A Representation of Programs for Learning and Reasoning
Draft — This content has not been approved for publication.
← Back to Publications
A Representation of Programs for Learning and Reasoning
Authors:
Year: 2008
Venue: Second Conference on Artificial General Intelligence (AGI-09)
Links: paper (PDF)
Summary
Argues that computer programs are a natural representation for the relational, typed, hierarchical, and complex-functional structure of real-world data, and proposes a programmatic representation — well-specified, compact, combinatorial, and hierarchical — for AGI systems that must represent, learn, and reason about programs and their naturally-occurring variations.
Relevance to Hyperon
A foundational MOSES-lineage source by Moshe Looks (the creator of MOSES) on why and how programs should be represented for learning and reasoning in integrative AGI — underpinning the program-representation choices described in the MOSES Deep Dive. (Companion to the 2009 Atlantis-published version.)