5th Conference on Artificial General Intelligence (AGI-12)

Continuing the mission of the first four AGI conferences, AGI-12@Oxford gathers an international group of leading academic and industry researchers involved in scientific and engineering work aimed directly toward the goal of artificial general intelligence.    Appropriately for this  Alan Turing centenary year, this is the first AGI conference to be held in the UK.

The AGI conferences are the only major conference series devoted wholly and specifically to the creation of AI systems possessing general intelligence at the human level and ultimately beyond. By gathering together active researchers in the field, for presentation of results and discussion of ideas, we accelerate our progress toward our common goal.

AGI-12@Oxford will feature contributed talks and posters, keynotes, and a Special Session on Neuroscience and AGI.  It will be held immediately preceding the first conference on AGI Safety and Impacts, which is organized by Oxford’s Future of Humanity Institute; AGI-12 registrants will receive free admission to the latter conference.  Proceedings will be published as a book in Springer’s Lecture Notes in AI series.

“Artificial General Intelligence”

The original goal of the AI field was the construction of “thinking machines” – that is, computer systems with human-like general intelligence. Due to the difficulty of this task, for the last few decades the majority of AI researchers have focused on what has been called “narrow AI” – the production of AI systems displaying intelligence regarding specific, highly constrained tasks.

In recent years, however, more and more researchers have recognized the necessity – and feasibility – of returning to the original goals of the field. Increasingly, there is a call for a transition back to confronting the more difficult issues of “human level intelligence” and more broadly artificial general intelligence (AGI).  The AGI conference series has played, and continues to play, a significant role in this resurgence of research on artificial intelligence in the deeper, original sense of the term.

Schedule

Each contributed-talks session (Sessions 1-6, plus the AGI & Neuroscience session) will consist of a number of 15-minute talks, followed by a lengthy panel discussion involving the speakers.   Specific papers corresponding to each contributed-talks section are listed at the end of this page, below the summary schedule.

Sat. Dec. 8

8:30AM.  Conference intro: Ben Goertzel

8:45AM.  Keynote talk by Margaret Boden: Creativity and AGI

9:45 AM.  Keynote talk by David Hanson:  Open source genius machines who care: Growing Friendly AGI via GENI, Glue and Lovable Characters

10:45AM.  Morning break

11:15AM.  Keynote talk by Angelo Cangelosi, From Sensorimotor Intelligence to Symbols: Developmental Robotics Experiments

12:15PM.  Lunch

1:30-2:30PM.  Keynote talk by Nick Bostrom:  The Superintelligence Control Problem

2:35PM.  Session 1: Cognitive Architectures & Models A

4:20PM.  Break

4:45-6:30PM.  Session 2: Cognitive Architectures & Models B

Sun. Dec. 9

8:30 AM.  Session 3: Universal Intelligence and its Formal Approximations

10:30 AM.   Break

11AM.  Session 4: Conceptual and Contextual Issues

12:40PM.  Lunch

1:15PM.  (Attendance optional) AGI Society / conference series business meeting

2PM.  Session 5: Cognitive Architectures and Models C

4PM.  Break

4:30 PM.  Session 6: Mathematical Formalisms and Tools

Mon. Dec. 10

8:30-10:30AM.  AGI-12 tutorial with Aaron Sloman.  Topic:  Meta-morphogenesis: How a planet can produce Minds, Mathematics and Music

10AM.  break

10:30AM-12:30.   Hands-on AGI-12 tutorial on the CHREST cognitive architecture

1:45 AGI-Impacts, Session 1

Tues Dec. 11

8:30AM.   Special Session on AGI & Neuroscience

10:30AM.  Break

11AM.  AGI Roadmap discussion, led by Ben Goertzel and Joscha Bach.   To what extent, and in what way, can we articulate a common development and evaluation roadmap spanning multiple AGI approaches.  What tools might make this easier?  See related article  and video.

12PM.  Lunch

1:30PM AGI-Impacts, Session 2

AGI-12 Contributed Paper Sessions:

Session 1: Cognitive Architectures & Cognitive Modeling A

The Next Generation of the  MicroPsi Framework. Joscha Bach, Humboldt University of Berlin, Germany

Motivation Management in AGI Systems.  Pei Wang, Temple University, United States

On Attention Mechanisms for AGI Architectures: A Design Proposal. Helgi Helgason, Eric Nivel and Thórisson Kristinn, Reykjavik University / School of Computer Science, Iceland

Pursuing Artificial General Intelligence By Leveraging the Knowledge Capabilities Of ACT-R.   Alessandro Oltramari and Christian Lebiere, Carnegie Mellon University, USA

CHREST models of implicit learning and board game interpretation.  Peter Lane and Fernand Gobet, University of Hertfordshire, UK

Session 2: Cognitive Architectures & Cognitive Modeling B

Perception Processing for General Intelligence: Bridging the Symbolic/Subsymbolic Gap, Ben Goertzel, Novamente LLC, USA & Hong Kong

Pattern Mining for General Intelligence: The FISHGRAM Algorithm for Frequent and Interesting Subhypergraph Mining.  Jade O’Neill and Ben Goertzel.   Hong Kong Polytechnic University.

