• LuxLogAI 2018

    Luxembourg Logic for AI Summit

    17-26 September 2018

    Luxembourg

SST ABCDE 2018

Short Summer Tutorial on Advanced Belief Change and Default Entailment

Organizer: Emil Weydert
Teachers: Richard Booth (Cardiff), Giovanni Casini (Luxembourg), Emil Weydert (Luxembourg)
Date: September 20, 2018
Room: MSA 3.120

The ABCDE summer tutorial 2018 gives a survey of exciting recent developments in default reasoning and belief change and offers a look at some interesting but hardly known facts from these areas.

It takes place on Thursday, Sep 20, at the Unicampus Belval in the context of the Luxembourg Logic for AI Summit 2018.

About

Intelligent agents have to meet many challenges. One of them is to revise their beliefs or theories in rational ways in the face of conflicting new information. Another one is to be able to draw, whenever required, plausible but potentially defeasible conclusions from incomplete, uncertain, if not inconsistent information, e.g.~expressed by default implications allowing exceptions.

Humans are rather good at handling such instances of commonsense reasoning. However, the formal modeling of belief revision and defeasible reasoning has turned out to be surprisingly difficult to achieve. Simple benchmark examples have been difficult to handle in a coherent way, and rationality principles have been hard to satisfy. The existence of different paradigms and the complexity of the real world are further challenges.

But today, around 40 years after the first seminal publications started to address these questions in a systematic way, we have gained a pretty good theoretical understanding of belief change and defeasible reasoning, including their mutual relationship. Both fields have now reached a degree of maturity where they may serve as a guide for real-world artificial intelligence.

Nevertheless, it is fair to say that even today there are still important problems without fully convincing conceptual and theoretical solutions, calling for further research. And the rise of AI with its more specific empirical demands promises to add many more. This is therefore a good moment to take stock on the theoretical-conceptual side.

Tutorial

Our 1-day tutorial extends and complements existing introductions into belief change and default entailment:

  1. It offers an advanced theoretical perspective on belief revision and defeasible reasoning without obscuring the big picture with too many technical details.
  2. It pays particular attention to crucial but less well known problems and facts which may even surprise some experts.
  3. It looks at the state of the art and recent developments, notably new techniques, general principles, and semantic frameworks.

The speakers are experienced researchers which have been active in this field for up to 30 years and who are eager to communicate their insights to a rising generation of researchers investigating and developing new paradigms, techniques, and applications for belief change and plausible inference.

Content

Belief change:
Introduction, interval-based revision, trust-sensitive revision, revision in nonmonotonic theories, revision in description logics.

Default entailment:
The nonmonotonic reasoning zoo, old and new quasi-probabilistic default formalisms, beyond KLM – principles and puzzles, first-order default reasoning, default revision.

Registration

Registration fees are 50 EUR. Note that access to SST ABCDE is included without any additional expenses if you already registered for RuleML+RR/GCAI at LuxLogAI.

Registration to SST ABCDE is done via the EasyChair registration process of LuxLogAI, see here for further information.