Perspectives of Neural-Symbolic IntegrationBarbara Hammer, Pascal Hitzler Springer, 14. 8. 2007. - 319 страница The human brain possesses the remarkable capability of understanding, - terpreting, and producing human language, thereby relying mostly on the left hemisphere. The ability to acquire language is innate as can be seen from d- orders such as speci?c language impairment (SLI), which manifests itself in a missing sense for grammaticality. Language exhibits strong compositionality and structure. Hence biological neural networks are naturally connected to processing and generation of high-level symbolic structures. Unlike their biological counterparts, arti?cial neural networks and logic do not form such a close liason. Symbolic inference mechanisms and statistical machine learning constitute two major and very di?erent paradigms in ar- ?cial intelligence which both have their strengths and weaknesses: Statistical methods o?er ?exible and highly e?ective tools which are ideally suited for possibly corrupted or noisy data, high uncertainty and missing information as occur in everyday life such as sensor streams in robotics, measurements in medicine such as EEG and EKG, ?nancial and market indices, etc. The m- els, however, are often reduced to black box mechanisms which complicate the integration of prior high level knowledge or human inspection, and they lack theabilitytocopewitharichstructureofobjects,classes,andrelations. S- bolic mechanisms, on the other hand, are perfectly applicative for intuitive human-machine interaction, the integration of complex prior knowledge, and well founded recursive inference. Their capability of dealing with uncertainty andnoiseandtheire?ciencywhenaddressingcorruptedlargescalereal-world data sets, however, is limited. Thus, the inherent strengths and weaknesses of these two methods ideally complement each other. |
Садржај
Comparing Sequence Classification Algorithms | 23 |
Mining StructureActivity Relations in Biological Neural | 49 |
Adaptive Contextual Processing of Structured Data | 67 |
Markovian Bias of Neuralbased Architectures | 95 |
Logic and Neural Networks | 181 |
Connectionist Model Generation | 204 |
Learning Models of Predicate Logical Theories with Neural | 233 |
Connectionist Representation of MultiValued Logic | 283 |
315 | |
Друга издања - Прикажи све
Чести термини и фразе
activation algorithm allows annotated applications approach approximation architecture arrows Artificial associated Assume atom called clauses cognitive complex compute connectionist connections consider construction contains context corresponding defined Definition denote depends described dynamic elements encoding equations error example extended fact Figure finite first-order fixed function given graph grasping implemented inference input integration Intelligence International interpretation introduced kernel knowledge label layer learning length logic programs machine means memory methods neural networks neurons nodes Note objects operator output parameters performance position possible prediction presented problem processing properties protein reasoning recurrent relation represent representation respectively robot rules Science semantic sequence shown shows similar simple space step string structure symbolic Table task theory tion transductions transition tree true types unit University values variables vector vertex weights
Популарни одломци
Страница xiii - Department of Electronics and Computer Science University of Southampton, Southampton SO17 1BJ, UK +44(0)1703592418 {sgb97r, dder}@ecs.soton.ac.uk 1.
Страница xiii - International Computer Science Institute 1947 Center Street, Suite 600 Berkeley, CA 94704...
Страница 9 - ... more similar they are - rather than only considering contiguous n-grams, the degree of contiguity of the subsequence in the input string s determines how much it will contribute to the comparison. In order to deal with non-contiguous substrings, it is necessary to introduce a decay factor A € (0,1) that can be used to weight the presence of a certain feature in a string.