Page 1
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Simulation as a tool in GI ScienceSandbox Geography continued …
Florian Twaroch
Annual Scientific Meeting – GeoGERAS 2005Pernegg, Austria
Institute for Geoinformation & Cartography27th – 28th June 2005
Page 2
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
MotivationMotivation BackgroundBackground HypothesisHypothesis MethodMethod ConclusionConclusion
Overview
● Motivation● Background - Literature● Hypothesis● Methodology
– How to organize knowledge of the world– How to change the knowledge
● Conclusions
Page 3
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation BackgroundBackgroundBackgroundBackground Hypothesis Method Conclusion
Example for experiment
● Hidden object experiments [Baillargeon 2003]
– Confront child with real and „magic“ event
– Independent observer records
● Reaction time
● Reaction behavior
● Preferential looking studies
● Predictive action studies
http://www.psych.uiuc.edu/~rbaillar/ICL/welcome.html
Page 4
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation BackgroundBackgroundBackgroundBackground Hypothesis Method Conclusion
Developmental Psychology & AI
(Mareschal 2001)
Nativism Empiricism
symbolic
sub-symbolic
Page 5
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation BackgroundBackgroundBackgroundBackground Hypothesis Method Conclusion
Theory Theory
● Theory theory: “Infants learn about the world by forming and revising theories. Their conceptual development is theory formation and change, their semantic development is theory dependent.”
● Development is driven by three forces
– Innate knowledge– Powerful learning abilities– Unconcious tuition from adults
(Meltzoff & Gopnik 2002)
Page 6
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation Background HypothesisHypothesisHypothesisHypothesis Method Conclusion
Hypothesis
Theories of space can be described by a set of axioms. It is possible to revise such a theory by adding, deleting or exchanging axioms.
Page 7
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation Background Hypothesis MethodMethodMethodMethod Conclusion
Page 8
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation Background Hypothesis MethodMethodMethodMethod Conclusion
Graphical Models
● Graphical models are graphs in which nodes represent random variables.
● A Bayesian Network is kind of directed graphical model , which takes into account the directionality of the arcs. (arrows between nodes)
● Advantage of a directed graphical model is that one can regard an arc from A to B as indicating that A ``causes'' B..
● Models how beliefs change in the light of new evidence
A B
Page 9
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation Background Hypothesis MethodMethodMethodMethod Conclusion
Stochastic Modelling
● P1(sunrise) = 0.5 P1(no sunrise) = 0.5● P2(sunrise) = 0.66 P2(no sunrise) = 0.33● P100(sunrise) = 0.98 P100(no sunrise) = 0.02
Page 10
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation Background Hypothesis MethodMethodMethodMethod Conclusion
Page 11
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation Background Hypothesis MethodMethodMethodMethod Conclusion
Page 12
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation Background Hypothesis MethodMethodMethodMethod Conclusion
Page 13
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation Background Hypothesis Method ConclusionConclusionConclusionConclusion
Conclusion
● The big question is how to connect the simple parts of object ontologies.
● Stochastic methods are a step towards linking them.
● Agent based simulations help to get sound theories in developmental psychology.
● Psychology serves as an input for new computational models in other disciplinces.
Page 14
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation Background Hypothesis Method ConclusionConclusionConclusionConclusion
To be continued …To be continued …
Page 15
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation Background Hypothesis Method ConclusionConclusionConclusionConclusion
Page 16
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation Background Hypothesis Method ConclusionConclusionConclusionConclusion
Agent Based Approach
● Modular: A modular agent that observes a simple toy world.
● Hierarchical: The agents knowledge is structured hierarchically.
● Dynamic: Exchange of Axioms leads to new knowledge.
● Action driven: The acting of the agent in the world elicits changes in the model.
Page 17
Florian TwarochDepartment of Geoinformation and Cartography, TU Vienna
[email protected]
Motivation Background Hypothesis Method ConclusionConclusionConclusionConclusion
Concepts of the world