From synapses to rules: The self-referential perspective

Apolloni, B; Biella, G; Stafylopatis, A

dc.contributor.author	Apolloni, B	en
dc.contributor.author	Biella, G	en
dc.contributor.author	Stafylopatis, A	en
dc.date.accessioned	2014-03-01T01:48:36Z
dc.date.available	2014-03-01T01:48:36Z
dc.date.issued	1999	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/25526
dc.relation.uri	http://www.scopus.com/inward/record.url?eid=2-s2.0-4944267170&partnerID=40&md5=2d9253e9316ce3e5e49bda8b32bcf25b	en
dc.subject	Neural networks	en
dc.subject	Rule extraction	en
dc.subject	Symbolic/subsymbolic knowledge	en
dc.subject.other	Algorithms	en
dc.subject.other	Artificial intelligence	en
dc.subject.other	Brain	en
dc.subject.other	Codes (symbols)	en
dc.subject.other	Constraint theory	en
dc.subject.other	Formal logic	en
dc.subject.other	Functions	en
dc.subject.other	Robustness (control systems)	en
dc.subject.other	Genetic changes	en
dc.subject.other	Rule extraction	en
dc.subject.other	Symbolic knowledge	en
dc.subject.other	Neural networks	en
dc.title	From synapses to rules: The self-referential perspective	en
heal.type	journalArticle	en
heal.publicationDate	1999	en
heal.abstract	We consider the extraction of formal knowledge from a trained neural network in the perspective of identifying this network within our brain and the final user of this information with our brain again. We first analyze theoretical issues - mainly coming from AI, but also from neurophysiology and information theory - on relations and links between subsymbolic and symbolic knowledge in our brain. From this analysis a bipartition derives of the considered algorithms. From one side, there are direct methods for discovering Horn clauses and extensions from trained networks, a usual subject in many review papers. From the other side, we will identify symbolic knowledge with tools for efficiently managing concepts discovered in subsymbolic way in a self-referential framework where a neural network is however the user of the concepts. At first glance, this alternative perspective would just reconsider the direct methods in respect to the functionalities of the hidden_to_output nodes connections. But exactly after self referentiality, discovering formal connection should require a heavy training of the involved neural network, namely: a training capable of simulating the architectural and parametric refinement achieved by our brain along millennia. This calls for algorithms for symbolical learning that comply with neurophysiological functionality constraints, but shorten the mentioned long training phase, by using facilities now available to our brain - such as a preexisting formal knowledge and the capabilty of generating suitable examples by ourself. By definition, the output of these algorithms is exactly the goal knowledge springing from neural networks we are searching for.	en
heal.publisher	World Scientific and Engineering Academy and Society	en
heal.journalName	Computational Intelligence and Applications	en
dc.identifier.spage	61	en
dc.identifier.epage	66	en