Bayesian Statistical Methods
Bayesian methodology group lead by Aki Vehtari conducts research in the field of modern computational science and develops generic computational statistical methods. These include fast approximative inference methods as well as model assessment and selection methods for complex hierarchical, non-parametric, graphical, dynamic and spatio-temporal models. The developed methods can handle larger datasets using more elaborate and computationally intensive models than before while keeping the computation time reasonable to aid researchers in other scientific fields. The methods are applied to challenging scientific problems in, e.g., brain signal analysis (MEG, fMRI), epidemiology, genetics, bioinformatics, medicine, tomography, animal population research, audio signal processing, and tracking.
Group members
Senior reseachers
- Aki Vehtari (group leader)
- Simo Särkkä
Post Doc Reseachers
Doctoral students
Research students
- Mudassar Abbas
- Sakari Cajanus
- Olli-Pekka Koistinen
- Jani Kuula
- Sasu Mäkelä
- Arno Solin
- Saara Suikkanen
- Ville Tolvanen
- Ville Väänänen
Downloads
Current projects
- Bayesian Model Assessment and Selection Using Expected Utilities
- Gaussian processes in Bayesian modelling of complex systems
- Bayesian Estimation of Dynamic Systems
- Statistical Brain Signal Analysis
- Computational Health
- Public health Genomics to Practice in Cardiovasculaur diseases
- Large predator population structure and size estimation
Past projects
- Sequential Monte Carlo Methods in Multiple Target Tracking
- New Analysis Methods for Healthcare Data
- Dynamic state estimation and reliability analysis of sensor networks
- Bayesian Methods for Neural Networks
- Bayesian Modeling of the Concrete Quality
- Building Spatial Choice Models from Aggregate Data
- Neural Networks in Electrical Impedance Tomography
- Optimizing the Web Cache
- Prediction of Steel Jominy Curves
- Probability Density Model for the Self-Organizing Map
