Universitšt Karlsruhe
Dr. Martin Spott
IPD Goos
BTexact Technologies, MLB1 PP 12, Adastral Park
IP53RE Ipswich
Email martin dot spott at bt dot com

Sprechzeiten: Fragen Sie mich!


The key part of my research work is covered by approximate reasoning with fuzzy systems. The investigations are based on the fact that humans are able to make reasonable decisions though their domain knowledge is often only vague or uncertain. The aim of my work is to find a flexible representation of vague and/or uncertain knowledge and an efficient way to reason on the basis of the knowledge. I especially focus on rule based systems.

The approximate reasoning techniques are applied in the DFG-project Fynesse (FuzzY-NEuro-SyStEm) that deals with autonomous learning of strategies for optimal sequential decisions in large domains. The important questions for the fuzzy part are

  1. the integration of imperfect prior knowledge about the strategy into a reinforcement learning approach,
  2. the interpretation of a learned strategy in terms of fuzzy rules (rule extraction).


Reasoning With Vague And Uncertain Terms

Betreute Studien- und Diplomarbeiten

Integration eines adaptiven Strategieelementes in die hybride Reglerarchitektur Fynesse (abgeschlossen)
Das Konzept der Sicherheit bei adaptiven, unscharfen Reglern (abgeschlossen)


Spott, Efficient Reasoning With Fuzzy Words
Weisbrod, Spott, Conditional constraints, implication based rules, and possibilistic rule bases: Are they any good?
Schoknecht, Spott, Riedmiller, Fynesse: A new architecture for sequential decision problems
Schoknecht, Spott, Riedmiller, Design of selflearning controllers using Fynesse
Riedmiller, Spott, Weisbrod, Fynesse: A hybrid architecture for selflearning control
Schoknecht, Spott, Liekweg, Riedmiller, Search Space Reduction for Strategy Learning in Sequential Decision Processes
Spott, Schoknecht, Riedmiller, Approaches for the integration of a priori knowledge into an autonomously learning control architecture
Spott, A theory of possibility distributions
Gaul, Riedmiller, Schoknecht, Spott, Fuzzy-Neuro-Controlled Verified Instruction Scheduler
Schoknecht, Spott, Riedmiller, Goos, Menzel, Selbständiges Lernen von Regelungen mit Fynesse
Spott, Using classic approximation techniques for approximate reasoning
Spott, Riedmiller, Improving a priori control knowledge by reinforcement learning
Goos, Spott, Weisbrod, Riedmiller, Interpretation und Adaption unscharfer Relationen innerhalb einer hybriden selbstlernenden Steuerungsarchitektur
Riedmiller, Spott, Weisbrod, First Results on the Application of the Fynesse Control Architecture
Bolten, Spott, Fuzzy Rule Extraction from Fuzzy Relations
Spott, Weisbrod, A new approach to the adaptation of fuzzy relations
Spott, Unscharfes Schließen mit komplexen Wissensbasen