Projects / Programmes
Advanced machine learning methods for automated modelling of dynamic systems
| Code |
Science |
Field |
Subfield |
| 2.07.00 |
Engineering sciences and technologies |
Computer science and informatics |
|
| Code |
Science |
Field |
| P176 |
Natural sciences and mathematics |
Artificial intelligence |
machine learning, equation discovery, dynamic systems, automated modelling, ecological modelling, systems biology
Researchers (12)
Organisations (3)
Abstract
Dynamic systems, the state of which changes over time, are ubiquitous in both science and engineering. The task that we address in this proposal is the task of automated modeling of dynamic systems, i.e., the process of establishing models from observations and measurements of system behavior. Recent approaches to the task at hand, developed in the area of computational scientific discovery, suffer from a number of limitations: deterministic behavior of the modeled system is commonly assumed, where the state of the system in the future is completely determined by its present state; it is also assumed that a single model is valid over the entire lifetime of the modeled system; current methods employ standard approaches to parameter estimation, mainly based on gradient descent; finally, they suffer from computational complexity problems. The project will develop methods that overcome each of the four major limitations mentioned above: it will develop methods for learning probabilistic models of dynamic system behavior; methods for learning structurally dynamic models, whose structure and/or parameters change over time; improved methods for parameter estimation in the context of automated modeling of dynamic systems; and parallel algorithms for automated modeling. It will also evaluate the developed methods and demonstrate their utility by applying them to practical problems from the areas of ecological modeling (focusing mainly on aquatic ecosystems and waste-water treatment plants) and systems biology (learning metabolic and gene regulation networks).
Significance for science
The research carried out in this project is relevant for the development of several scientific disciplines. First and foremost, it contributed to the field of computer science (information technologies) broadly speaking, and the discipline of machine learning and the area of computational scientific discovery more specifically. The project moved well beyond the state of the art in that area, by developing methods for the automated construction of new types of models for dynamic systems, as well as improving upon key aspects of the construction of standard types of models.
The results of the project are also relevant for the scientific fields where we applied the developed methods, namely ecological modeling and systems biology. The topic of structurally dynamic modeling was originated in the field of ecological modeling, yet at the beginning of the project few (if any) approaches existed for the automated construction of such models. The emerging field of systems biology has a strong need for automated methods for modeling dynamic systems. By addressing the pressing needs in these application areas, the research carried out in this project has greatly facilitated their development.
Significance for the country
In the area of information technologies, it is conceivable that the developed methods for automated modeling of dynamic systems would give rise to a software product, which could be marketed to a potentially broad customer base in many disciplines (incl. various types of engineering). Slovenian industry has a strong IT sector that would be capable of turning the achieved research results into a product. The developed methods were used for automated modeling of agricultural ecosystems, including genetically modified crops. The learned models improved the understanding of the studied ecosystems and facilitated their management. The developed methods were also used in the area of systems biology, i.e., for automated modeling of various processes at the cellular level. The gene regulation pathways discovered in this fashion can be of use in the development of new therapies for the studied diseases, which could be relevant for the pharmaceutical industry in general and the Slovenian pharmaceutical industry in particular. Diseases/bacteria that were studied include Salmonella and Tuberculosis.
The project promoted the visibility of Slovenian researchers and Slovenia in the specific research areas considered (machine learning, ecological modeling, and systems biology) and the corresponding wider scientific areas (information technology, ecology, and biology). It also increased international cooperation of Slovenian researchers and facilitated the transfer of knowledge in the area of systems biology to Slovenia. A concrete example of this was the organization of the MLSB 2009 and 2010 workshops (Third and Fourth Workshop on Machine Learning in Systems Biology) in Ljubljana and Edinburgh. Systems biology is an emerging research area that will receive increased attention over the coming years: Slovenian researchers have limited expertise in this area, especially in learning models of the dynamics of processes in the cell. Finally, the project contributed to the development of researchers in its areas of interest, both at the PhD student and at the PostDoc level.
Most important scientific results
Annual report
2008,
2009,
final report,
complete report on dLib.si
Most important socioeconomically and culturally relevant results
Annual report
2008,
2009,
final report,
complete report on dLib.si
