Tunnelling Through Time Series: A Probabilistic Visibility Graph for Local and Global Pattern Discovery

TL;DR

The paper introduces the Probabilistic Visibility Graph (PVG), a novel method inspired by quantum tunnelling, to better capture both local and global patterns in complex time series data, especially long-range dependencies.

Contribution

It extends the classical Visibility Graph by incorporating probabilistic connections, enabling more effective analysis of complex signals with interacting temporal scales.

Findings

PVG reveals distinct network properties between rest and anesthesia states.

Rest state shows stronger small-world and scale-free network features.

PVG provides new insights into neural dynamics and complex signal analysis.

Abstract

The growing availability of high-resolution, long-term time series data has highlighted the need for methods capable of capturing both local and global patterns. To address this, we introduce the Probabilistic Visibility Graph (PVG), a novel approach inspired by the quantum tunnelling phenomenon. The PVG extends the classical Visibility Graph (VG) by introducing probabilistic connections between time points that are obstructed in the VG due to intermediate values. We demonstrate the PVG's effectiveness in capturing long-range dependencies through simulations of amplitude-modulated signals and analysis of electrocorticography (ECoG) data under rest and anesthesia conditions. Key results show that the PVG presents distinct network properties between rest and anesthesia, with rest exhibiting stronger small-worldness and scale-free behavior, reflecting a hub-dominated, centralized connectivity structure, compared to anesthesia. These findings highlight the PVG's potential for analyzing complex signals with interacting temporal scales, offering new insights into neural dynamics and other real-world phenomena.