arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2210.11698cs.LGcs.AI
收藏

基于变分稀疏门的鲁棒动力学学习

Learning Robust Dynamics through Variational Sparse Gating

Université de Montréal(蒙特利尔大学);Mila- Quebec AI Institute(米拉-魁北克人工智能研究所);École de technologie supérieure(高等技术学院);University of Toronto(多伦多大学);Google Brain(谷歌大脑)

Arnav Kumar Jain, Shivakanth Sujit, Shruti Joshi, Vincent Michalski, Danijar Hafner, Samira Ebrahimi-Kahou

摘要:

从感官输入中学习世界模型,使智能体能够通过想象未来的结果来计划行动。世界模型以前已经被证明可以在具有很少对象的模拟环境中提高采样效率,但是还没有成功地应用于具有许多对象的环境。在有许多对象的环境中,通常只有少数对象同时移动或交互。在本文中,我们将研究整合这种稀疏交互的归纳偏差到从像素训练的世界模型的潜在动态中。首先,我们介绍变分稀疏门控(VSG),一个潜在的动态模型,更新其特征尺寸稀疏通过随机二进制门。此外,我们提出了一个简化的架构简单变分稀疏门控(SVSG),删除了以前的模型的确定性路径,从而产生一个完全随机的过渡函数,利用VSG机制。我们评估了两个模型架构的BringBackstrike(BBS)环境中,具有大量的移动对象和部分可观测性,表现出明显的改进,比以前的模型。

英文摘要:

Learning world models from their sensory inputs enables agents to plan for actions by imagining their future outcomes. World models have previously been shown to improve sample-efficiency in simulated environments with few objects, but have not yet been applied successfully to environments with many objects. In environments with many objects, often only a small number of them are moving or interacting at the same time. In this paper, we investigate integrating this inductive bias of sparse interactions into the latent dynamics of world models trained from pixels. First, we introduce Variational Sparse Gating (VSG), a latent dynamics model that updates its feature dimensions sparsely through stochastic binary gates. Moreover, we propose a simplified architecture Simple Variational Sparse Gating (SVSG) that removes the deterministic pathway of previous models, resulting in a fully stochastic transition function that leverages the VSG mechanism. We evaluate the two model architectures in the BringBackShapes (BBS) environment that features a large number of moving objects and partial observability, demonstrating clear improvements over prior models.

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。

↑