Saturday, October 08, 2005

The Problem With Jeff Hawkins 

The Problem With Jeff Hawkins
"After those chapters were published, Jeff Hawkins published OnIntelligence, started a new company (Numenta) to develop software on his ideas and came close to getting it right.

But not right.

It is admirable that Hawkins has been searching for how the brain works. He started Redwood Neuroscience Institute to research the brain. He obviously has put great thought into how it works. But he has committed the one single error most known to neuroscience. He has imposed what he believes he perceives upon what he has perceived."

Connectionism in Action 

Mixing Memory: Connectionism in Action: Lexicon ex nihilo: "Connectionism in Action: Lexicon ex nihilo
After describing the basics of connectionist models the other day, I thought it might be interesting to give an example of their use. A responsible cognitive science blogger would choose a typical model that is designed to implement some individual aspect of cognition. But a responsible cognitive science blogger would be boring! I, on the other hand, not being boring, have chosen something a little different to use as an example: an agent-based model of the development of a lexicon in a community. Agent-based models are a relatively new breed in cognitive science, and I have to admit that I'm not really qualified to speak to all of their strengths and weaknesses, though an in press Trends in Cognitive Sciences paper by Rob Goldstone (aka Mr. Percepts and Concepts; see the picture at the linked site) seems to do a nice job of laying these out. If you want to learn more about agent-based models in general, I recommend reading it."

Meaning as a Social Phenomenon :: View topic - Meaning as a Social Phenomenon: "Meaning as a Social Phenomenon"
"But the main point of the model was to show that a community of communicating individuals could create a structured system of shared symbol-world relationships where no system had previously existed. This implies that the creation of meaning is a social phenomenon that occurs through the interaction of multiple agents in a shared environment. Interestingly, in their simulations, each of the agents developed different internal representations of the visual scenes. This indicates that agents need not have the same internal representations in order to be able to communicate with each other. "

"Well, we have a common system called the physical world and the structure of our bodies. The original quote starts with too basic of an assumption. If you wanted to develop a robot for exploring a planet where we didn't know the range of visible light, atmosphere, gravity, terrain, behavour of "life" forms, etc, then the problem posed in the quote is (almost) valid. Even then, you would program a "go out and explore" subroutine into the robot.

The author is quite correct with the qualification "the creation of meaning is a social phenomenon that occurs through the interaction of multiple agents in a shared environment." However, he does not assume a common brain structure, so internal representations of the world, and therefore meaning, may be very different from agent to agent.

In my opinion, there's no point developing an agent that has no built-in assumptions of it's intended environment or co-agents. You might have to wait the same amount of time since the first cell formed on earth. In which case, read the Hitch Hikers Guide to the Galaxy, for some theories on that.

To me the most facinating aspect isn't that internal representations and meaning are different, but that they're so much the same across time and cultures."

Monday, October 03, 2005

FW: Call for participation: 2005 AAAI Fall Symposium on Agents and the Semantic Web 

-----Original Message-----
[]On Behalf Of Valentina Tamma
Sent: Monday, October 03, 2005 8:34 AM
To: Valentina Tamma
Subject: Call for participation: 2005 AAAI Fall Symposium on Agents and
the Semantic Web

Apologies for multiple postings

--- Agents and the Semantic Web ---

2005 AAAI Fall Symposium Series
Call for Papers

Arlington, Virginia, USA
3rd-6th November, 2005

for information please contact Bijan Parsia <>

Invited Speakers:
James Hendler, University of Maryland
Michael Wooldridge, University of Liverpool

Accepted papers:
Dynamic Sub-Ontology Evolution for Collaborative Problem-Solving
Y. Mao, W. Cheung, Z. Wu, J. Liu

SEA: a Semantic Web Services Context-aware Execution Agent
A. Lopes, L. Botelho

Integrating Language Understanding Agents Into the Semantic Web
A. Java, T. Finin, S. Nirenburg

Integrating knowledge modeling and multi-agent systems
M. Gomez, E. Plaza

A Temporal Aggregates Ontology in OWL for the Semantic Web
F. Pan

Web Service Composition as a Planning Task: Experiments using
Knowledge-Based Planning
E. Martinez, Y. Lesperance

Service Level Agreements for Semantic Web Agents
N. Oren, A. Preece, T. Norman

Semantic Web Service Composition Planning with OWLS-Xplan
M. Klusch, A. Gerber, M. Schmidt

Integrating Agents, Ontologies, and Semantic Web Services for
Collaboration on the Semantic Web
M. Stollberg, S. Thomas

Protocols for Web Service Invocation
C. Walton

OWLS-MX: Hybrid OWL-S Service Matchmaking
M. Klusch, B. Fries, M. Khalid, K. Sycara

Template-based Composition of Semantic Web Services
E. Sirin, B. Parsia, J. Hendler

Interleaving Semantic Web Reasoning and Service Discovery
J. Rao, N. Sadeh

Preparing Semantic Agents for an Unsuspecting and Unreliable World
L. McDowell

Topics of interest
The Semantic Web is based on the idea of dynamic, heterogeneous,
shared knowledge sources providing machine-readable content in a
similar way to that in which information is shared on the World
Wide Web. Integral to this vision was a synergy with Multi-Agent
Systems technology; agents could utilize this knowledge to achieve
their own goals, producing new knowledge that could be disseminated
or published within a common framework. Conversely, the Semantic
Web would benefit from autonomous, distributed agents responsible
for gathering/aggregating knowledge, reasoning and inferring new
facts, identifying and managing inconsistencies, and providing trust
and security mechanisms.

