CoFlow: Coordinated Few-Step Flow for Offline Multi-Agent Decision Making

2026-05-02 · Guowei Zou, Haitao Wang, Beiwen Zhang, Boning Zhang, Hejun Wu arxiv

Generative models have emerged as a promising paradigm for offline multi-agent reinforcement learning (MARL), but existing approaches require many iterative sampling steps. Recent few-step acceleration methods either distill a joint teacher into independent students or apply averaged velocity fields independently to each agent. Unfortunately, these few-step approaches hurt inter-agent coordination. We show that the efficiency-coordination trade-off is not inherent: single-pass multi-agent generation can preserve coordination when the velocity field is natively joint-coupled. We propose Coordinated few-step Flow (CoFlow), an architecture that combines Coordinated Velocity Attention (CVA) with Adaptive Coordination Gating. A finite-difference consistency surrogate further replaces memory-prohibitive Jacobian-vector product backpropagation through the averaged velocity field with two stop-gradient forward passes. Across 60 configurations spanning MPE, MA-MuJoCo, and SMAC, CoFlow matches or surpasses Gaussian policies, value-based methods, transformer policies, diffusion models, and prior flow baselines on episodic return. Three independent coordination probes confirm that CoFlow's improvements arise from inter-agent coordination rather than per-agent capacity. A denoising-step sweep shows that single-pass inference suffices on every configuration. CoFlow reaches state-of-the-art coordination quality in 1-3 denoising steps under both centralized and decentralized execution. Project Page: https://guowei-zou.github.io/coflow/

📄 PDF Abstract BibTeX arXiv:2605.01457

Code (0)

등록된 구현이 없습니다.

Tasks

Multi-agent Reinforcement Learning Decision Making

A Scalable Deep Reinforcement Learning Model for Online Scheduling Coflows of Multi-Stage Jobs for High Performance Computing

2021-12-21 · Xin Wang, Hong Shen

Coflow is a recently proposed networking abstraction to help improve the communication performance of data-parallel computing jobs. In multi-stage jobs, each job consists of multiple coflows and is represented by a Direc…

Deep Reinforcement Learning Graph Neural Network Reinforcement Learning (RL)Scheduling

EcoFlow: Efficient Convolutional Dataflows for Low-Power Neural Network Accelerators

2022-02-04 · Lois Orosa, Skanda Koppula, Yaman Umuroglu, Konstantinos Kanellopoulos 외

Dilated and transposed convolutions are widely used in modern convolutional neural networks (CNNs). These kernels are used extensively during CNN training and inference of applications such as image segmentation and high…

Generative Adversarial Network Image Generation Image Segmentation Semantic Segmentation

DeCoFlow: Structural Decomposition of Normalizing Flows for Continual Anomaly Detection

2026-06-25 · Hun Im, Jungi Lee, Subeen Cha, Pilsung Kang arxiv

In industrial environments, new product categories arrive sequentially, requiring continual anomaly detection without access to past data. Normalizing Flows (NFs) provide exact density estimation but suffer from catastro…

Density Estimation Anomaly Detection

Coordinated Frequency Control through Safe Reinforcement Learning

2022-01-30 · Yi Zhou, Liangcai Zhou, Di Shi, Xiaoying Zhao

With widespread deployment of renewables, the electric power grids are experiencing increasing dynamics and uncertainties, with its secure operation being threatened. Existing frequency control schemes based on day-ahead…

AI Agent Decision Making reinforcement-learning Reinforcement Learning+2

OM2P: Offline Multi-Agent Mean-Flow Policy

2025-08-08 · Zhuoran Li, Xun Wang, Hai Zhong, Qingxin Xia 외 arxiv

Generative models, especially diffusion and flow-based models, have been promising in offline multi-agent reinforcement learning. However, integrating powerful generative models into this framework poses unique challenge…

Multi-agent Reinforcement Learning