IPC분류정보
| 국가/구분 |
United States(US) Patent
등록
|
| 국제특허분류(IPC7판) |
|
| 출원번호 |
US-0813556
(1991-12-26)
|
|
발명자
/ 주소 |
- Villarreal James A. (Friendswood TX) Shelton Robert O. (Houston TX)
|
| 출원인 / 주소 |
- The United States of America as represented by the Administrator of the National Aeronautics and Space Administration (Washington DC 06)
|
| 인용정보 |
피인용 횟수 :
35 인용 특허 :
0 |
초록
▼
Neural network algorithms have impressively demonstrated the capability of modelling spatial information. On the other hand, the application of parallel distributed models to processing of temporal data has been severely restricted. The invention introduces a novel technique which adds the dimension
Neural network algorithms have impressively demonstrated the capability of modelling spatial information. On the other hand, the application of parallel distributed models to processing of temporal data has been severely restricted. The invention introduces a novel technique which adds the dimension of time to the well known back-propagation neural network algorithm. In the space-time neural network disclosed herein, the synaptic weights between two artificial neurons (processing elements) are replaced with an adaptable-adjustable filter. Instead of a single synaptic weight, the invention provides a plurality of weights representing not only association, but also temporal dependencies. In this case, the synaptic weights are the coefficients to the adaptable digital filters.
대표청구항
▼
A processing element (i) for use in a space-time neural network for processing both spacial and temporal data, wherein the neural network comprises a plurality of layers of said processing elements, the plurality of layers comprising a first layer and at least one additional layer, the network furth
A processing element (i) for use in a space-time neural network for processing both spacial and temporal data, wherein the neural network comprises a plurality of layers of said processing elements, the plurality of layers comprising a first layer and at least one additional layer, the network further comprising connections between processing elements of the first layer and processing elements of an additional layer: each said processing element adapted to receive a sequence of signal inputs X(n), X(n-1), X(n-2) . . . , each input X(n) comprising K signal components x1(n), x2(n), . . . xj(n), . . . xk(n), each said processing element comprising, in combination: (a) a plurality K of adaptable filters (F1i, F2i, . . . Fji, . . . Fki) each filter Fji having an input for receiving a respective component xj(n), xj(n-1), xj(n-2), . . . , of said sequence of inputs, where xj(n) is the most current input component, and providing a filter output yj(n) in response to the input xj(n) which is given by: yj(n)=f(amjYj(n-m), bkjXj(n-k)), where amj and bkj are coefficients of the filter Fji and f denotes the operation of the filter; (b) a junction, coupled to each of said adaptive filters, providing a non-linear output pi(Si(n)) in response to the filter outputs yj(n) which is given by: pi(Si(n))=f(yj(n)), where Si(n) is the sum of the filter outputs, whereby said junction presents a sequence of output signals, pi(Si(n)), pi(Si(n-1)), pi(Si(n-2)).
※ AI-Helper는 부적절한 답변을 할 수 있습니다.