arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2406.08404cs.LGcs.AI
收藏

将价值迭代网络扩展到5000层以实现超长期规划

Scaling Value Iteration Networks to 5000 Layers for Extreme Long-Term Planning

Center of Excellence for Generative AI, King Abdullah University of Science and Technology(生成人工智能卓越中心,国王 Abdullah 科学与技术大学);The Swiss AI Lab IDSIA/USI/SUPSI, Switzerland(瑞士人工智能实验室 IDSIA/USI/SUPSI,瑞士);The University of Southampton(南安普顿大学);National University of Singapore(新加坡国立大学)

Yuhui Wang, Qingyuan Wu, Dylan R. Ashley, Francesco Faccio, Weida Li, Chao Huang, Jürgen Schmidhuber

AI总结:

针对价值迭代网络难以处理长期大规模规划的问题,提出动态转移VIN(DT-VIN),通过动态转移核增强潜在MDP表示能力,并引入自适应高速公路损失缓解梯度消失,成功扩展到5000层并在迷宫导航、连续控制及月球车导航等任务中取得突破。

摘要:

价值迭代网络(VIN)是一种用于规划的端到端可微分神经网络架构。它通过在潜在马尔可夫决策过程(MDP)上运行的可微分规划模块,展现出对未见领域的强大泛化能力。然而,VIN难以扩展到长期和大规模规划任务,例如在100x100迷宫中导航——这一任务通常需要数千个规划步骤才能解决。我们观察到这一不足源于两个问题:潜在MDP的表示能力以及规划模块的深度。我们通过为潜在MDP增加动态转移核来解决这些问题,从而大幅提升其表示能力,同时为了缓解梯度消失问题,引入一种“自适应高速公路损失”,构建跳跃连接以改善梯度流动。我们在2D/3D迷宫导航环境、连续控制以及真实世界的月球车导航任务上评估了我们的方法。我们发现,我们提出的新方法——动态转移VIN(DT-VIN)——能够扩展到5000层,并解决上述任务中具有挑战性的版本。总之,我们相信DT-VIN代表了在复杂环境中执行长期大规模规划方面迈出的实质性一步。

英文摘要:

The Value Iteration Network (VIN) is an end-to-end differentiable neural network architecture for planning. It exhibits strong generalization to unseen domains by incorporating a differentiable planning module that operates on a latent Markov Decision Process (MDP). However, VINs struggle to scale to long-term and large-scale planning tasks, such as navigating a 100x100 maze -- a task that typically requires thousands of planning steps to solve. We observe that this deficiency is due to two issues: the representation capacity of the latent MDP and the planning module's depth. We address these by augmenting the latent MDP with a dynamic transition kernel, dramatically improving its representational capacity, and, to mitigate the vanishing gradient problem, introduce an "adaptive highway loss" that constructs skip connections to improve gradient flow. We evaluate our method on 2D/3D maze navigation environments, continuous control, and the real-world Lunar rover navigation task. We find that our new method, named Dynamic Transition VIN (DT-VIN), scales to 5000 layers and solves challenging versions of the above tasks. Altogether, we believe that DT-VIN represents a concrete step forward in performing long-term large-scale planning in complex environments.

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。

↑