Structure in patterns in ordered datasets with applications in astrophysics, neuroscience and archaeology
The project will delve into classic and powerful techniques related to complex systems, such as graph theory, probability theory and statistics, as well as modern and promising ones such as Network Science, machine learning and data mining techniques and tools. The techniques will be explored together with application based specialised knowledge in order to provide a rigorous theoretical and algorithmic framework for the identification of semantics networks applicable to the data from different disciplines, such as Astrophysics, Neuroscience and Archaeology. For the astrophysics application, the project will address problems related to galaxy evolution and cosmology. For the astrophysics application, we wish to classify stages of processing and identifying transitions of regional activations of the brain system, using EEG and MEG data and the slow hemodynamic measures (PET and fMRI). The archaeology work involves investigation of possible relation of the evolution of ancient pottery images to how the visual system analyses information.
Objectives
- To apply Bayesian inferencing and graph-theory based methodologies to facilitate the development of a powerful theoretical framework for the identification of semantics networks applicable to the data from different disciplines;
- Development of efficient algorithms for automatically extracting the semantics networks from the data and the domain knowledge and their implementation in a unified software tool;
- Application and adaptation of the software tool for the analysis of data in astrophysics, neuroscience and archaeology.
Expected Results
- Understanding the minimal common mathematical structures that can be used to derive identification of semantics networks that transform the data in each discipline into meaningful descriptions;
- Translating the knowledge in (1) into software tools;
- Apply these tools to the targeted disciplines.