arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.10211cs.LGcs.ROeess.SP
收藏

用于连续序列到序列建模的分层状态空间模型

Hierarchical State Space Models for Continuous Sequence-to-Sequence Modeling

Carnegie Mellon University(卡内基梅隆大学);FAIR, Meta(Meta FAIR实验室);New York University(纽约大学)

Raunaq Bhirangi, Chenyu Wang, Venkatesh Pattabiraman, Carmel Majidi, Abhinav Gupta, Tess Hellebrekers, Lerrel Pinto

AI总结:

本文提出 HiSS,通过堆叠结构化状态空间模型形成时间层次,以解决真实传感器非线性、漂移和小样本连续序列预测问题,并在六个数据集上显著优于现有序列模型。

摘要:

从原始感官数据序列中进行推理是从医疗设备到机器人技术等众多领域中普遍存在的问题。这些问题通常涉及使用长序列的原始传感器数据(例如磁力计、压敏电阻)来预测期望物理量的序列(例如力、惯性测量值)。尽管经典方法对于局部线性预测问题非常强大,但在使用真实世界传感器时往往表现不佳。这些传感器通常是非线性的,会受到无关变量(例如振动)的影响,并表现出依赖于数据的漂移。对于许多问题而言,由于获取真值标签需要昂贵的设备,预测任务因标注数据集较小而进一步恶化。在本研究中,我们提出了 Hierarchical State-Space Models(HiSS,分层状态空间模型),这是一种概念简单、用于连续序列预测的新技术。HiSS 将结构化状态空间模型相互堆叠,以构建时间层次结构。在六个真实世界传感器数据集上,从基于触觉的状态预测到基于加速度计的惯性测量,HiSS 在 MSE 上比 causal Transformers、LSTMs、S4 和 Mamba 等最先进的序列模型至少高出 23% 的性能。我们的实验进一步表明,HiSS 能够高效扩展到较小的数据集,并且与现有数据过滤技术兼容。代码、数据集和视频可在 https://hiss-csp.github.io 获取。

英文摘要:

Reasoning from sequences of raw sensory data is a ubiquitous problem across fields ranging from medical devices to robotics. These problems often involve using long sequences of raw sensor data (e.g. magnetometers, piezoresistors) to predict sequences of desirable physical quantities (e.g. force, inertial measurements). While classical approaches are powerful for locally-linear prediction problems, they often fall short when using real-world sensors. These sensors are typically non-linear, are affected by extraneous variables (e.g. vibration), and exhibit data-dependent drift. For many problems, the prediction task is exacerbated by small labeled datasets since obtaining ground-truth labels requires expensive equipment. In this work, we present Hierarchical State-Space Models (HiSS), a conceptually simple, new technique for continuous sequential prediction. HiSS stacks structured state-space models on top of each other to create a temporal hierarchy. Across six real-world sensor datasets, from tactile-based state prediction to accelerometer-based inertial measurement, HiSS outperforms state-of-the-art sequence models such as causal Transformers, LSTMs, S4, and Mamba by at least 23% on MSE. Our experiments further indicate that HiSS demonstrates efficient scaling to smaller datasets and is compatible with existing data-filtering techniques. Code, datasets and videos can be found on https://hiss-csp.github.io.

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。

↑