Seminar announcement

time: Thursday, 21 September 2023, 12h15-13h15

room: Campus Solvay, R42.2.113

speaker: François Kamper (EPFL)

title: Advances in Time Series Analysis via Optimal Transport

abstract:
Optimal transport (OT) is becoming increasingly popular in machine learning and
its rich theoretical basis has inspired advancements in a variety of domains,
including time series analysis. In this talk I will discuss some recent
advances in time series analysis domains inspired by ideas from OT. More
specifically, I will discuss how optimal transport maps can be used to estimate
distributional auto-regressive models, how Wasserstein barycenters can be used
in state-space modeling, and how optimal transport can be used in time series
domain adaption and forecasting problems.