Previous workshops and discussion fora devoted to this topic have
mainly focused on either the semantic web aspect or the agent aspect
of the problem, and have failed to achieve an agreement on the
common research themes. Thus there is a risk of missing significant
opportunities for sharing results in areas such as:

* Knowledge sharing. The agent paradigm is successfully employed
in those applications where autonomous, heterogeneous, and
systems need to interoperate in order to achieve a common goal,
however this is possible if agents are able to share knowledge.
Ontologies are a powerful tool to achieve semantic interoperability
among heterogeneous, distributed systems.

* Syntactic Unification. Data exchanged between service providers
are typically based on different syntaxes and conceptual schemas,
raising the problem of data mediation for interoperability.
and mechanisms for mapping and translating across ontologies
can address these problems.

* Discovery of agent capabilities. Semantic-based discovery
mechanisms and languages/ontologies for describing agent capabilities
and predefined coordination mechanisms are needed to make the
automatic discovery of services offered by agents and other

* Agent coordination. Goal-directed composition typically involves
planning across a space of existing actions, ensuring that data
and control flow constraints are satisfied. Model checking
techniques are required to ensure valid compositions, as well as
temporal reasoning to validate control flow dependences. Such
techniques need to accommodate semantic descriptions as well as
avoiding live-lock situations that may lead to failure.

* Interaction Protocols. Different agents expect specific messages
to be choreographed in a precisely defined manner. Integration
has to guarantee and enforce the communication protocols.
Interoperable description frameworks are thus required to ensure
that both parties understand and adhere to interaction protocols.
The semantics of the terms used in these protocols is made explicit
in ontologies.

This symposium aims to promote and foster a greater understanding
of the synergy between Multi-Agent Systems and the Semantic Web.

Topics of Interest include:
- Semantic interoperability and integration
- Distributed, autonomous knowledge-management
- Dynamic, semantic mapping across ontologies;
- Use of negotiation techniques for reaching consensus;
- Evolution of ontologies in multi-agent systems;
- Scalability and versioning of ontologies in multi-agent systems;
- Centralized and Distributed mechanisms for service invocation/
- Failure and Recovery mechanisms
- Semantic descriptions of Autonomic mechanisms for robust,
coordinated service communities
- Semantic description, discovery, and selection of services and
- Semantic Web Services (including OWL-S and WSMO)
- Semantics in Agent Communication Languages
- Semantics in Interaction Protocols
- Semantics in Electronic Institutions
- Semantics for service delegation and knowledge aggregation
- Architectures for supporting Agents and Web Services within the
Semantic Web

Registration Deadline: October 7th
Symposium: November 3rd-6th

Organizing Committee

Terry Payne (Co-chair) University of Southampton
Valentina Tamma (Co-chair) University of Liverpool
Bijan Parsia University of Maryland
David Martin SRI International
Simon Parsons City University of New York
Nick Gibbins University of Southampton

Program Committee
Alice Mulvehill, BBN Technologies, MA
Alun Preece, University of Aberdeen, UK
Andreas Hess, University College, Dublin, Ireland
Anupam Joshi, University of Maryland, Baltimore County, MD
Boi Faltings, EPFL, Switzerland
Brian Blake, Georgetown University, Washington, DC
Fabio Casati, HP Labs, CA
Chiara Ghidini, ITC, Italy
Chris Priest, HP Labs, UK
Chris Walton, University of Edinburgh, UK
Chris van Aart, NL
David Robertson, University of Edinburgh, UK
Dave de Roure, University of Southampton, UK
Enrico Motta, Open University, UK
Evren Sirin, University of Maryland, MD
Fabien Gandon, Inria, Sophia Antipolis, France
Filip Perich, Cougar Software, McLean, VA
Frank van Harmelen, Vrije Universiteit Amsterdam, NL
Grit Denker, SRI, CA
Michael Huhns, University of South Carolina, SC
John Domingue, Open University, UK
Jeffrey M. Bradshaw, University of West Florida, FL
Jim Blythe, ISI, University of South California, CA
Julian Padget, University of Bath, UK
Katia Sycara, Carnegie Mellon University, PA
Ian J. Dickinson, HP Labs, UK
Ion Costantinescu, EPFL, Switzerland
Kaoru Hiramatsu, NTT Corporation, Japan
Mark Burstein, BBN Technologies, MA
Mark S. Fox, Univerity of Toronto, Canada
Marta Sabou, Vrije Universiteit Amsterdam, NL
Matthias Klusch, DFKI, Germany
Michael Klein, University of Karlsruhe, Germany
Monica Crubezy, University of Stanfard, CA
Monika Solanki, De Montfort Univeristy, UK
Munindar Singh, North Carolina State University, NC
Norman Sadeh, Carnegie Mellon University, PA
Onn Shehory, IBM, Israel
Ora Lassila, Nokia
Paul Buhler, College of Charleston, SC
Ramanathan Guha, IBM Almaden Research Center, CA
Pete Edwards, University of Aberdeen, UK
Richard Benjamins, iSOCO, Spain
Ronald Ashri, University of Southampton, UK
Ryusuke Masuoka, Fujitsu Labs of America, MD
Sheila MacIlraith, University of Toronto, Canada
Simon Thompson, BT Labs, UK
Stefan Decker, DERI, Ireland
Steffen Staab, University of Koblenz, Germany
Steve Battle, HP Labs, UK
Stephen Cranefield, Otago University, New Zeland
Steven Willmott, University of Catalunya, Spain
David Trastour, HP Labs, UK
Mike Uschold, Boeing, WA
Walter Binder, EPFL, Switzerland
Tim Finin, University of Maryland, Baltimore County, MD
Lalana Kagal, MIT, MA
Carine Bournez, W3C
Ashok Mallya, North Carolina State, NC
Mary Pulvermacher, MITRE Corporation, CO
Yannis Labrou, Fujitsu Labs of America, MD
Yolanda Gil, ISI, University of South California, CA

This page is powered by Blogger. Isn't yours?