From Sensorimotor Graphs to Rules: An Agent Learns from a Stream of Experience

TitleFrom Sensorimotor Graphs to Rules: An Agent Learns from a Stream of Experience
Publication TypeConference Paper
Year of Publication2011
AuthorsRaab, M., Wernsdorfer M., Kitzelmann E., & Schmid U.
Volume6830
Page(s)333-339
Other Numbers3215
Keywordscognitive architecture, inductive rule learning, symbol grounding, temporal Hebbian learning
Abstract

In this paper we argue that a philosophically and psychologicallygrounded autonomous agent is able to learn recursive rules frombasic sensorimotor input. A sensorimotor graph of the agent’s environmentis generated that stores and optimises beneficial motor activationsin evaluated sensor space by employing temporal Hebbian learning. Thisresults in a categorized stream of experience that feeds in a Minervamemory model which is enriched by a time line approach and integratedin the cognitive architecture Psi—including motivation and emotion.These memory traces feed seamlessly into the inductive rule acquisitiondevice Igor2 and the resulting recursive rules are made accessible in thesame memory store. A combination of cognitive theories from the 1980iesand state-of-the-art computer science thus is a plausible approach to thestill prevailing symbol grounding problem.

Acknowledgment

This work was partially funded by the Deutscher Akademischer Austausch Diesnst (DAAD) through a postdoctoral fellowship.

URLhttp://www.icsi.berkeley.edu/pubs/ai/RaabWernsdorferEtAl2011.pdf
Bibliographic Notes

Proceedings of the Fourth International Artificial General Intelligence Conference (AGI 2011), Vol. 6830, pp. 333-339, Mountain View, California

Abbreviated Authors

M. Raab, M. Wernsdorfer, E. Kitzelmann, and U. Schmid

ICSI Research Group

AI

ICSI Publication Type

Article in conference proceedings