WINFlowNets: Warm-up Integrated Networks Training of Generative Flow Networks for Robotics and Machine Fault Adaptation

2026-03-17

AI Summary

This work addresses the limitations of conventional generative flow networks (CFlowNets) in robotic control, which rely on pretrained retrieval networks and struggle to adapt to dynamic or failure-prone environments—especially when pretrained data are unavailable or unrepresentative. To overcome this dependency, the authors propose WINFlowNets, a novel framework that enables, for the first time, end-to-end joint training of the flow and retrieval networks. By integrating a warm-start strategy, a shared replay buffer, and a unified training architecture, WINFlowNets eliminates the need for pretrained data. Experiments in simulated robotic tasks demonstrate that the approach significantly enhances training stability and average reward under sample-scarce conditions, while exhibiting superior adaptability to dynamic changes and system failures.

Technology Category

• Intelligent Robots: Behavior Learning & Control
• Machine Learning: Adversarial Learning & Robustness
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Generative Flow Networks for continuous scenarios (CFlowNets) have shown promise in solving sequential decision-making tasks by learning stochastic policies using a flow and a retrieval network. Despite their demonstrated efficiency compared to state-of-the-art Reinforcement Learning (RL) algorithms, their practical application in robotic control tasks is constrained by the reliance on pre-training the retrieval network. This dependency poses challenges in dynamic robotic environments, where pre-training data may not be readily available or representative of the current environment. This paper introduces WINFlowNets, a novel CFlowNets framework that enables the co-training of flow and retrieval networks. WINFlowNets begins with a warm-up phase for the retrieval network to bootstrap its policy, followed by a shared training architecture and a shared replay buffer for co-training both networks. Experiments in simulated robotic environments demonstrate that WINFlowNets surpasses CFlowNets and state-of-the-art RL algorithms in terms of average reward and training stability. Furthermore, WINFlowNets exhibits strong adaptive capability in fault environments, making it suitable for tasks that demand quick adaptation with limited sample data. These findings highlight WINFlowNets' potential for deployment in dynamic and malfunction-prone robotic systems, where traditional pre-training or sample inefficient data collection may be impractical.

Problem

• Generative Flow Networks
• robotic control
• pre-training dependency
• dynamic environments
• fault adaptation

Innovation

• Generative Flow Networks
• Co-training
• Warm-up Phase
• Robotic Adaptation
• Sample Efficiency