SecurityAnalysisFL

This project demonstrates a federated learning (FL) setup using PyTorch for training a MobileNetV2 model, with the Flower framework to manage the federated learning process. The project also incorporates differential privacy for enhanced security, using local differential privacy (DP) mechanisms.

Table of Contents

• Introduction
• Features
• Installation
• Usage
  • Running the Federated Learning Simulation
• Contributing
• License

Introduction

In this project, we leverage federated learning to train a neural network model in a distributed manner across multiple clients. Each client trains the model locally using its own data and shares only the updated model parameters with the central server. No raw data is shared, ensuring privacy.

Key Components:

• Flower: A federated learning framework used to simulate the FL process.
• PyTorch: For building and training the MobileNetV2 model

Features

• Federated Learning Simulation: Manage multiple clients and a central server to orchestrate model training.
• MobileNetV2: Utilize a pre-trained MobileNetV2 model, which is fine-tuned on custom datasets.
• Data Preprocessing: Includes image transformations such as resizing, normalization, and grayscale conversion.
• Scalable Simulation: Number of clients and resource allocation are parameterizable.

Installation

Prerequisites

• Conda (Anaconda or Miniconda)
• Python 3.10+

Setting Up the Environment

To set up the environment using Conda, follow these steps:

1. Clone the repository:

   git clone <repository-url>
   cd <repository-directory>

2. Create the Conda environment

   conda env create -f environment.yml

3. Activate the environment:

   conda activate federated-learning

usage

Running multi Experiments

4. Run multi Experiments

   cd scripts/ && bash multirun.sh