High-Level Connectionist Models
Abstract
The major achievement of this semiannum was the significant revision and extension of the Recursive Auto-Associative Memory (RAAM) work for publication in the journal Artificial Intelligence. Included as an appendix to this report, the article includes several new elements: (1) Background - The work was more clearly set into the area of recursive distributed representations, machine learning, and the adequacy of the connectionist approach for high-level cognitive modeling; (2) New Experiment - RAAM was applied to finding compact representations for sequences of letters; (3) Analysis - The developed representations were analyzed as features which range from categorical to distinctive. Categorical features distinguish between conceptual categories while distinctive features vary within categories and discriminate or label the members. The representations were also analyzed geometrically; and 4) Applications - Feasibility studies were performed and described on inference by association, and on using RAAM-generated patterns along with cascaded networks for natural language parsing. Both of these remain long-term goals of the project. (kr)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1989
- Accession Number
- ADA216581
Entities
People
- Jordan B. Pollack
Organizations
- Ohio State University