UTCS Artificial Intelligence
courses
talks/events
demos
people
projects
publications
software/data
labs
areas
admin
Using a Sequential SOM to Parse Long-Term Dependencies (1999)
Marshall R. Mayberry III
and
Risto Miikkulainen
Simple Recurrent Networks (SRNs) have been widely used in natural language processing tasks. However, their ability to handle long-term dependencies between sentence constituents is somewhat limited. NARX networks have recently been shown to outperform SRNs by preserving past information in explicit delays from the network's prior output. However, it is unclear how the number of delays should be determined. In this study on a shift-reduce parsing task, we demonstrate that comparable performance can be derived more elegantly by using a SARDNET self-organizing map. The resulting architecture can represent arbitrarily long sequences and is cognitively more plausible.
View:
PDF
,
PS
Citation:
In
Proceedings of the 21st Annual Conference of the Cognitive Science Society
, Martin Hahn and Scott C. Stoness (Eds.), pp. 367-372 1999. Hillsdale, NJ: Erlbaum.
Bibtex:
@InProceedings{mayberry:cogsci99, title={Using a Sequential SOM to Parse Long-Term Dependencies}, author={Marshall R. Mayberry III and Risto Miikkulainen}, booktitle={Proceedings of the 21st Annual Conference of the Cognitive Science Society}, editor={Martin Hahn and Scott C. Stoness}, publisher={Hillsdale, NJ: Erlbaum}, pages={367-372}, url="http://www.cs.utexas.edu/users/ai-lab?mayberry:cogsci99", year={1999} }
People
Marshall R. Mayberry III
Ph.D. Alumni
marty mayberry [at] gmail com
Risto Miikkulainen
Faculty
risto [at] cs utexas edu
Areas of Interest
Cognitive Science
Natural Language Processing (Cognitive)
Labs
Neural Networks