arXiv subjects

Keman Huang

Publications and source records attributed to Keman Huang.

7 recordsLinked to original sources

arXiv2026

Finding Where the Buck Stops: An Automated Failure Attribution-Based Reflection Framework for Multi-Agent Collaboration

Multi-agent systems (MAS) powered by large language models have shown promise for complex tasks but suffer from high failure rates. Current self-reflection methods for MAS require all agents to reflect upon failure, overlooking a critical reality: failures typically stem from a specific agent leading the task astray, namely the decisive error agent, while others merely fulfill their regular duties. Forcing regular-behaving agents to reflect contaminates their memory with wrong insights. Hence, we propose DoCtOR (Diagnose-then-Correct PPO-enhanced Reflection), a novel reflection framework that enhances multi-agent collaboration. DoCtOR first identifies the decisive error step and decisive error agent through automated failure attribution, then employs counterfactual reasoning to generate a corrected decisive error step, and finally engages only the decisive error agent to produce targeted reflections. Experimental results show DoCtOR achieves 22%, 26%, and 27% improvements over initial success rates on HotPotQA, ChartQAPro, and Mind2Web datasets, outperforming Reflexion, Retroformer, and COPPER. We further establish the generalizability of our diagnose-then-correct paradigm and demonstrate that in low-resource settings, focusing reflection on reasoning steps after the decisive error step achieves comparable quality to reflecting on the complete failure trajectory.

cs.AI↗

arXiv2026

Double-Edged Sword or Sharp Tool? Designing and Evaluating Triadic LLM-Teacher Collaboration for K-12 Writing at Scale

The double-edged sword of integrating Large Language Models (LLMs) requires an effective triadic collaboration mechanism among LLMs, teachers and students, especially for K-12 education. By developing a triadic collaboration system to support K-12 writing learning, a multidimensional evaluation framework grounded in Systemic Functional Linguistics and the suggestion trajectory tracing pipeline, this paper contributes a large-scale empirical dataset involving $57,954$ essays from $10,195$ students across $120$ schools over two years. Our findings confirm the efficacy of this system in improving writing quality through a strategic labor division: the LLM serves as a generative engine to mitigate teacher burnout, and the teacher acts as a pedagogical gatekeeper and bridge to guarantee feedback quality. While both LLM and teacher are critical for skill improvement, we uncover a ceiling effect where excessive linguistic expansion yields diminishing marginal utility. These suggest a dynamically adaptive LLM-teacher collaboration as student proficiency increases.

cs.AI↗

arXiv2025

AgentODRL: A Large Language Model-based Multi-agent System for ODRL Generation

The Open Digital Rights Language (ODRL) is a pivotal standard for automating data rights management. However, the inherent logical complexity of authorization policies, combined with the scarcity of high-quality "Natural Language-to-ODRL" training datasets, impedes the ability of current methods to efficiently and accurately translate complex rules from natural language into the ODRL format. To address this challenge, this research leverages the potent comprehension and generation capabilities of Large Language Models (LLMs) to achieve both automation and high fidelity in this translation process. We introduce AgentODRL, a multi-agent system based on an Orchestrator-Workers architecture. The architecture consists of specialized Workers, including a Generator for ODRL policy creation, a Decomposer for breaking down complex use cases, and a Rewriter for simplifying nested logical relationships. The Orchestrator agent dynamically coordinates these Workers, assembling an optimal pathway based on the complexity of the input use case. Specifically, we enhance the ODRL Generator by incorporating a validator-based syntax strategy and a semantic reflection mechanism powered by a LoRA-finetuned model, significantly elevating the quality of the generated policies. Extensive experiments were conducted on a newly constructed dataset comprising 770 use cases of varying complexity, all situated within the context of data spaces. The results, evaluated using ODRL syntax and semantic scores, demonstrate that our proposed Orchestrator-Workers system, enhanced with these strategies, achieves superior performance on the ODRL generation task.

cs.MA↗

arXiv2025

Shadows in the Code: Exploring the Risks and Defenses of LLM-based Multi-Agent Software Development Systems

