ms@thalamus.sans.kth.se (Magnus Stensmo) (05/09/91)
Archive-name: ai/neural-nets/ms-icann91/1991-04-29 Archive: thalamus.sans.kth.se:/pub/ms-icann91.ps.Z [130.237.219.100] Original-posting-by: ms@thalamus.sans.kth.se (Magnus Stensmo) Original-subject: Re: Connectionist expert systems (& Paper announcement) Reposted-by: emv@msen.com (Edward Vielmetti, MSEN) In article <1991Apr4.074916.28091@cs.utk.edu> raj@utkux1.utk.edu (Rajendra Patil) writes: >Hi friends: > > Anyone working with connectionist expert systems (S. Gallant) >I will like to get some idea about the details of implementations. >Almost any information (references, code etc) is appreciated. > The following paper is accepted for oral presentation at the International Conference on Artificial Neural Networks ICANN-91 in Helsinki, Finland, June 24-28, 1991 A QUERY-REPLY SYSTEM BASED ON A RECURRENT BAYESIAN NEURAL NETWORK Magnus Stensmo, Anders Lansner, and Bj\"orn Levin Studies of Artificial Neural Systems Department of Numerical Analysis and Computing Science Royal Institute of Technology, S-100 44 Stockholm, Sweden Abstract A query-reply system based on a fully-connected recurrent Bayesian neural network is presented. A database describing the classes to classify and their attributes is loaded onto the network. Each attribute and class are coded using two network units, where one unit represents the existence of an attribute or class and the other its non-existence. Initially the system asks about the property it is most undetermined about. After a while a hypothesis is formed and a verifying question phase is entered. Inconsistencies in user input are detected and corrected. The paper is available using anonymous ftp from thalamus.sans.kth.se as pub/ms-icann91.ps.Z If this fails please send me mail or e-mail and I will post it to you. Please try ftp first as stamps are expensive to us. Magnus Stensmo ms@sans.kth.se Dept. of Numerical Analysis and Computing Science ms%sans@sekth.bitnet Royal Institute of Technology Phone: +46 8 790 6908 S-100 44 Stockholm, Sweden Fax: +46 8 790 0930 -- Magnus Stensmo ms@sans.kth.se Dept. of Numerical Analysis and Computing Science ms%sans@sekth.bitnet Royal Institute of Technology Phone: +46 8 790 6908 S-100 44 Stockholm, Sweden Fax: +46 8 790 0930 -- comp.archives file verification thalamus.sans.kth.se -rw-r--r-- 1 0 0 61529 Apr 29 14:06 /pub/ms-icann91.ps.Z found ms-icann91 ok thalamus.sans.kth.se:/pub/ms-icann91.ps.Z