Module responsible: Massimiliano Ruocco
The module focuses on modern machine learning methods for time series analysis.
The course covers four main topics:
- Generative Models for Time Series
- Forecasting for Time Series
- Classification for Time Series
- Anomaly Detection for Time Series
The course material will consist mainly of selected scientific papers.
The module is organized as four seminar sessions. Each student will be assigned one scientific paper and present it during the corresponding session.
The presentation should be approximately 12–15 minutes and should cover:
- Motivation and problem
- Main method
- Experimental results
- Conclusions and critical discussion
Each presentation will be followed by questions and discussion.
| Session | Topic | Date | Papers |
|---|---|---|---|
| 1 | Generative Models for Time Series | TBA | TBA |
| 2 | Forecasting for Time Series | TBA | TBA |
| 3 | Classification for Time Series | TBA | TBA |
| 4 | Anomaly Detection for Time Series | TBA | TBA |
The papers and student assignments will be announced soon.
Students must:
- present one assigned scientific paper;
- attend all four seminar sessions;
- actively participate in the discussions and ask questions.
The module concludes with a short individual oral exam.
Each student will be examined on the three topics corresponding to the sessions in which they did not present a paper.
For example, a student presenting in Generative Models for Time Series will be examined on Forecasting, Classification, and Anomaly Detection.
The oral exam will focus on the main concepts, methods, and challenges discussed during the seminar sessions.