Now published: The SP theory of intelligence: distinctive features and advantages (PDF, http://bit.ly/21gv2jT).

0 views

Skip to first unread message

Gerry Wolff

unread,

Mar 8, 2016, 9:28:21 AM3/8/16

to ai-robotics, Artificial General Intelligence, Big Data Europe, EUCog General News, euRobotics AISBL mailing list, Genetic Programming group (Yahoo), Machine Learning News, Reinforcement Learning Mailing List

The paper: The SP theory of intelligence: distinctive features and advantages (PDF, J G Wolff, IEEE Acess, 4, 216-246, 2016, http://bit.ly/21gv2jT) has recently been published. It is an open-access paper that may be downloaded for no charge.

The paper includes an outline of the SP theory and the computer model in which it embodied, with enough detail to ensure that the rest of the paper makes sense.

As its title suggests, the paper describes what is distinctive about the SP theory. But more importantly, it describes why the SP theory provides firmer foundations for the development of AI than the main AI-related alternatives.

In particular, the paper describes several problems with deep learning and several problems with the Watson system, and how those problems may be overcome in the SP system.

Of course it makes sense to continue work in promising areas like deep learning and Watson. But there is now a strong case for developing parallel strands of research in the development of the SP system. How this may be done is described in a Proposal for the creation of a research facility for the development of the SP machine (PDF, http://bit.ly/1zZjjIs).

If such a facility would be of interest to you, please let me know. This would be helpful in demonstrating to funding agencies that the proposed system has support.