DOC PREVIEW
Chico CSCI 397 - Hybrid Neural Systems

This preview shows page 1-2-3-4-5-6 out of 18 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 18 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

Hybrid Neural SystemsStefan WermterRon SunSpringer, Heidelberg, New YorkJanuary 2000PrefaceThe aim of this book is to present a broad spectrum of current research inhybrid neural systems, and advance the state of the art in neural networks andartificial intelligence. Hybrid neural systems are computational systems whichare based mainly on artificial neural networks but which also allow a symbolicinterpretation or interaction with symbolic components.This book focuses on the following issues related to different types of rep-resentation: How does neural representation contribute to the success of hybridsystems? How does symbolic representation supplement neural representation?How can these types of representation be combined? How can we utilize theirinteraction and synergy? How can we develop neural and hybrid systems for newdomains? What are the strengths and weaknesses of hybrid neural techniques?Are current principles and methodologies in hybrid neural systems useful? Howcan they be extended? What will be the impact of hybrid and neural techniquesin the future?In order to bring together new and different approaches, we organized aninternational workshop. This workshop on hybrid neural systems, organized byStefan Wermter and Ron Sun, was held during December 4–5, 1998 in Denver.In this well-attended workshop, 27 papers were presented. Overall, the work-shop was wide-ranging in scope, covering the essential aspects and strands ofhybrid neural systems research, and successfully addressed many important is-sues of hybrid neural systems research. The best and most appropriate papercontributions were selected and revised twice. This book contains the best re-vised papers, some of which are presented as state-of-the-art surveys, to coverthe various research areas of the collection.This selection of contributions is a representative snapshot of the state of theart in current approaches to hybrid neural systems. This is an extremely activearea of research that is growing in interest and popularity. We hope that thiscollection will be stimulating and useful for all those interested in the area ofhybrid neural systems.We would like to thank Garen Arevian, Mark Elshaw, Steve Womble andin particular Christo Panchev, from the Hybrid Intelligent Systems Group ofthe University of Sunderland for their important help and assistance during thepreparations of the book. We would like to thank Alfred Hofmann from Springerfor his cooperation. Finally, and most importantly, we thank the contributors tothis book.January 2000Stefan WermterRon SunTable of ContentsAn overview of hybrid neural systems . . . . . . . . . . . . . . . . . . . . . 1S. Wermter and R. SunStructured Connectionism and Rule RepresentationLayered hybrid connectionist models for cognitive science 14Jerome Feldman and David BaileyTypes and quantifiers in SHRUTI: A connectionist modelof rapid reasoning and relational processing . . . . . . . . . . . 28Lokendra ShastriA recursive neural network for reflexive reasoning . . . . . . 46Steffen H¨olldobler, Yvonne Kalinke and J¨org WunderlichA novel modular neural architecture for rule-based andsimilarity-based reasoning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63Rafal Bogacz and Christophe Giraud-CarrierAddressing knowledge-representation issues in connec-tionist symbolic rule encoding for general inference . 78Nam Seog ParkTowards a hybrid model of first-order theory refinement 92Nelson A. Hallack, Gerson Zaverucha and Valmir C. BarbosaDistributed Neural Architectures and Language ProcessingDynamical recurrent networks for sequential data pro-cessing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108Stefan C. Kremer and John KolenFuzzy knowledge and recurrent neural networks: A dy-namical systems perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124Christian W. Omlin, Lee Giles and Karvel K. ThornberCombining maps and distributed representations forshift-reduce parsing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146Marshall R. Mayberry and Risto MiikkulainenTowards hybrid neural learning internet agents . . . . . . . . . . 160Stefan Wermter, Garen Arevian and Christo PanchevA connectionist simulation of the empirical acquisition ofgrammatical relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177William C. Morris, Garrison W. Cottrell and Jeffrey L. ElmanLarge patterns make great symbols: An example of learn-ing from example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194Pentti KanervaContext vectors: A step toward a grand unified represen-tation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204Stephen I. GallantIntegration of graphical rules with adaptive learning ofstructured information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212Paolo Frasconi, Marco Gori and Alessandro SperdutiTransformation and ExplanationLessons from past, current issues and future research di-rections in extracting the knowledge embedded in ar-tificial neural networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227Alan B. Tickle, Frederic Maire, Guido Bologna, Robert Andrews andJoachim DiederichSymbolic rule extraction from the DIMLP neural net-work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241Guido BolognaUnderstanding state space organization in recurrent neu-ral networks with iterative function systems dynamics 256Peter Tino, Georg Dorffner and Christian SchittenkopfDirect explanations and knowledge extraction from amultilayer perceptron network that performs low backpain classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271Marilyn L. Vaughn, Steven J. Cavill, Stewart J. Taylor, Michael A. Foyand Anthony J.B. FoggHigh order eigentensors as symbolic rules in competitivelearning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . …


View Full Document
Download Hybrid Neural Systems
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Hybrid Neural Systems and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Hybrid Neural Systems 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?