arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1906.00302cs.LGstat.ML
收藏

从时间序列数据中学习低维状态嵌入和亚稳态簇

Learning low-dimensional state embeddings and metastable clusters from time series data

Carnegie Mellon University(卡内基梅隆大学);Princeton University(普林斯顿大学)

Yifan Sun, Yaqi Duan, Hao Gong, Mengdi Wang

摘要:

本文研究了如何从高维马尔可夫状态轨迹中找到紧致状态嵌入,其中转移核具有小的内在秩。在扩散映射的精神,我们提出了一个有效的方法来学习低维状态嵌入和捕捉过程的动态。这个想法也导致核整形方法更准确的非参数估计的过渡函数。状态嵌入可以用于将状态聚类到亚稳态集合中,从而识别慢动态。尖锐的统计误差界和误分类率证明。对一个模拟动力系统的实验表明,状态聚类方法确实揭示了亚稳态结构。我们还实验了在玩Atari游戏时由Deep-Q-Network层生成的时间序列。嵌入方法识别游戏状态是相似的,如果他们共享类似的未来事件,即使他们的原始数据是远远不同的。

英文摘要:

This paper studies how to find compact state embeddings from high-dimensional Markov state trajectories, where the transition kernel has a small intrinsic rank. In the spirit of diffusion map, we propose an efficient method for learning a low-dimensional state embedding and capturing the process's dynamics. This idea also leads to a kernel reshaping method for more accurate nonparametric estimation of the transition function. State embedding can be used to cluster states into metastable sets, thereby identifying the slow dynamics. Sharp statistical error bounds and misclassification rate are proved. Experiment on a simulated dynamical system shows that the state clustering method indeed reveals metastable structures. We also experiment with time series generated by layers of a Deep-Q-Network when playing an Atari game. The embedding method identifies game states to be similar if they share similar future events, even though their raw data are far different.

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。

↑