本文提出了一種名為「快速莫比烏斯變換」的新方法,利用代數性質有效計算信息分解,克服了傳統方法的計算瓶頸,並展示了其在神經科學和音樂分析中的應用。
This paper introduces the fast Möbius transform, a novel method leveraging algebraic properties of the redundancy lattice to efficiently compute information decomposition, enabling previously intractable analyses of synergy, redundancy, and unique information in complex systems with up to five variables.
Practical methodology using machine learning for information decomposition in complex systems.