Algorithmic Systems: Mission Statement

Computer science is the science of studying how things can be automated. When automating intelligent behaviour, modeling plays a central role as an attempt to formalize the properties of processes characterizing learning, inference and intervention (actions). Due to the uncertainty and incompleteness of available information in application domains of computer science - the artificial intelligence and machine learning domains in particular - such models are commonly based on probabilities. The aim of our research is fundamental understanding and development of computationally efficient probabilistic and information-theoretic modeling techniques, and their multi-disciplinary applications from engineering to sciences.

The work has a strong basic research component, being at the intersection of computer science, information theory and mathematical statistics. The results of this methodological work are applied both in science and industrial applications resulting in advanced prototypes and fully fielded applications. The recent applied research areas include industrial collaboration related to user modeling, next generation information search and signal denoising, as well as multi-disciplinary applications of probabilistic modeling techniques in social sciences, medicine, historical studies, biology and neuroinformatics.

The main research challenges addressed are:

• Theoretical frameworks for probabilistic modeling. Develop computationally efficient, general-purpose methods for probabilistic modeling, focusing on issues related to model selection, parameter estimation and inference.
• Models for intelligent information access. In many modern information networks (like the Internet and various sensor networks), the data can not be found in a well structured format, and accessing the information may be a problem even if the information is in principle available. The goal is to apply probabilistic models for performing information retrieval tasks in this type of environments.
• Models for image analysis. Develop probabilistic methods for processing two- or three-dimensional measurement data, with applications in data visualization and denoising, and in the analysis of brain imaging data.
• Models for information processing in the visual system of the brain. Develop probabilistic computational models on how vision is possible in the brain, and to generalize these principles to different domains of computational neuroscience and computational intelligence.
• Models for probabilistic data fusion. Develop probabilistic methods for combining inputs originating from heterogeneous data sources.
• User modeling. Develop probabilistic modeling methods for personalization, profiling and segmentation.