Forschung

Forschungsschwerpunkte

Die theoretischen und angewandten Forschungen auf dem Gebiet der wissensbasierten Systeme haben zum Ziel, wissensintensive Probleme mithilfe von Computern zu lösen. Tätigkeiten, die bislang den Einsatz menschlicher Experten erforderlich machen, sollen durch die Entwicklung „intelligenter“ Programme zu weiten Teilen automatisch durchgeführt werden können. Entsprechend vielfältig sind die Forschungsgebiete:

In den Forschungsvorhaben des Bereichs Logik entwickeln wir effiziente Schlussfolgerungsverfahren und untersuchen damit zusammenhängende Entscheidungsprobleme (Satisfiability), speziell für die Aussagenlogik und die Quantifizierte Boolesche Logik.

Wir beschäftigen uns außerdem mit der Integration wissensbasierter Verfahren in Methoden der Ingenieurwissenschaften. Hierzu zählt die Automatisierung von Modellbildungs-, Diagnose- und Entwurfsaufgaben für technische Systeme.

 

Aktuelle Projekte

  • MinUnsat - Minimal unerfüllbare Formeln
  • QBF - Quantified Boolean Formulas
 

Abgeschlossene Projekte

 

Konferenzen

 

Spin-Offs

Einige Projekte der AG Wissensbasierte Systeme haben inzwischen zu Ausgründungen (Spin-Offs) durch wissenschaftliche Mitarbeiter der Arbeitsgruppe geführt. Für nähere Informationen zu den betreffenden Projekten wird deshalb im Folgenden auf die entsprechenden Firmenseiten verwiesen.

Art Systems Logo Art Systems Software GmbH
Klarinet Logo Klarinet Software GmbH

 

 

Research Group „Knowledge-Based Systems”

Research Topics

Main Research Areas

Multiagent Systems and Machine Learning

Our multiagent system research concentrates on the coordination of a team of agents and the learning of both emergent and single group-independent behavior. We assume that the knowledge and capabilities of agents are restricted due to resource limitations. Our activities focus on partitioning the problems and hierarchical organization of multiple agents as well as acting in highly dynamic environments. Techniques taken from evolutionary computation and swarm intelligence contributed to powerful learning approaches in this research context, i.e. the evolution of sets of rules for reactive agent control.

Logic

Mathematical logic is a fundamental formalism in computational complexity theory and AI. Many problems are closely connected to the expression of logical formulas and their feasibility. Our research activities concentrate on the efficiency of algorithms for the decision problem ‘satisfiability’ (SAT) in propositional logic and related questions, e.g., the structure of minimal unsatisfiable formulas or the representation of formulas as circuits. A recent research focus is on Quantified Boolean Formulas (QBF). We investigate short encodings of propositional formulas as QBFs, satisfiability and equivalence models for QBFs, especially for QHORN, and consider applications such as model checking.

Modeling, Analysis, and Synthesis of Technical Systems

The automatic synthesis of technical systems forms the core of many problems from the field of configuration, design, and feasibility analysis. To master it, questions from the field of economics, engineering sciences, intelligent search, and knowledge processing need to be answered. In various research projects and industrial ventures we have investigated questions on formal models of technical system design, model generation with design graph grammars, and the learning of optimum design decisions. Our research led to the development of remarkable tools that support fluidic system design and world-class software in education and uncausal simulation (FluidSIM).

 

Current Projects

Multiagent Systems and Machine Learning

Publications on Multiagent Systems and Machine Learning

Logic

  • QBF - Quantified Boolean Formulas
  • MinUnsat - Minimal unsatisfiable formulas
Publications on Logic

Modeling, Analysis, and Synthesis of Technical Systems

  • Analysis and Diagnosis of Distributed Systems
Publications on Modeling, Analysis, and Synthesis of Technical Systems

 

Closed Projects

Multiagent Systems and Machine Learning

Logic

Modeling, analysis, and synthesis of technical systems

  • DejaVue - Diagnosis through Model Compilation
  • AutoWed - Automatic Design of Wave Digital Structures
  • Modelica
  • Case-Based Design
  • FluidSIM - Simulation of Fluidic Systems
  • OFT - Optimized Configuration of Controlled Fluidic Systems with Knowledge-Based Techniques
  • LANeCo - Local Area Network Configuration
  • WiMom - Knowledge-based Modeling and Optimization of Mixing Processes
  • Wita - Wissensbasierte Inbetriebnahme technischer Anlagen
  • Multimedia Authoring System
  • Insafa - Information integrated Sensor/Actuator-Systems for Fluidic Drives

 

Conferences

 

Spin-Offs

Art Systems Logo Art Systems Software GmbH
Klarinet Logo Klarinet Software GmbH

Impressum | Webmaster | Letzte Änderungen am : 18.10.2017