Budapest University of Technology and Economics, Faculty of Electrical Engineering and Informatics

Dr. András Pataricza, full professor, Dept. of Measurement and Inf. Systems

Dr. Imre Kocsis, assistant professor, Dept. of Measurement and Inf. Systems

Computer-based systems are getting more and more complex at an increasing rate. Therefore, guaranteeing their extra-functional properties during design as well as in operation is becoming a more and more critical, too.

In addition to the increases in the number of components - most of which are typically integrated, not newly created - the number and complexity of various component-relationships is increasing, too. Thus, modeling contemporary systems for design and operation support requires the design- and runtime use of techniques which would be called „system identification" in a classic system theoretic context.

The course discusses the key techniques for connecting the realms of continuous metrics and discrete, qualitative models of IT systems and touches on the most important application areas.

Key techniques of Exploratory Data Analysis (EDA) and Confirmatory Data Analysis for deriving phenomenological models from observations.

Basics of hybrid modeling, discretization techniques and the continuous-discrete model transition. Basics of qualitative modeling, statistical validation of basic properties. Mathematical handling of qualitative models.

The basics of rough set theory, its applications in modeling for dependability assurance, when only partial information/knowledge is available.

Answer set programming and its application for approximative modeling and diagnosis. Model validation.

Representation of complex models as knowledge graphs, capturing a priori knowledge in knowledge graphs, consistency checking of observation-derived data and a priori knowledge.

Model identification case studies (dependable and resilient IT systems).

The role and application of empirically derived models in modern system design and operation. Key processes (e.g., modern capacity planning, chaos engineering, ...); the Digital Twin paradigm; knowledge bases of self-* processes (from Event-Condition-Action models to semantic reasoning support).

Outlook: protections against model errors, continuous model reassessment.

a. During the semester:

1 major mid-term homework assignment; for outstanding work, we waive the examination requirement and propose a term grade based on the homework (may require solving additional, noncompulsory homework tasks).

b. In the examination period: oral examination

c. Early exams before the examination period: none

Appointments shall be made with the lecturers on a case-by-case basis.

S. Akama, T. Murai, Y. Kudo: Reasoning with Rough Sets Logical Approaches to Granularity-Based Framework. Springer 2018.

M. S. Raza, U. Qamar: Understanding and Using Rough Set Based Feature Selection: Concepts, Techniques and Applications. Springer 2017.

D. Ciucci, T. Mihálydeák, Z. E. Csajbók: On Exactness, Definability and Vagueness in Partial Approximation Spaces. Technical Sciences 18(3), 2015, 203-212

F. Harmelen, V. Lifschitz, and B. Porter, "The Handbook of Knowledge Representation," Elsevier Science San Diego, USA, 2007.

R. Murch, Autonomic Computing. IBM Press, 2004.