How Should Intelligence be Abstracted in AI Research: MDPs, Symbolic Representations, Artificial Neural Networks, or _____?


AAAI Fall Symposium 2013
November 15-17
Westin Arlington Gateway in Arlington, Virginia adjacent to Washington, DC

* DEADLINE EXTENDED * Paper/Extended Abstract Submissions: June 24, 2013

Overview

Artificial intelligence is a disparate field that consists of many smaller communities with diverse approaches towards the common goal of creating computational intelligence. While they are unified by this common goal, and most take inspiration from human or biological intelligence, these separate communities are driven by different abstractions of intelligence (e.g. symbolic manipulation, connectionist networks, or embodiment) and the processes that might produce it (e.g. reinforcement learning, evolutionary computation, or statistical machine learning). Because the particular choice of abstraction profoundly impacts how a problem is framed, reaching the ambitious goals of AI may be aided by carefully examining the promise, drawbacks, and motivations of popular abstractions.

Interestingly, there are often many different levels from which to abstract the same underlying phenomenon, which causes subfields of artificial intelligence to diversify from a core idea. For example, deep learning networks, models in computational neuroscience, and neuroevolution all take inspiration from biological neural networks as a potential pathway to AI. Most researchers choose to pursue the subfield (and by extension, abstraction) they see as most promising for leading to AI, which naturally results in significant debate and disagreement among researchers as to what abstraction is best. Similarly, researchers more directly studying a phenomenon may believe that those studying it more abstractly are ignoring important details or failing to grasp the crux of the studied system. A better understanding and less polarized debate may be facilitated by a clear presentation and discussion of abstractions by their most knowledgeable proponents.

Because communities can often become insulated from one another, useful insights at one level of abstraction might filter only slowly to other related communities. Additionally, researchers at one level may increasingly view the motivations of those at other levels only in caricature, unintentionally simplifying opposing perspectives from a lack of meaningful interaction. Thus it may be useful to bring together researchers from fields that abstract AI at different levels or in different ways, both to disperse knowledge and to critically examine the value and promise of different abstractions.

For these reasons, we propose a symposium that aims to bring together a diverse and multi-disciplinary group of AI researchers interested in discussing and comparing different abstractions of both intelligence and processes that might create it. In this way, we hope to provide a common ground for these diverse perspectives; the result will be cross-pollination of ideas between levels and types of abstraction, and new ideas for revising and creating abstractions of intelligence and intelligence-generating processes.

Call for papers

We invite contributions related to how intelligence can or should be abstracted in artificial intelligence research. Papers that provide a high-level overview of existing work or summarize the results of an extended research program along these lines are most welcome, as are position papers or contributions describing speculative work or work in progress. Works bridging traditionally separate AI paradigms are encouraged. Interested participants may submit either full-length papers (up to 6 pages in AAAI format) or short papers/extended abstracts (up to 2 pages) in PDF format to sebastian.risi@cornell.edu

Topics of Interest:

Keynote Speakers

Important Dates

Full Paper/Extended Abstract Submissions: June 24, 2013
Notification: July 7, 2013
Final camera-ready paper/extended abstract: September 12, 2013

Program

A PDF version of the program can be found here.

Registration

AAAI Fall Symposium registration website

Organizing Committee

Sebastian Risi - sebastian.risi@cornell.edu
Joel Lehman - joel@cs.utexas.edu
Jeff Clune - jeffclune@uwyo.edu

Program Committee

Dave Knoester
Martin Pyka
Stéphane Doncieux
Joshua Auerbach

Student Scholarships for FSS13

Thanks to the generous support of Artificial Intelligence (AIJ), multiple scholarships are available to help support student travel to the symposia series. Accepted student authors will be given priority.

http://www.aaai.org/Symposia/Fall/fss13.php

Scholarship application deadline: September 15, 2013
Scholarship notification date: October 1, 2013

To apply for a scholarship, please send the following information to Matt Taylor (taylorm at eecs dot wsu dot edu) by the scholarship application deadline. Put "FSS13 Scholarship" in the subject line of your email.

1. The name of the symposium you wish to attend
2. Point of origin for traveling to/from the symposium
3. An itemized list of your expected travel costs (registration, transportation, hotel, etc.)
4. Your CV (PDF preferred)
5. Are you a co-author of an accepted symposium paper? If so, what is the title and paper #?
6. Are you presenting?
7. Have you previously attended an AAAI symposium?

As a scholarship recipient, you will commit to attending the 2.5 day symposium and submit a post-conference summary of your experience.

Questions? Contact Matt Taylor.