_client.json`. Here, we use the \"Sparse Vector Technique\", i.e. the [SVTPrivacy](https://nvflare.readthedocs.io/en/main/apidocs/nvflare.app_common.filters.svt_privacy.html) protocol, as utilized in [Li et al. 2019](https://arxiv.org/abs/1910.00962) [7] (see [Lyu et al. 2016](https://arxiv.org/abs/1603.01699) [8] for more information)."
    ]
   },
   {
@@ -83,33 +42,8 @@
     "## Run experiments with FL simulator\n",
     "### Training with FL simulator\n",
     "FL simulator is used to simulate FL experiments or debug codes, not for real FL deployment.\n",
-    "In this example, we assume four local GPUs with at least 12GB of memory are available."
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "8d256389-4112-46e6-86bd-115c9bf2e189",
-   "metadata": {},
-   "source": [
-    "Then, we can run the FL simulator with 1 client for centralized training"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "a9583165-af58-45ed-a86d-9fbfc74d80ca",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "!python3 -u -m nvflare.private.fed.app.simulator.simulator './configs/brats_central' -w './workspace_brats/brats_central' -n 1 -t 1 -gpu 0"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "f1121bc2-1118-4b64-8d61-06e1a49bc7ef",
-   "metadata": {},
-   "source": [
-    "Similarly, run the FL simulator with 4 clients for federated learning by running"
+    "\n",
+    "First, train a model using the FedAvg algorithm with four clients without DP."
    ]
   },
   {
@@ -119,7 +53,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!nvflare simulator './configs/brats_fedavg' -w './workspace_brats/brats_fedavg' -n 4 -t 4 -gpu 0,1,2,3"
+    "!nvflare simulator './configs/brats_fedavg' -w './workspace_brats/brats_fedavg' -n 4 -t 4"
    ]
   },
   {
@@ -137,7 +71,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "!nvflare simulator './configs/brats_fedavg_dp' -w './workspace_brats/brats_fedavg_dp' -n 4 -t 4 -gpu 0,1,2,3"
+    "!nvflare simulator './configs/brats_fedavg_dp' -w './workspace_brats/brats_fedavg_dp' -n 4 -t 4"
    ]
   },
   {