Volterra signature offers new feature map for time series learning

Researchers have introduced the Volterra signature (VSig) as a novel feature representation for time series data, aiming to improve interpretability and long-horizon training for non-Markovian systems. This method leverages tensor algebra and the Volterra-Chen identity to provide theoretical guarantees, including identifiability and universal approximation. The Volterra signature offers a computationally tractable approach, enabling the use of the kernel trick and demonstrating improved performance on dynamic learning tasks compared to classical path signatures. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Introduces a new mathematical framework for feature representation in time series, potentially improving model performance and interpretability in AI applications.

RANK_REASON The cluster contains an academic paper detailing a new method for time series analysis. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv stat.ML →

• Luca Pelizzari
• Volterra signature

• paper
• other