Neural Orchestration for Multi-Agent Systems: A Deep Learning Framework for Optimal Agent Selection in Multi-Domain Task Environments

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Agrawal, Kushagra, Nargund, Nisharg
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911602198970368
author Agrawal, Kushagra
Nargund, Nisharg
author_facet Agrawal, Kushagra
Nargund, Nisharg
contents Multi-agent systems (MAS) are foundational in simulating complex real-world scenarios involving autonomous, interacting entities. However, traditional MAS architectures often suffer from rigid coordination mechanisms and difficulty adapting to dynamic tasks. We propose MetaOrch, a neural orchestration framework for optimal agent selection in multi-domain task environments. Our system implements a supervised learning approach that models task context, agent histories, and expected response quality to select the most appropriate agent for each task. A novel fuzzy evaluation module scores agent responses along completeness, relevance, and confidence dimensions, generating soft supervision labels for training the orchestrator. Unlike previous methods that hard-code agent-task mappings, MetaOrch dynamically predicts the most suitable agent while estimating selection confidence. Experiments in simulated environments with heterogeneous agents demonstrate that our approach achieves 86.3% selection accuracy, significantly outperforming baseline strategies including random selection and round-robin scheduling. The modular architecture emphasizes extensibility, allowing agents to be registered, updated, and queried independently. Results suggest that neural orchestration offers a powerful approach to enhancing the autonomy, interpretability, and adaptability of multi-agent systems across diverse task domains.
format Preprint
id arxiv_https___arxiv_org_abs_2505_02861
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Neural Orchestration for Multi-Agent Systems: A Deep Learning Framework for Optimal Agent Selection in Multi-Domain Task Environments
Agrawal, Kushagra
Nargund, Nisharg
Multiagent Systems
Artificial Intelligence
Neural and Evolutionary Computing
Multi-agent systems (MAS) are foundational in simulating complex real-world scenarios involving autonomous, interacting entities. However, traditional MAS architectures often suffer from rigid coordination mechanisms and difficulty adapting to dynamic tasks. We propose MetaOrch, a neural orchestration framework for optimal agent selection in multi-domain task environments. Our system implements a supervised learning approach that models task context, agent histories, and expected response quality to select the most appropriate agent for each task. A novel fuzzy evaluation module scores agent responses along completeness, relevance, and confidence dimensions, generating soft supervision labels for training the orchestrator. Unlike previous methods that hard-code agent-task mappings, MetaOrch dynamically predicts the most suitable agent while estimating selection confidence. Experiments in simulated environments with heterogeneous agents demonstrate that our approach achieves 86.3% selection accuracy, significantly outperforming baseline strategies including random selection and round-robin scheduling. The modular architecture emphasizes extensibility, allowing agents to be registered, updated, and queried independently. Results suggest that neural orchestration offers a powerful approach to enhancing the autonomy, interpretability, and adaptability of multi-agent systems across diverse task domains.
title Neural Orchestration for Multi-Agent Systems: A Deep Learning Framework for Optimal Agent Selection in Multi-Domain Task Environments
topic Multiagent Systems
Artificial Intelligence
Neural and Evolutionary Computing
url https://arxiv.org/abs/2505.02861