Zijiu Lyu

Abstract

Feature extraction plays a pivotal role in modeling streaming data. Recent research has increasingly focused on the signature method and unitary Cartan development due to their characteristicity and universality, such as PCF-GAN. In this talk, we shall propose a formula that explicitly recovers the signature from unitary Cartan development. As a corollary, the linear extension of certain families of linear mappings separates signatures.

Our speaker

Zijiu Lyu is currently a research assistant at University College London (UCL), working with Prof. Hao Ni. His research focuses on theories of rough paths, especially the signature method.

Prior to this, Zijiu earned an MSc degree in Mathematical and Computational Finance (MCF) at the University of Oxford. His master’s dissertation researched the control problem of the Nash equilibrium in stochastic mean-field games.

To become a member of the Rough Path Interest Group, register here for free.