Learning Actions Models: Qualitative Approach

DOI10.1007/978-3-662-48561-3_4MaRDI QIDQ3457685zbMATH OpenOpenAlexDBLPWikidataFDO

Authors Thomas Bolander, Nina Gierasimczuk

Publication date 17 December 2015

Published in Logic, Rationality, and Interaction (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1507.04285, https://orbit.dtu.dk/en/publications/6cf95a69-3a82-4ab9-b7d5-823a40e3db22

Mathematics Subject Classification ID

Logics of knowledge and belief (including belief change) (03B42) Computational learning theory (68Q32) Logic in artificial intelligence (68T27)

Abstract: In dynamic epistemic logic, actions are described using action models. In this paper we introduce a framework for studying learnability of action models from observations. We present first results concerning propositional action models. First we check two basic learnability criteria: finite identifiability (conclusively inferring the appropriate action model in finite time) and identifiability in the limit (inconclusive convergence to the right action model). We show that deterministic actions are finitely identifiable, while non-deterministic actions require more learning power-they are identifiable in the limit. We then move on to a particular learning method, which proceeds via restriction of a space of events within a learning-specific action model. This way of learning closely resembles the well-known update method from dynamic epistemic logic. We introduce several different learning methods suited for finite identifiability of particular types of deterministic actions.

Recommendations

• Learning to act: qualitative learning of deterministic action models
• scientific article; zbMATH DE number 1134972
• Model-based reasoning about learner behaviour
• scientific article; zbMATH DE number 622664
• Learning qualitative models of dynamic systems
• Qualitative case-based reasoning and learning

Cites work

• Action emulation between canonical models
• Conditional epistemic planning
• Epistemic planning for single- and multi-agent systems
• scientific article; zbMATH DE number 2201583 (Why is no real title available?)
• Inductive inference of formal languages from positive data
• Learning by Erasing in Dynamic Epistemic Logic
• Logic and learning
• Semantic results for ontic and epistemic change
• The logic of public announcements, common knowledge, and private suspicions

Cited in

(5)

• Truth-tracking by belief revision
• Learning to act: qualitative learning of deterministic action models
• Learning by intervention in simple causal domains
• Learning to act and observe in partially observable domains
• Learnability with PAC semantics for multi-agent beliefs

This page was built for publication: Learning Actions Models: Qualitative Approach

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3457685)