analiza aiqbench
This paper presents DHCG, a novel framework that dynamically constructs hierarchical collaboration graphs for multi-agent systems using LLMs. By reframing agent orchestration as a partially observable Markov decision process, it addresses limitations in composition and scalability through action-aware preference optimization. The approach demonstrates state-of-the-art performance across code generation, mathematical reasoning, and domain-specific tasks by adaptively expanding or finalizing agent roles based on execution feedback.
DHCG: Dynamic Construction of Hierarchical Collaboration Graphs for LLM-Based Multi-Agent Reasoning
arXiv:2610.07835v1 Announce Type: new Abstract: LLM-based multi-agent systems (MAS) have demonstrated strong capabilities in solving complex problems across diverse domains. Recently, the dynamic orchestration of agent systems has become an important research direction. However, existing methods suffer from limited composition, misaligned dependencies, and inflexible scale, restricting their ability to adapt to reasoning requirements during execution. To address these limitations, we reframe MAS design as a partially observable Markov decision process, in which both the composition and scale of the MAS are dynamically determined. We propose DHCG, a novel framework that coordinates three modules (Planner, Worker, and Generator) to progressively construct a dynamic hierarchical collaboration graph from scratch based on the query and evolving execution feedback. At each step, guided by feedback, the Planner generates a set of distinct and complementary roles tailored to the current reasoning needs and selectively routes relevant information to each role. It can also finalize the hierarchical collaboration graph early or progressively expand it when additional reasoning is required. We further introduce action-aware preference optimization to train the Planner to make more effective decisions when constructing hierarchical collaboration graphs. We systematically evaluate DHCG across code generation, mathematical reasoning, and domain-specific reasoning benchmarks. DHCG achieves state-of-the-art average performance among the compared methods, improving over the single-agent baseline by 13.06 points and outperforming both static and dynamic MAS baselines by 2.77-8.02 points. Additional experiments further demonstrate its generalization across different Planner backbones and unseen Worker models.
Abstract:LLM-based multi-agent systems (MAS) have demonstrated strong capabilities in solving complex problems across diverse domains. Recently, the dynamic orchestration of agent systems has become an important research direction. However, existing methods suffer from limited composition, misaligned dependencies, and inflexible scale, restricting their ability to adapt to reasoning requirements during execution. To address these limitations, we reframe MAS design as a partially observable Markov decision process, in which both the composition and scale of the MAS are dynamically determined. We propose DHCG, a novel framework that coordinates three modules (Planner, Worker, and Generator) to progressively construct a dynamic hierarchical collaboration graph from scratch based on the query and evolving execution feedback. At each step, guided by feedback, the Planner generates a set of distinct and complementary roles tailored to the current reasoning needs and selectively routes relevant information to each role. It can also finalize the hierarchical collaboration graph early or progressively expand it when additional reasoning is required. We further introduce action-aware preference optimization to train the Planner to make more effective decisions when constructing hierarchical collaboration graphs. We systematically evaluate DHCG across code generation, mathematical reasoning, and domain-specific reasoning benchmarks. DHCG achieves state-of-the-art average performance among the compared methods, improving over the single-agent baseline by 13.06 points and outperforming both static and dynamic MAS baselines by 2.77-8.02 points. Additional experiments further demonstrate its generalization across different Planner backbones and unseen Worker models.
Artificial Intelligence (cs.AI)
arXiv:2610.07835 [cs.AI]
(or arXiv:2610.07835v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.07835 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 327 cuvinte
Subiecte conexe
DRG-MAPPO: Hierarchical Dynamic Role-Graph Multi-Agent Reinforcement Learning for Cooperative Air Combat
arXiv · 12 sept. 2026
A Unified Framework for Context-Aware and Relation-Aware Graph Retrieval-Augmented Generation
arXiv · 17 iun. 2026
Towards Security-Auditable LLM Agents: A Unified Graph Representation
arXiv · 15 oct. 2026
Learning Multi-Scale Hypergraph for High-Order Brain Connectivity Analysis
arXiv · 3 iun. 2026