Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
The paper introduces TADP-RME, a trust-adaptive differential privacy framework designed to enhance the reliability of data-driven systems in adversarial settings by dynamically adjusting the privacy budget based on an inverse trust score. To further mitigate privacy leakage from inference attacks, TADP-RME employs Reverse Manifold Embedding, a nonlinear transformation that disrupts local geometric relationships while maintaining differential privacy guarantees. Experiments demonstrate that TADP-RME improves the privacy-utility trade-off, reducing attack success rates by up to 3.1% compared to existing methods.
By adaptively modulating the privacy budget based on user trust and disrupting geometric data structures, TADP-RME achieves a better privacy-utility trade-off than fixed-budget DP methods in adversarial settings.
Ensuring reliability in adversarial settings necessitates treating privacy as a foundational component of data-driven systems. While differential privacy and cryptographic protocols offer strong guarantees, existing schemes rely on a fixed privacy budget, leading to a rigid utility-privacy trade-off that fails under heterogeneous user trust. Moreover, noise-only differential privacy preserves geometric structure, which inference attacks exploit, causing privacy leakage. We propose TADP-RME (Trust-Adaptive Differential Privacy with Reverse Manifold Embedding), a framework that enhances reliability under varying levels of user trust. It introduces an inverse trust score in the range [0,1] to adaptively modulate the privacy budget, enabling smooth transitions between utility and privacy. Additionally, Reverse Manifold Embedding applies a nonlinear transformation to disrupt local geometric relationships while preserving formal differential privacy guarantees through post-processing. Theoretical and empirical results demonstrate improved privacy-utility trade-offs, reducing attack success rates by up to 3.1 percent without significant utility degradation. The framework consistently outperforms existing methods against inference attacks, providing a unified approach for reliable learning in adversarial environments.