Gaussian Processes to Study the Joint Dynamics of Networks and Time Series
Event details
| Date | 22.06.2016 |
| Hour | 08:30 › 10:30 |
| Speaker | Victor Kristof |
| Location | |
| Category | Conferences - Seminars |
EDIC Candidacy Exam
Exam President: Prof. Wulfram Gerstner
Thesis Director: Prof. Patrick Thiran
Thesis Co-director: Prof. Matthias Grossglauser
Co-examiner: Prof. Pascal Frossard
Background papers
Recurrent Gaussian Processes, by C. L. C. Mattos et al.
COEVOLVE: A Joint Point Process Model for Information Diffusion and Network Co-evolution, by M.Farajtabar et al.
Spatial and Spatio-Temporal Log-Gaussian Cox Processes: Extending the Geostatistical Paradigm, by P. J. Diggle et al.
Abstract
Gaussian processes are a class of statistical models whose popularity in Machine Learning has been increasing dramatically in recent years. They provide a powerful framework for regression and classification, while exhibiting great flexibility and interpretability. We describe in this work how they can be extended into a recurrent structure to learn dynamical patterns and how they can model spatio-temporal point processes. This will lead us to the study of co-evolutionary network and time series, and how Gaussian processes can be exploited to improve the existing joint dynamics models.
Exam President: Prof. Wulfram Gerstner
Thesis Director: Prof. Patrick Thiran
Thesis Co-director: Prof. Matthias Grossglauser
Co-examiner: Prof. Pascal Frossard
Background papers
Recurrent Gaussian Processes, by C. L. C. Mattos et al.
COEVOLVE: A Joint Point Process Model for Information Diffusion and Network Co-evolution, by M.Farajtabar et al.
Spatial and Spatio-Temporal Log-Gaussian Cox Processes: Extending the Geostatistical Paradigm, by P. J. Diggle et al.
Abstract
Gaussian processes are a class of statistical models whose popularity in Machine Learning has been increasing dramatically in recent years. They provide a powerful framework for regression and classification, while exhibiting great flexibility and interpretability. We describe in this work how they can be extended into a recurrent structure to learn dynamical patterns and how they can model spatio-temporal point processes. This will lead us to the study of co-evolutionary network and time series, and how Gaussian processes can be exploited to improve the existing joint dynamics models.
Practical information
- General public
- Free
Contact
- Cecilia Chapuis EDIC