Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Protecting Activity Sensing Data Privacy Using Hierarchical Information Dissociation

Guangjing Wang, Hanqing Guo, Yuanda Wang, Bocheng Chen, Ce Zhou, Qiben Yan

TL;DR

Hippo is the first unified model that achieves two goals: perturbing the sensitive attributes and controlling the disclosure of sensitive information in mobile sensing data and is designed as a latent guidance-based diffusion model.

Abstract

Smartphones and wearable devices have been integrated into our daily lives, offering personalized services. However, many apps become overprivileged as their collected sensing data contains unnecessary sensitive information. For example, mobile sensing data could reveal private attributes (e.g., gender and age) and unintended sensitive features (e.g., hand gestures when entering passwords). To prevent sensitive information leakage, existing methods must obtain private labels and users need to specify privacy policies. However, they only achieve limited control over information disclosure. In this work, we present Hippo to dissociate hierarchical information including private metadata and multi-grained activity information from the sensing data. Hippo achieves fine-grained control over the disclosure of sensitive information without requiring private labels. Specifically, we design a latent guidance-based diffusion model, which generates multi-grained versions of raw sensor data conditioned on hierarchical latent activity features. Hippo enables users to control the disclosure of sensitive information in sensing data, ensuring their privacy while preserving the necessary features to meet the utility requirements of applications. Hippo is the first unified model that achieves two goals: perturbing the sensitive attributes and controlling the disclosure of sensitive information in mobile sensing data. Extensive experiments show that Hippo can anonymize personal attributes and transform activity information at various resolutions across different types of sensing data.

Protecting Activity Sensing Data Privacy Using Hierarchical Information Dissociation

TL;DR

Hippo is the first unified model that achieves two goals: perturbing the sensitive attributes and controlling the disclosure of sensitive information in mobile sensing data and is designed as a latent guidance-based diffusion model.

Abstract

Smartphones and wearable devices have been integrated into our daily lives, offering personalized services. However, many apps become overprivileged as their collected sensing data contains unnecessary sensitive information. For example, mobile sensing data could reveal private attributes (e.g., gender and age) and unintended sensitive features (e.g., hand gestures when entering passwords). To prevent sensitive information leakage, existing methods must obtain private labels and users need to specify privacy policies. However, they only achieve limited control over information disclosure. In this work, we present Hippo to dissociate hierarchical information including private metadata and multi-grained activity information from the sensing data. Hippo achieves fine-grained control over the disclosure of sensitive information without requiring private labels. Specifically, we design a latent guidance-based diffusion model, which generates multi-grained versions of raw sensor data conditioned on hierarchical latent activity features. Hippo enables users to control the disclosure of sensitive information in sensing data, ensuring their privacy while preserving the necessary features to meet the utility requirements of applications. Hippo is the first unified model that achieves two goals: perturbing the sensitive attributes and controlling the disclosure of sensitive information in mobile sensing data. Extensive experiments show that Hippo can anonymize personal attributes and transform activity information at various resolutions across different types of sensing data.
Paper Structure (22 sections, 1 theorem, 10 equations, 4 figures, 2 tables)

This paper contains 22 sections, 1 theorem, 10 equations, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Metadata Information Protection
  4. Semantic Information Protection
  5. Problem Formulation
  6. User Privacy Leakage in Activity Sensing Data
  7. Mobile Activity Sensing Data Utility
  8. System Design
  9. System Model
  10. Multi-Resolution Feature Extraction
  11. Multi-grained Sensing Data Generation
  12. Implementation and Evaluation
  13. Experimental Setup
  14. Metadata-level Information Evaluation
  15. Feature-level Information Evaluation
  16. ...and 7 more sections

Key Result

Lemma 4.1

Given two stacked CAEs whose outputs are layer features $z_i$ and $z_j$ respectively. Then $z_j$ is low-resolution version of $z_i$ such that $\mathcal{H}(z_j)<\mathcal{H}(z_i)$ when $i<j$.

Figures (4)

  • Figure 1: The latent feature guidance diffusion model of Hippo. Hippo acts as a middleware between the OS Sensor Manager and Apps. Hippo reconstructs raw data and the Apps obtain data from Hippo. In multi-grained data generation, right arrows are the forward diffusion process, and left arrows refer to the reverse diffusion process conditioned on multi-resolution layer features.
  • Figure 2: Comparison of different generative models.
  • Figure 3: The impact of modality and inference steps. Acc. means 3-axis in accelerometer modality, Gyro. means gyroscope, Mag. means magnetometer, and All means the 9-axis modalities. The utility ratio is the accuracy of the model on reconstructed data over the accuracy of raw data.
  • Figure 4: The pedometer utility on multi-grained data.

Theorems & Definitions (1)

  • Lemma 4.1