COMPUTATIONAL KNOWLEDGE DISCOVERY IN SCIENTIFIC APPLICATIONS

The proposed program is strongly multidisciplinary. Computer science is the integrating element for all proposed projects but at the same time each project has its own theoretical and practical goals in its target research field. Each project is a research entity able to define and realize own goals independently of other projects. It also means that most of the work will be done independently and in parallel. The gain of being a part of the same program is at the methodological and the application level. There are four projects that dominantly develop methodology and three projects that dominantly use methodology. The application projects will gain by using novel computer science methods developed for their specific needs and methodological projects will gain by having chance to evaluate their algorithms on real scientific tasks. Besides that, stronger cooperation among methodological projects working on complementary tasks is expected.

The distribution of tasks will be as follows: There is one project developing mainly the technology of measurement and analysis methods of real-life signals, time series, complex systems and dynamical processes. It presents program connection to scientific instrumentation and modern electronic measurement hardware and software, enabling complete information flow from signal measurement to data collection and direct connection of knowledge technologies with scientific experiments. The research topics besides high resolution timing measurements are also complex system modeling procedures and indexing of static databases. The task of the second project is development of machine learning algorithms and their application in intelligent data analysis and very different knowledge discovery tasks. Connected with this is also implementation of decision support and knowledge representation tasks. The project will share interests with the third methodological project targeted at implementation of computational intelligence in system biology domains. The significant difference is that in contrast to the former, representing typical computer science research, the later will develop algorithms especially for biological and biochemical applications in a multidisciplinary collaboration with domain experts. Its research topics include procedures for functional classification of proteins and development of machine learning based models in the drug discovery processes. Some of very difficult knowledge discovery tasks from data sequences can be transformed into image analysis problems. The task of the last methodological project is extraction of complex features from images. The intention is to enable application of standard machine learning techniques also for the image analysis and their automated classification. The research topics are implementations of signal processing techniques, including applications of multiresolution wavelet analysis.

At the application side, in the field of bioinformatics the tasks can be divided into a few groups. The first is acquisition, storage and specific structuring of selected biological data like DNA and protein sequences of genes. Follows production, application and further development of algorithms and programs for the prediction of protein functions, the determination of biosynthetic pathways, and modeling of homologous recombination process. The third group is in silico annotation of gene-clusters present in different genomes by the individual similarity searches, profile searches, detailed analysis of individual genes, analysis of individual gene transcription and the determination of novel genetics and/or chemistry. We will work also on experimental verification of the in silico hypothesis by cloning, homologous recombination and analysis of novel gene-clusters, expression of their products in surrogate host by the biosynthesis of recombinant strains and the analysis of chemical structures of their products.

->hrvatski

The program is supported by Croatian Ministry of Science Education and Sports
contract number: 0982560
starting date: July 2007

Participating projects

Intelligent image features extraction in knowledge discovery systems - University of Zagreb, Faculty of Electrical Engineering and Computing

Real life data measurement and characterization - Rudjer Boskovic Institute

Machine learning algorithms and their applications - Rudjer Boskovic Institute

Computational Intelligence Methods in Measurement Systems - Rudjer Boskovic Institute

Predictive models in health care - University of Zagreb, Medical School, Andrija Stampar School of Public Health

Documents

Scientific hypothesis
Collaboration plan
First program meeting - IRB 03/09/2007 presentation