Beyond Knowing That: Axiomatizing Ability-Based Logics with Regularity Constraints

Title

Beyond Knowing That: Axiomatizing Ability-Based Logics with Regularity Constraints

Speaker

Ziqi Wang (Student of the MCS Program at GTIIT)

Time and Location

October 17, Thursday, 18:45, E509, Education Building,  North Campus

Pizza will be provided after the seminar.

Abstract

Ability-based epistemic  are ubiquitous in computer science, as arguably they provide us with a formal approach for strategic reasoning and AI planning. Several logics in this family have been studied lately. In particular, epistemic logics describing the notion of knowing how received a lot of attention, due to their simplicity and good meta-logical properties.

We studied one of the variations of the knowing how logic with regularity constraints. In particular, we revisited the uncertainty-based logic, encoding the uncertainty by a regular language. We demonstrated that this new logic, introduced in [2], retains the strong completeness of the original axioms from [1]. We proved this by defining a new concept: the basic automaton, and constructing the corresponding canonical model.

[1] C. Areces, R. Fervari, A. R. Saravia, and F. R. Velázquez-Quesada. Uncertainty-based semantics for multi-agent knowing how logics. In Proceedings of the 18th Conference on Theoretical Aspects of Rationality and Knowledge (TARK 2021), volume 335 of EPTCS, pages 2337, 2021.

[2] S. Demri and R. Fervari. Model-checking for ability-based logics with constrained plans. In Proceedings of the 37th AAAI Conference on Artificial Intelligence (AAAI-23), volume 37, pages63056312, 2023.