Rehearsal-based continual learning in Medical IoT (MIoT) autoencoders requires replaying latent codes of past clinical tasks to prevent catastrophic forgetting. When these latent codes are derived from patient data, each replay constitutes a fresh privacy query, and existing differential privacy (DP) composition theorems-designed for finite, pre-specified query sequences-fail to bound cumulative privacy leakage over an unbounded task horizon. This paper presents PRISM-DP, a Privacy-Resilient Infinite-Sequence Memory framework that introduces a martingale-based DP composition theorem for unbounded replay sequences. PRISM-DP proves that a geometrically decaying per-task privacy schedule combin... More >

The rise of multi-factor authentication (MFA) has significantly enhanced cybersecurity postures, yet its effectiveness is increasingly challenged by sophisticated social engineering attacks, particularly those exploiting MFA fatigue. MFA fatigue, a tactic where attackers inundate users with authentication prompts, aims to induce erroneous approvals, as notably exemplified by the 2022 Uber breach. This phenomenon undermines the very security MFA is designed to provide by leveraging human vulnerabilities. Game theory, a powerful mathematical framework for analyzing strategic decision-making, offers a robust methodology to model the dynamic interactions between attackers and defenders. By apply... More >