Title :
Inheritance reasoning in connectionist networks
Author :
Jones, Melissa A
Author_Institution :
AT&T Bell Lab., Murray Hill, NJ, USA
Abstract :
Summary form only given, as follows. A bidirectional network model is described for inheritance reasoning which processes queries by combinations of top-down and bottom-up reasoning. The model, which is based on theoretical work in nonmonotonic reasoning, permits multiple inheritance paths in acyclic inheritance theories and allows an arbitrary preference relation among the inferences in the theory (to handle exceptions, for example). Unlike other inheritance models which compute extensions serially (maximally consistent models), the network gains substantially more parallelism by simultaneously reasoning in multiple extensions when possible.<>
Keywords :
inference mechanisms; neural nets; bidirectional network model; bottom-up reasoning; connectionist networks; inference mechanisms; inheritance reasoning; multiple inheritance paths; neural nets; nonmonotonic reasoning; preference relation; top-down reasoning; Inference mechanisms; Neural networks;
Conference_Titel :
Neural Networks, 1989. IJCNN., International Joint Conference on
Conference_Location :
Washington, DC, USA
DOI :
10.1109/IJCNN.1989.118335