Weitere Kapitel dieses Buchs durch Wischen aufrufen
Useful IT systems may operate in domains in which results are fundamentally equivocal, and this is at odds with the correct-or-incorrect basis of computing technology. In reality, all IT system results are to some degree equivocal, and, instead of retrofitting conventional discrete technology to accommodate this feature, we might compute in terms of well-founded continua – precise approximationcomputation (PAC). An IT system with PAC-based components can determine (from the manner in which the computation proceeds) whether a computed result is suspect or not, and valuable use could be made of this ‘knowledge’. PAC, in several manifestations, appears to offer the possibility of providing the necessary ‘self-knowledge’. Like neural-computing technology, the precise approximations technologies are currently limited in scope and would apply to specialist components of IT systems rather to IT systems as a whole. The move to continuous probability distributions or spaces does circumvent some of the problems associated with discrete computation, but it also introduces issues of discrete interpretation, or summaries, of the continuities computed. Although the precise approximation strategies illustrated are based on sound statistical principles, fundamental assumptions are involved, and these need to be thoroughly researched before full weight can put on the probabilistic continua computed.
Bitte loggen Sie sich ein, um Zugang zu diesem Inhalt zu erhalten
Sie möchten Zugang zu diesem Inhalt erhalten? Dann informieren Sie sich jetzt über unsere Produkte:
- A Computer That Knows When It’s Wrong
- Springer London
- Chapter 23
Neuer Inhalt/© ITandMEDIA, Product Lifecycle Management/© Eisenhans | vege | Fotolia