diffGHOST: Diffusion based Generative Hedged Oblivious Synthetic Trajectories

TL;DR

diffGHOST is a diffusion-based generative model that creates privacy-preserving synthetic trajectories by mitigating memorization and ensuring utility, addressing privacy concerns in trajectory data sharing.

Contribution

The paper introduces diffGHOST, a novel diffusion model with latent space segmentation that reduces memorization and enhances privacy in synthetic trajectory generation.

Findings

diffGHOST effectively reduces memorization of sensitive data.

The model preserves utility while providing privacy guarantees.

It outperforms existing methods in privacy preservation.

Abstract

Trajectories are nowadays valuable information for a wide range of applications. However they are also inherently sensitive, as they contain highly personal information about individuals. Facing this challenge, synthesizing mobility trajectories has emerged as a promising solution to leverage mobility information while preserving privacy. State-of-the-art models, often rely on the false assumptions of generative models implicit privacy and fails to provide privacy guarantees while preserving trajectories utility. Here, we introduce diffGHOST, a conditional diffusion model based on latent space segmentation, designed to answer this challenge. Thus, this paper propose a methodology that identify and mitigate memorization of critical samples using condition segments of a learn latent space.