Research
Alper Sarıtaş works on probabilistic inference and uncertainty quantification at the Institute of Industrial Information Technology (IIIT), Karlsruhe Institute of Technology (KIT). His work connects mathematical models, sensor data, and research software.
Current directions
- Gaussian mixture inference: representing multimodal uncertainty and propagating it through probabilistic graphical models.
- Hybrid models: combining discrete and continuous variables in learned networks and factor graphs.
- Mixture reduction: controlling model size while measuring the approximation introduced by reduction.
- Research software: turning inference methods into reusable, testable components.
Work and sources
Fusion GMBP brings these directions together in a software family. Its public KIT GitLab project describes the inference core and identifies Alper Saritas as its implementer.
The publication page records the MFI 2026 paper on neural-network-based Gaussian mixture reduction, with links to the official conference program and KIT announcement.