VVZ API is not affiliated with ETH Zurich. Data might be outdated or incorrect. Please view the official ETHZ Vorlesungsverzeichnis for binding information.

401-3660-25L 4 Credits MSC D-MATH

Rough Path Theory and Machine Learning

Lecturers & Examiners: Dr. Alexander Schell
Number of participants limited to 11.
VVZ CR n/a

Last Updated: 2026-06-01 11:33:48

Content

This course examines how rough path theory, particularly its signature-based methodologies, can enable and enhance the machine learning and statistical modelling of irregular and multidimensional time-dependent data. Starting with an introduction to rough integration, controlled differential equations and path signatures, the course explores the interplay of these concepts with modern sequence-processing architectures (e.g., RNNs, transformers, and neural SDE/CDE frameworks), and investigates how signature transforms and related rough-path techniques have driven algorithmic advances that demonstrate superior performance and interpretability. Students will critically engage with recent and current research through presentations and discussions, gaining foundational insights to the rapidly expanding interface between stochastic analysis and machine learning. Prerequisites include measure-theoretic probability, familiarity with time series or stochastic processes, and a basic knowledge of machine learning and statistics.

General Information

Language
English
Levels
MSC

Examination

Type
ungraded semester performance

Registration & Places

Limited places (Special selection)
Signup End
08.02.2025
Priority: Registration for the course unit is only possible for the primary target group

Course Components

Type Title Time & Place Hours
seminar Rough Path Theory and Machine Learning
Fully booked.
  • Wed 18:15-20:00 (ML J 34.3)
2 h weekly

Offered In

      • Seminare (ZUR BEACHTUNG: Damit die Zuteilung der verfügbaren Seminarplätze sich nicht primär auf den Zeitpunkt des Einschreibens in die Warteliste stützen muss, haben die meisten Mathematik-Seminare ein spezielles Auswahlverfahren. Eine direkte Belegung in myStudies ist dann nicht möglich, alle kommen zuerst auf die Warteliste. Ausserdem gilt: Die Auswahl an Mathematik-Seminaren wird auf 1 Seminar pro Semester beschränkt. Falls Sie in diesem Semester 2 Seminare absolvieren müssen, melden Sie sich bitte beim Studiensekretariat (E-Mail: ). Beachten Sie auch die Lerneinheit 401-0002-99L Generic Seminar - Second Priority / Third Priority.)
  • Quantitative Finance Master (siehe Studierende im Joint Degree Master-Studiengang "Quantitative Finance" müssen Module der UZH direkt an der UZH buchen. Die entsprechenden Module sind hier nicht aufgelistet.)