Research conducted by the Institute focuses on:
- parallel and distributed computing in cluster and grid environments,
- computer methods in economics,
- advanced multimedia and biometric techniques,
- development of engineering software.
The Institute is the coordinator and main participant of the CLUSTERIX project (National CLUSTER of LInuX Systems) -
a new generation cluster.
In the frame of cooperation with ID-IMAG Laboratory (France) research is being done in the field of parallel bioinformatics.
A new methods for large scale multiple sequence alignment with phylogeny reconstruction have been developed.
Another research activity is focused on simulation and multiobjective optimization of processes in production and economy.
This includes application of statistical methods, the fuzzy set theory (type-2 FS and hyper-FS) and fuzzy-interval mathematics to
computer aided decision making problems. The work being done covers, among other things, optimization of investments in a risk setting,
solution of logistic problems and aiding multicriteria decision making under uncertainty. Dedicated software is being developed for
solution of the abovementioned problems.
Scientific activity in the area of advanced multimedia and biometric techniques concerns methods of lossy compression of non-media
signals with on-line quality optimisation, and audio-video recognition of Polish speech. Moreover, an application of hidden Markov
models in face recognition problem is being investigated, as well as techniques of authentication based on face asymmetry or on emotions
expressed by face.
Several research projects deal with development of tools for numerical simulations of physical phenomena and technological processes.
In the context of numerical software the work concerns problems such as heat transfer or elastoplasticity, and is based on the adaptive
finite element method and on the object-oriented paradigm. A software package for parallel finite element computations has been created
using a distributed memory model (with message passing), as well as a hybrid model which combines message passing and multithreading models.
Currently the system is being ported to grid environment.