Yu Su

The paper introduces extsc{Spore}, a novel, training-free, and highly efficient privacy extraction attack that targets sensitive information stored in the memory of LLM agents during inference, outperforming existing state-of-the-art methods.

The paper proposes SubPopMark, a novel subpopulation-driven framework that injects harmless, verifiable markers into distilled datasets to prevent copyright infringement and data leakage.

The paper introduces HIDBench, a new benchmark for evaluating LLMs' ability to perform host-based intrusion detection using complex, noisy system logs, finding that model performance degrades significantly with increased data complexity.

VIPER-MCP is a novel, end-to-end automated framework that detects and dynamically confirms the exploitability of taint-style vulnerabilities in Model Context Protocol (MCP) servers, achieving high-fidelity vulnerability discovery in real-world systems.

The paper proposes In-Context Reward Adaptation, a transformer-based framework that uses in-context learning and auxiliary signals (like human response time) to robustly model diverse and unseen human preferences for better AI alignment.

The paper introduces AGENTCL, a rigorous evaluation framework that uses controlled task streams to accurately measure an agent's ability to accumulate and reuse knowledge across multiple tasks, thereby addressing limitations in current continual learning benchmarks.

EVA-Net proposes a two-stage framework that uses action videos as semantic priors to achieve strong subject-independent EEG motor decoding, significantly outperforming text-based methods.

The paper introduces SkillHarm, a comprehensive benchmark and automated framework for evaluating skill-based attacks across the entire agent skill-use lifecycle, demonstrating that current agents remain highly vulnerable to both fixed-payload and self-mutating poisoning attacks.

The paper proposes Resonant Context Anchoring (RCA), a lightweight, training-free method that enhances factual faithfulness in LLMs by dynamically amplifying the signal of external context evidence during inference.

SkillGuard introduces a novel, skill-centric permission framework to secure LLM agent skill ecosystems by jointly regulating both context influence and runtime action side effects.

The paper presents BEATS, a human-in-the-loop LLM framework for bootstrapping product attribute taxonomies from scratch.

This paper proposes a preconditioning layer for stable weight conditioning in LLM training.

This paper identifies two novel location inference attacks against k-nearest neighbor queries (kNNQ) and proposes DPRS, a differential privacy framework that effectively protects location privacy while maintaining high query utility.