Reinforcement Learning: a Brief Overview

  • Michael M. Richter


Learning is considered as an essential aspect of intelligence. It takes usually place in some context where one learns from an environment. There are various forms of learning: How to learn and what to learn. Here we are concerned with learning of informal concepts. Informal concepts occur in many forms: Heuristics, personal judgements, utterances about taste etc. Such concepts provide to major difficulties:
  1. 1)

    Informal concepts do not have a precise definition and often not a definition at all.

  2. 2)

    Informal concepts are subjective and their interpretation depends on persons or groups of persons.

  3. 3)

    Not the concepts themselves play the major role but rather the way one uses them. The use is manifold but mainly connected with decisions for or against a behavior or an action.

  4. 4)

    The concepts and the use of the concepts have to be learned.

  5. 5)

    There is no sharp measurement of what the meaning of ‘successful learning’ is: The learning success is again something imprecise. As a consequence, the approximation character of the learning process is central.



Optimal Policy Reinforcement Learning Reward Function Learn Automaton Optimal Learning 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Bergmann, R. (2001): Experience Management, Manuscript Kaiserslautern 2001 ).Google Scholar
  2. 2.
    Burkhard, K-D., Richter, M.M. (2000): On the Notion of Similarity in Case Based Reasoning and Fuzzy Theory. In: Soft Computing in Case Based Reasoning (ed. Sankar K Pal et al), Springer Verlag 2000, p. 29–46Google Scholar
  3. 3.
    Globig, C., Jantke, K, Lange, S. Sakakibara, Y., (1997): Krechel, D. On case-based learn-ability of languages. New Generation Computing 15 (1), p. 57–63.CrossRefGoogle Scholar
  4. 4.
    Hardy, G.H., Littlewood„ J.E., Polya, G. (1934): Inequalities. Cambridge Univ. Press. Google Scholar
  5. 5.
    Lackoff, G. (1987): Women, Fire and Dangerous Things. The University of Chicago Press 1987.Google Scholar
  6. 6.
    Lenz, M., Bartsch-Spörl, B., Burkhard, H.-D., Wess, S. (eds.) (1998): Case-Based Reasoning Technology. Springer LNAI 1998.Google Scholar
  7. 7.
    Pape,Ch., Krechel, D., Richter, M:M. (1999): Wissensbasierte Bilddeutung in der Medizin am Beispiel des CYCLOPS-Systems. In: Angewandte Mathematik, insbesondere Informatik (ed. Patrick Horster) Vieweg-Verlag, ISBN 3-528-05720-3, 1999.Google Scholar
  8. 8.
    Pawlak, Z. (1984): Rough Classification. Intern.J. of Man-Machine Studies 20 (1984), p. 469–483.CrossRefMATHGoogle Scholar
  9. 9.
    Pfleger, T. (2000): Private communication.Google Scholar
  10. 10.
    Richter, M.M., Schmitt, S. (2001): Kundenmodellierung und Dialogführung: Eine Herausforderung fü eCRM. In: Electronic Customer Relationship Management, ed. Eggert, A., Fas-sot, G. Schäffer-Poeschel Verlag 2001, p. 175–198.Google Scholar
  11. Stahl, A. (2001).: Learning Feature Weights from Case Order Feedback. To appear in Proc. ICCBR 2001.Google Scholar
  12. 12.
    Wangenheim, A.V. (1996): Cyclops: Ein Konfigurationsansatz zur Integration hybrider Systeme am Beispiel der Bildauswertung, Reihe DISK!, infix-Verlag 1996.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2003

Authors and Affiliations

  • Michael M. Richter

There are no affiliations available

Personalised recommendations