Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

The insertion method to invert the signature of a path

About

The signature is a representation of a path as an infinite sequence of its iterated integrals. Under certain assumptions, the signature characterizes the path, up to translation and reparameterization. Therefore, a crucial question of interest is the development of efficient algorithms to invert the signature, i.e., to reconstruct the path from the information of its (truncated) signature. In this article, we study the insertion procedure, originally introduced by Chang and Lyons (2019), from both a theoretical and a practical point of view. After describing our version of the method, we give its rate of convergence for piecewise linear paths, accompanied by an implementation in Pytorch. The algorithm is parallelized, meaning that it is very efficient at inverting a batch of signatures simultaneously. Its performance is illustrated with both real-world and simulated examples.

Adeline Fermanian, Jiawei Chang, Terry Lyons, G\'erard Biau• 2023

Related benchmarks

TaskDatasetResultRank
Signature fidelityS&P 500 2023–2025 (Out-of-sample)
Relative MSE0.03
7
Signature fidelitylog-GBM
Relative MSE0.99
7
Signature fidelitylog-fBM
Relative MSE1.01
7
Signature fidelityou
Relative MSE1.13
7
Signature fidelityS&P 500 2009–2022 (In-sample)
Relative MSE0.03
7
Parameter Estimationlog-GBM processes (synthetic)
Sigma Error1.93
7
Parameter Estimationlog-fBM synthetic processes
Sigma Error1.91
7
Parameter EstimationOU (Ornstein-Uhlenbeck) synthetic processes
Sigma Error1.9
7
Stylized-fact diagnosticsS&P 500 2009–2022 (In-sample)
Volatility MAE0.13
7
Stylized-fact diagnosticsS&P 500 2023–2025 (Out-of-sample)
Volatility (MAE)0.11
7
Showing 10 of 10 rows

Other info

Follow for update