Task-adaptive physical reservoir computing.
Task-adaptive physical reservoir computing.
复制标题
DOI：
10.1038/s41563-023-01698-8
复制
发表时间：
2024-01
期刊：
影响因子：
41.2
通讯作者：
Kurebayashi, Hidekazu
中科院分区：
文献类型：
--
作者：
Lee, Oscar;Wei, Tianyi;Stenning, Kilian D;Gartside, Jack C;Prestwood, Dan;Seki, Shinichiro;Aqeel, Aisha;Karube, Kosuke;Kanazawa, Naoya;Taguchi, Yasujiro;Back, Christian;Tokura, Yoshinori;Branford, Will R;Kurebayashi, Hidekazu
水库计算是一种神经形态架构，可以为机器学习不断增长的能源成本提供可行的解决方案。在基于软件的机器学习中，计算性能可以通过调整超参数来重新配置以适应不同的计算任务。在利用物理系统的非线性和历史相关响应进行数据处理的“物理”储层计算方案中缺少这一关键功能。在这里，我们克服了这个问题的“任务自适应”的方法来物理水库计算。通过利用几何相空间来重新配置关键的储层属性，我们优化了不同任务集的计算性能。我们使用的旋波光谱的手性磁体Cu2OSeO3主机skyrmion，锥形和螺旋磁相，提供按需访问不同的计算水库响应。任务自适应的方法是适用于各种各样的物理系统，我们显示在其他手性磁体通过以上（和附近）的室温演示Co8.5Zn8.5Mn3（和FeGe）。当前物理神经形态计算面临着如何重新配置系统的关键物理动力学以适应计算性能以匹配各种任务的关键挑战。在这里，作者提出了一个任务自适应的方法，物理神经形态计算的基础上按需控制的计算性能，使用各种磁相位的手性磁铁。
Reservoir computing is a neuromorphic architecture that may offer viable solutions to the growing energy costs of machine learning. In software-based machine learning, computing performance can be readily reconfigured to suit different computational tasks by tuning hyperparameters. This critical functionality is missing in ‘physical’ reservoir computing schemes that exploit nonlinear and history-dependent responses of physical systems for data processing. Here we overcome this issue with a ‘task-adaptive’ approach to physical reservoir computing. By leveraging a thermodynamical phase space to reconfigure key reservoir properties, we optimize computational performance across a diverse task set. We use the spin-wave spectra of the chiral magnet Cu2OSeO3 that hosts skyrmion, conical and helical magnetic phases, providing on-demand access to different computational reservoir responses. The task-adaptive approach is applicable to a wide variety of physical systems, which we show in other chiral magnets via above (and near) room-temperature demonstrations in Co8.5Zn8.5Mn3 (and FeGe). Current physical neuromorphic computing faces critical challenges of how to reconfigure key physical dynamics of a system to adapt computational performance to match a diverse range of tasks. Here the authors present a task-adaptive approach to physical neuromorphic computing based on on-demand control of computing performance using various magnetic phases of chiral magnets.