1.

Systems identification from measured data

The principal investigator manager was appointed the lecturer in the course Systems modelling from the measured data in the autumn semester 2009/2010 for the MS programme at the Faculty of transportation sciences, Czech Technical University in Prague. The course dealt with system identification methods , including modelling with Gaussian process models.

B.05 Guest lecturer at an institute/university

COBISS.SI-ID: 23079719

2.

Associate editor of the international journal Evolving systems, Springer

The principal investigator become an associate editor and a member of editorial board of the scientific journal Evolving systems, Springer, in year 2012, and valid since start of 2013. The scientific journal is expected to gain SCI impact factor since year 2012 and is an important journal in the field of artificial evolving systems.

C.04 Editorial board of an international magazine

3.

Control algorithms based of Gaussian process models

A detailed state-of-the-art survey of control algorithms based on Gaussian process models is given in the paper. Beside the survey of control algorithms also trends, challenges and future opportunities are discussed in the paper. The contribution was prepared for the invited lecture at the international conference from the topic of control, communications and computing.

B.04 Guest lecture

COBISS.SI-ID: 25087783

4.

Prognosis of gear health using Gauissian process model

Application of Guassian process models are demonstrated for the prognosis of gear health in a gear-box. This application is meant for the diagnostics of rotational tool machinery for prediction of gear failure and for enabling replacement of the failed part in time.

B.03 Paper at an international scientific conference

COBISS.SI-ID: 24721447

5.

Accelerated learning of Gaussian process models

An application of algorithm for the parallel processing of Gaussian process models' training on a multicore graphic unit for personal computer is shown in the paper. This training is accelerated a few times in comparison with conventional processor computing due to multicore use and algorithm adaptation which improves efficiency and applicability of the training algorithm.