|
|
Sponsored by:
Intelligent Systems Group (Macquarie University)
Intelligent Systems Laboratory (University of Western Sydney)
Conventicle Date and Venue:
Friday, 7th March 2008;
Dunmore Lang College
130 -134 Herring Rd, North Ryde
How to get to the Conventicle Venue?
Dunmore Lang College is very close to Macquarie Centre. If you are driving, get to Herring Road off Epping Road (to left if coming from Epping, to right if coming from City). Go through a round about. Dunmore Lang College is on the left (please see map here ). Limited onsite parking available in front of the college. Off site parking available on the other side of the Herring Road, as well as on Windsor drive and Lachlan Ave. If you prefer public transportation, take a bus to Macquarie Centre, then walk to Dunmore Lang College.
Organisers:
Abhaya Nayak, Macquarie University
Maurice Pagnucco, University of New South Wales
Yan Zhang , University of Western Sydney
Program Schedule
| Time | Presentation |
|---|---|
| 9:00 - 9:15 |
Opening Address by Prof. Stephen Thurgate, Dean of ICS, Macquarie University. |
| 9:15 - 9:45 |
Hans Rott, University of Regensburg. Belief change and free will |
| 9:45 - 10:05 |
Abhaya Nayak, Macquarie University. Iterated Belief Contraction |
| 10:05 - 10:25 |
Arthur Ramer,
University of New South Wales. GM meets Dirichlet: iterated revision of probabilistic beliefs |
| 10:25 - 10:45 | COFFEE/TEA BREAK |
| 10:45 - 11:15 |
Torsten Schaub, University of Potsdam. Biological Knowledge Representation and Reasoning |
| 11:15 - 11:45 |
Kewen Wang, Griffith University Semantic forgetting and its application in the semantic web |
| 11:45 - 12:05 |
Ka Shu Wong, University of New South Wales Forgetting on logic programs under strong equivalence |
| 12:05 - 12:25 |
Ron van der Meyden, University of New South Wales What, indeed, is intransitive noninterference? |
| 12:25 - 13:30 | LUNCH |
| 13:30 - 14:00 |
Michael Thielscher, University of Dresden. From General Game Playing to Trading Agents |
| 14:00 - 14:30 |
John Lloyd, The Australian National University. A Reasoning System for Quantified Modal Logic |
| 14:30 - 14:50 |
Wayne Wobcke, University of New South Wales. An Axiomatization of a Logic for Reasoning about Rational Agents! |
| 14:50 - 15:10 |
David Rajaratnam, University of New South Wales. Motivating Resource Bounded Logics |
| 15:10 - 15:30 | COFFEE/TEA BREAK |
| 15:30 - 16:30 | Panel/Open Session |
Talk Abstracts
A Reasoning System for Quantified Modal Logic
John Lloyd
The Australian National University
Abstract:
I will begin with some motivation for the need to consider quantified modal logic,
especially for agent applications. Then I will outline a reasoning system for the logic,
consisting of a proof component and a computation component that can call each other,
and give some illustrations of its use.
Background material for the talk can be found here.
| Back to Top |
Abhaya Nayak
Macquarie University
Abstract:
Importance of contraction for belief change notwithstanding, literature on
iterated belief change has by and large centered around the issue of iterated belief revision, ignoring the problem of iterated belief contraction.
In this paper we examine iterated belief contraction in a principled way, starting with Qualified Insertion, a proposal by Hans Rott.
We show that a judicious combination of Qualified Insertion with a well-known Factoring principle leads to what is arguably a pivotal principle of iterated belief contraction. We show that this principle is satisfied by the account of iterated belief contraction modelled by Lexicographic State Contraction, and outline its connection with Lexicographic Revision, Darwiche-Pearl's account of revision as well as Spohn's Ordinal ranking theory.
| Back to Top |
Motivating Resource Bounded Logics
David Rajaratnam
University of New South Wales
Abstract:
In this talk I argue that the logic based agent/AI community needs to be more
strongly focused on the intractability of classical logic in developing logical
models of agent reasoning. The traditional approach has assumed an idealised
view of agent reasoning and makes no consideration for the resource and computational
limitations of an agent. This approach has been justified based on normative grounds;
the intention is to model how an agent should behave,
its commitments, rather then how an agent actually behaves,
its performance.
In this talk I argue against this approach, both on practical as well as philosophical grounds.
Instead I argue that agent commitment and performance are inextricably linked and that normative
models can only be established empirically, based on the performance of actual agents,
rather than theoretically, based on idealised notions of rationality.
| Back to Top |
Arthur Ramer
University of New South Wales.
Abstract:
We present solutions to several open problems suggested by Gardenfors
in his book Knowledge in Flux.
The question is wheter inverse conditioning, imaging and related
probability kinematics can be effected as a minimal change.
The answers obtained through the use of entropies and graph entropies,
and their related Dirichlet generating functions.
