Yi Yang

The paper proposes SafeReview, a co-evolutionary adversarial training framework that significantly improves the robustness of LLM-based peer review systems against sophisticated adversarial hidden prompts.

The paper introduces Checkerboard, a novel, learning-free clean-label backdoor attack that efficiently poisons training data to compromise model integrity with minimal poisoning budget.

SecureForge is an automated pipeline that significantly reduces cybersecurity vulnerabilities in LLM-generated code by optimizing system prompts, achieving up to a 48% reduction in output vulnerabilities.

The paper introduces STORYLENSWRITER, a novel framework that significantly improves personalized story rewriting by incorporating context-aware narrative enrichment, outperforming style-only adaptation.

The paper introduces a unified framework to fairly evaluate LLM agentic capabilities by standardizing diverse benchmarks and separating the effects of the LLM model from the surrounding framework and environment.

The paper proposes SAAS, a novel RL framework that equips LLM agents with self-awareness to precisely regulate search behavior, significantly mitigating costly over-search without sacrificing accuracy.

The paper proposes MemoAttack, a memory-driven black-box jailbreak framework that systematically models, evolves, and selects attack experiences to significantly enhance LLM jailbreaking success rates.

The paper introduces SURE, a unified framework designed to standardize and improve the comparability and reproducibility of evaluations for advanced speech understanding models.

The paper proposes a highly reconfigurable 256x128 in-memory computing array that significantly improves efficiency and performance for analog computing by introducing novel components for ADC, weighted accumulation, and bitcell design.

This paper investigates if upper-face affective cues enhance audiovisual sentence recognition, especially when audio is degraded, finding that while mouth cues are crucial for robustness, upper-face cues improve overall confidence and performance under noise.

SelSkill introduces a dual-granularity preference learning framework that treats skill use as a 'skill-or-skip' decision, significantly improving agent performance and execution precision in complex agentic tasks.

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.