The rapid advancement of Large Language Model (LLM)-driven multi-agent systems has significantly streamlined software developing tasks, enabling users with little technical expertise to develop executable applications. While these systems democratize software creation through natural language requirements, they introduce significant security risks that remain largely unexplored. We identify two risky scenarios: Malicious User with Benign Agents (MU-BA) and Benign User with Malicious Agents (BU-MA). We introduce the Implicit Malicious Behavior Injection Attack (IMBIA), demonstrating how multi-agent systems can be manipulated to generate software with concealed malicious capabilities beneath seemingly benign applications, and propose Adv-IMBIA as a defense mechanism. Evaluations across ChatDev, MetaGPT, and AgentVerse frameworks reveal varying vulnerability patterns, with IMBIA achieving attack success rates of 93%, 45%, and 71% in MU-BA scenarios, and 71%, 84%, and 45% in BU-MA scenarios. Our defense mechanism reduced attack success rates significantly, particularly in the MU-BA scenario. Further analysis reveals that compromised agents in the coding and testing phases pose significantly greater security risks, while also identifying critical agents that require protection against malicious user exploitation. Our findings highlight the urgent need for robust security measures in multi-agent software development systems and provide practical guidelines for implementing targeted, resource-efficient defensive strategies.

cs.CR↗

arXiv2024

Practically implementing an LLM-supported collaborative vulnerability remediation process: a team-based approach

Incorporating LLM into cybersecurity operations, a typical real-world high-stakes task, is critical but non-trivial in practice. Using cybersecurity as the study context, we conduct a three-step mix-method study to incorporate LLM into the vulnerability remediation process effectively. Specifically, we deconstruct the deficiencies in user satisfaction within the existing process (Study 1). This inspires us to design, implement, and empirically validate an LLM-supported collaborative vulnerability remediation process through a field study (Study 2). Given LLM's diverse contributions, we further investigate LLM's double-edge roles through the analysis of remediation reports and follow-up interviews (Study 3). In essence, our contribution lies in promoting an efficient LLM-supported collaborative vulnerability remediation process. These first-hand, real-world pieces of evidence suggest that when incorporating LLMs into practical processes, facilitating the collaborations among all associated stakeholders, reshaping LLMs' roles according to task complexity, as well as approaching the short-term side effects of improved user engagement facilitated by LLMs with a rational mindset.

cs.CR↗

arXiv2024

Depending on yourself when you should: Mentoring LLM with RL agents to become the master in cybersecurity games

Integrating LLM and reinforcement learning (RL) agent effectively to achieve complementary performance is critical in high stake tasks like cybersecurity operations. In this study, we introduce SecurityBot, a LLM agent mentored by pre-trained RL agents, to support cybersecurity operations. In particularly, the LLM agent is supported with a profile module to generated behavior guidelines, a memory module to accumulate local experiences, a reflection module to re-evaluate choices, and an action module to reduce action space. Additionally, it adopts the collaboration mechanism to take suggestions from pre-trained RL agents, including a cursor for dynamic suggestion taken, an aggregator for multiple mentors' suggestions ranking and a caller for proactive suggestion asking. Building on the CybORG experiment framework, our experiences show that SecurityBot demonstrates significant performance improvement compared with LLM or RL standalone, achieving the complementary performance in the cybersecurity games.

cs.CR↗

arXiv2024

ISSF: The Intelligent Security Service Framework for Cloud-Native Operation

The growing system complexity from microservice architectures and the bilateral enhancement of artificial intelligence (AI) for both attackers and defenders presents increasing security challenges for cloud-native operations. In particular, cloud-native operators require a holistic view of the dynamic security posture for the cloud-native environment from a defense aspect. Additionally, both attackers and defenders can adopt advanced AI technologies. This makes the dynamic interaction and benchmark among different intelligent offense and defense strategies more crucial. Hence, following the multi-agent deep reinforcement learning (RL) paradigm, this research develops an agent-based intelligent security service framework (ISSF) for cloud-native operation. It includes a dynamic access graph model to represent the cloud-native environment and an action model to represent offense and defense actions. Then we develop an approach to enable the training, publishing, and evaluating of intelligent security services using diverse deep RL algorithms and training strategies, facilitating their systematic development and benchmark. The experiments demonstrate that our framework can sufficiently model the security posture of a cloud-native system for defenders, effectively develop and quantitatively benchmark different services for both attackers and defenders and guide further service optimization.

cs.CR↗