Syntax-Semantic Mapping for General Intelligence: Language Comprehension as Hypergraph Homomorphism, Language Generation as Constraint Satisfaction.  Ruiting Lian and Ben Goertzel, Hong Kong Polytechnic University and Xiamen University, China

Deconstructing Reinforcement Learning in Sigma.  Paul Rosenbloom, University of Southern California, USA

Extending Mental Imagery in Sigma.  Paul Rosenbloom, University of Southern California, USA

Session 3: Universal Intelligence and its Formal Approximations

On Ensemble Techniques for AIXI Approximation.  Joel Veness, Peter Sunehag and Marcus Hutter, University of Alberta,             Canada

OPTIMISTIC AIXI.  Peter Sunehag and Marcus Hutter. Australian National University

Memory issues of intelligent agents.  Laurent Orseau, AgroParisTech, France & IDSIA, Switzerland

Space-Time Embedded Intelligence, Laurent Orseau, AgroParisTech, France & IDSIA, Switzerland

Differences between Kolmogorov Complexity and Solomonoff Probability: Consequences for AGI, Alexey Potapov, Andrew Svitenkov and Yurii Vinogradov, AIDEUS, Russian Federation

Extending Universal Intelligence Models with a Formal Notion of Representation. Alexey Potapov and Sergey Rodionov, AIDEUS, Russian Federation

Session 4: Conceptual and Contextual Issues

On measuring social intelligence: experiments on competition and cooperation.  Javier Insa, Jose Hernandez-Orallo and José-Luis Benacloch-Ayuso, Universidad Politécnica de Valencia, Spain

An Extensible Language Interface for Robot Manipulation.  Jonathan Connell, Etienne Marcheret, Sharath Pankanti, Michiharu Kudoh and Risa Nishiyama, IBM T.J. Watson Research, US

Noisy Reasoners: Errors of Judgment in Humans and AIs.  Fintan Costello, University College Dublin, Ireland

Avoiding Unintended AI Behaviors.  Bill Hibbard, University of Wisconsin – Madison, USA

Decision Support for Safe AI Design.  Bill Hibbard, University of Wisconsin – Madison, USA

Session 5: Cognitive Architectures & Cognitive Modeling C

On the functional contributions of emotion mechanisms to (artificial) cognition and intelligence, Serge Thill and Robert Lowe, University of Skövde, Sweden

Perceptual time, perceptual reality, and general intelligence.  Leslie Smith, University of Stirling, United Kingdom

Creativity, Cognitive Mechanisms, and Logic.  Ahmed M. H. Abdel-Fattah.  University of Osnabrueck, Institute of Cognitive Science, Germany

Stupidity and the Ouroboros Model, Knut Thompson

Transparent Neural Networks: Integrating Concept Formation and Reasoning.   Claes Strannegård, Christian Balkenius, Olle Häggström and Johan Wessberg, Chalmers University of Technology, Sweden

Binary Space Partitioning as Intrinsic Reward.  Wojciech Skaba, AGINAO, Poland

Session 6: Mathematical Formalisms & Tools

A Framework for Representing Action Meaning in Artificial Systems via Force Dimensions.            Paul Hemeren and Peter Gärdenfors, University of Skövde, Sweden

Logical Prior Probabilities, Abram Demski, University of Southern California, USA

Fuzzy-probabilistic logic for common sense.  King Yin Yan, General Intelligence Research Group, Hong Kong

Fractal Analogies for General Intelligence.  Keith Mcgreggor and Ashok Goel. Georgia Institute of Technology, United States

An Intelligent Theory of Cost for Partial Metric Spaces.  Steve Matthews and Michael Bukatin, University of Warwick, UK

A Representation Theorem for Decisions about Causal Models.            Daniel Dewey, Future of Humanity Institute, UK

Modular Value Iteration Through Regional Decomposition. Linus Gisslen, Mark Ring, Matt Luciw and Juergen Schmidhuber, IDSIA, Switzerland & TU Munich, Germany

Special Session on AGI and Neuroscience

Toward tractable AGI: Challenges for System Identification in Neural Circuitry. Randal Koene, Carboncopies.org, USA.

The Connectome, WBE and AGI.  Diana Deca, IMPRS-LS, Germany

Hippocampal formation mechanism will inspire frame generation for building an artificial general intelligence, Hiroshi Yamakawa, FUJITSU LABORATORIES LTD., Japan           

What Is It Like to Be a Brain Simulation? Eray Ozkural, Bilkent University, Turkey

For registration (for one or both events), and accommodation at St. Anne's College, please see:

http://www.winterintelligence.org/travel-and-registration/

For any queries please contact:
[email protected]

“We can only see a short distance ahead, but we can see plenty there that needs to be done. “ Alan Turing.