arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.02186cs.LGcs.AI
收藏

进化引导的生成流网络

Evolution Guided Generative Flow Networks

National University of Singapore(新加坡国立大学);The Hong Kong University of Science and Technology(香港科学与技术大学)

Zarif Ikram, Ling Pan, Dianbo Liu

AI总结:

针对生成流网络在长时域、稀疏奖励场景下训练困难的问题,提出结合进化算法的EGFN方法,通过进化训练智能体参数、优先回放轨迹来增强训练,在多类任务上验证了有效性并开源代码。

摘要:

生成流网络(GFlowNets)是一类概率生成模型,用于学习按组合对象的奖励比例对其进行采样。GFlowNets的一大挑战是在处理长时域和稀疏奖励时难以实现有效训练。为解决该问题,我们提出进化引导的生成流网络(EGFN),这是一种利用进化算法(EA)增强GFlowNets训练的简单且高效的方法。我们的方法可基于任意GFlowNets训练目标运行:通过EA训练一组智能体参数,将生成的轨迹存入优先回放缓冲区,再利用存储的轨迹训练GFlowNets智能体。我们在大量玩具任务与真实基准任务上开展了全面研究,结果表明该方法在处理长轨迹和稀疏奖励方面具备有效性。我们已在http://github.com/zarifikram/egfn发布代码。

英文摘要:

Generative Flow Networks (GFlowNets) are a family of probabilistic generative models that learn to sample compositional objects proportional to their rewards. One big challenge of GFlowNets is training them effectively when dealing with long time horizons and sparse rewards. To address this, we propose Evolution guided generative flow networks (EGFN), a simple but powerful augmentation to the GFlowNets training using Evolutionary algorithms (EA). Our method can work on top of any GFlowNets training objective, by training a set of agent parameters using EA, storing the resulting trajectories in the prioritized replay buffer, and training the GFlowNets agent using the stored trajectories. We present a thorough investigation over a wide range of toy and real-world benchmark tasks showing the effectiveness of our method in handling long trajectories and sparse rewards. We release the code at http://github.com/zarifikram/egfn.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