Output details

A defeasible reasoning model of inductive concept learning from examples and communication

<22>This paper proposes a novel defeasible reasoning model of inductive generalization, and specifically of the machine learning task of inductive concept learning (ICL). The first contribution is the definition of a specific non-monotonic inductive consequence relation for ICL, that facilitates the integration of inductive reasoning with other forms of logical reasoning. The second one is a model of multi-agent ICL obtained by integrating our inductive consequence relation with computational argumentation, the latter being used to model the communication between agents, and ICL models their internal learning processes. The research has been partially funded by the ESF EUROCORES-LogICCC project "LoMoReVI".