TDT99 - Modern Machine Learning for Time Series Analysis
Modern Machine Learning for Time Series Analysis
Course overview [THE MODULE IS FULL]
- Responsible: Massimiliano Ruocco. For more information contact firstname.lastname@example.org (Contact the teacher if you want to enroll. ) or submit your request by using this link
- Work load: 3,75 SP
The course will focus on modern machine learning for the analysis of univariate and multivariate time series (i.e., anomaly detection, forecasting, classification, data imputation). In particular:
- Use of FNN, LSTM and CNN for time series modelling and forecasting.
- Attention mechanism in LSTM-based architecture for time series forecasting.
- The problem of small data and low-data regime in the time series domain:
- Unsupervised and Self-Supervised Learning for different time series related tasks
- Transfer Learning and Transformer architecture
- Few-Shot Learning and TS Classification in low-data regime
- Data Augmentation with Generative Adversarial Network (GAN)
- General use of GAN for time series analysis (i.e. Anomaly Detection, Data Imputation).
- (Deep) Echo State Network and Spiking Network for Time Series Analysis.
The course is structure in form of set of workshops where each group will present the papers in the course followed by group work and discussions. Information and list of papers will be available soon.
More information here.
The dates will be specified after the kick-off of the course.