A related application deals with Lewis' impossibility lemma for conditional objects. We prove a general nonexistence of atemporal objects in an arbitrary extension of the original probability domain.
| Back to Top |
Hans Rott
University of Regensburg.
Abstract:
In this talk I begin by giving a very short overview of the history of belief revision research of the past 25 years. An important development is that the objects of revision are now no longer conceived as belief sets (sets of sentences), but rather belief states (which are more complex structures). The classical AGM paradigm used doxastic preferences as structures guiding the revision of belief sets, and the way how these structures were applied was uncontroversial. The move from belief set revision to belief state revision involved the incorporation of doxastic preferences into the notion of a belief state, and the focus of research has shifted on (a fairly large number) methods of revising such preferences. It seems that we need principles for choosing between such methods, and that these principles should in turn count as part of the agent's mental state. Supposing this choice problem can be solved, the same kind of argument can be reiterated. We are thus lead into an infinite regress or, in an alternative presentation of the same problem, into a circularity of the definition of a belief revision function. In the second part of the talk I ask whether the choices involved in belief change might be manifestations of the agent's free will. I argue that this problem arrived at in the history of belief revision research is at least structurally identical with, and perhaps substantially an instantiation of, a well-known problem with free will put forward by Galen Stawson.
| Back to Top |
Torsten Schaub
University of Potsdam
Abstract:
Modeling and interacting with and within natural environments constitutes a
major challenge in informatics. This ranges from robotics, bio-informatics to
health care applications. Common problems in such domains include change,
incompleteness (due to partial observability), and inconsistencies. We address
these issues by appeal to techniques developed in the area of knowledge
representation and reasoning, using action languages and answer set programming
as the basic representation and reasoning framework. Although this choice has
several benefits like succinct and elaboration tolerant representations as well
as the availability of high performance systems that allow for dealing with
millions of variables, it is nonetheless a general purpose approach that is
sometimes difficult to access for domain experts, like physicians or biologists.
My talk will take up an ongoing project in systems biology and discuss by several examples how domain specific techniques can be used to address a domain experts needs, while remaining fully transparent as regards the underlying reasoning technology.
| Back to Top |
Michael Thielscher
University of Dresden
Abstract:
As one of the grand challenges for Artificial Intelligence, General Game
Playing (GGP) is concerned with the development of programs that accept
formal descriptions of arbitrary games and play these games without human
intervention. GGP requires to combine techniques from a wide range of areas,
including knowledge representation, reasoning, and game theory. In this talk,
we investigate how recent developments in GGP can be applied to the design
and analysis of automatic trading platforms and trading agents. Specifically,
we will show how the general Game Description Language can be adapted to a
high-level, declarative language for specifying arbitrary trading scenarios.
| Back to Top |
Ron van der Meyden
University of New South Wales
Abstract:
This paper argues that Haigh and YoungËs definition of noninterference for intransitive security
policies admits information flows that are not in accordance with the intuitions
it seeks to formalise.
Several alternative definitions are discussed, which are shown to be equivalent
to the classical
definition of noninterference with respect to transitive policies.
RushbyËs unwinding conditions for intransitive noninterference are shown to be sound and
complete for one of these definitions, TA-security.
Access control systems compatible with a policy are shown to be TA-secure,
and it is also shown that TA-security implies that the system can be interpreted
as an access control system.
| Back to Top |
Kewen Wang
Griffith University
Abstract:
The ability of discarding irrelevant information is a key feature
that an intelligent agent must possess to adequately handle reasoning tasks.
This ability is referred to as variable forgetting (or variable
elimination). In this talk we will first examine some requirements for a
suitable notion of forgetting in nonmonotonic reasoning and then introduce a
semantic notion of forgetting in the setting of answer set programming. We
will also briefly outline some applications of forgetting in editing,
reusing and merging ontologies for the semantic web.
| Back to Top |
An Axiomatization of a Logic for Reasoning about Rational Agents!
Wayne Wobcke
University of New South Wales
Abstract:
We provide a sound and complete axiomatization
of Agent Dynamic Logic (ADL), a logic for reasoning
about a class of BDI agent architectures. ADL combines
Computation Tree Logic, Propositional Dynamic Logic and
Rao and GeorgeffËs BDI Logic, and the main definition relates
intention to action, enabling a rigorous formal approach
to belief, desire and intention that is also computationally
grounded in the operational behaviour of agent architectures.
The completeness proof incorporates a result from process
algebra on equivalence of programs expressed as regular expressions,
and extends this result to the BDI agent framework to model
the actions, and indirectly the intentions, of an agent.
| Back to Top |
Forgetting on logic programs under strong equivalence
Ka-Shu Wong
University of New South Wales
Abstract:
In this talk I will look at the properties that a notion of forgetting on logic programs should have, particularly in relation to strong equivalence on answer set programs. I will then present some results about the forgetting operators which do have these properties.
| Back to Top |