DocumentCode
2785949
Title
Adaptive connection to the world through self-organizing sensors and effectors
Author
Cariani, Peter
Author_Institution
Massachusetts Eye & Ear Infirmary, Boston, MA, USA
fYear
1990
fDate
5-7 Sep 1990
Firstpage
73
Abstract
The realm of devices composed of adaptive sensing, computing, and effecting elements is surveyed, and contemporary devices are categorized accordingly. A new class of adaptive devices, those which adaptively construct their sensors and effectors, is presented. Such complement those which perform adaptive computations and could be used for adaptively constructing appropriate feature and action primitives for connectionist devices, thereby determining the real-world semantic categories within which the computational optimizations take place. Networks of such semantically adaptive elements would be capable of open-ended generation of new, orthogonal signal types, not logically reducible to preexisting types. An alternative conception of neurons as adaptive temporal elements utilizing the dynamic properties of excitable membranes is briefly outlined
Keywords
adaptive systems; neural nets; neurophysiology; self-adjusting systems; action primitives; adaptive computations; adaptive devices; adaptive sensing; adaptive temporal elements; computational optimizations; connectionist devices; excitable membranes; orthogonal signal types; real-world semantic categories; self-organizing sensors; semantically adaptive elements; Biomembranes; Ear; Laboratories; Machine learning; Neural networks; Neurons; Organisms; Predictive models; Sensor phenomena and characterization; Signal generators;
fLanguage
English
Publisher
ieee
Conference_Titel
Intelligent Control, 1990. Proceedings., 5th IEEE International Symposium on
Conference_Location
Philadelphia, PA
ISSN
2158-9860
Print_ISBN
0-8186-2108-7
Type
conf
DOI
10.1109/ISIC.1990.128442
Filename
128442
Link To Document