Learning temporary variable binding with dynamic links

Jurgen Schmidhuber

doi:10.1109/ijcnn.1991.170693

Learning temporary variable binding with dynamic links

Jurgen Schmidhuber

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

1 Scopus citations

Abstract

A novel gradient-based system for processing sequential time-varying inputs and outputs is described. With the method it is possible to train a system with time-varying inputs and outputs to use its dynamic links for temporarily binding variable contents to variable names as long as it is necessary for solving a particular task. Various learning methods for nonstationary environments are derived. Two experiments with unknown time delays illustrate the approach. A by-product of this work is the demonstration that a system consisting of two feedforward networks can solve tasks that only dynamic recurrent networks were supposed to solve.

Original language	English (US)
Title of host publication	1991 IEEE International Joint Conference on Neural Networks - IJCNN '91
Publisher	Publ by IEEEPiscataway
Pages	2075-2079
Number of pages	5
ISBN (Print)	0780302273
DOIs	https://doi.org/10.1109/ijcnn.1991.170693
State	Published - Jan 1 1991
Externally published	Yes

Access to Document

10.1109/ijcnn.1991.170693

Cite this

@inproceedings{ffde062a5674441cb78e0f1a09e27b65,

title = "Learning temporary variable binding with dynamic links",

abstract = "A novel gradient-based system for processing sequential time-varying inputs and outputs is described. With the method it is possible to train a system with time-varying inputs and outputs to use its dynamic links for temporarily binding variable contents to variable names as long as it is necessary for solving a particular task. Various learning methods for nonstationary environments are derived. Two experiments with unknown time delays illustrate the approach. A by-product of this work is the demonstration that a system consisting of two feedforward networks can solve tasks that only dynamic recurrent networks were supposed to solve.",

author = "Jurgen Schmidhuber",

note = "Generated from Scopus record by KAUST IRTS on 2022-09-14",

year = "1991",

month = jan,

day = "1",

doi = "10.1109/ijcnn.1991.170693",

language = "English (US)",

isbn = "0780302273",

pages = "2075--2079",

booktitle = "1991 IEEE International Joint Conference on Neural Networks - IJCNN '91",

publisher = "Publ by IEEEPiscataway",

}

TY - GEN

T1 - Learning temporary variable binding with dynamic links

AU - Schmidhuber, Jurgen

N1 - Generated from Scopus record by KAUST IRTS on 2022-09-14

PY - 1991/1/1

Y1 - 1991/1/1

N2 - A novel gradient-based system for processing sequential time-varying inputs and outputs is described. With the method it is possible to train a system with time-varying inputs and outputs to use its dynamic links for temporarily binding variable contents to variable names as long as it is necessary for solving a particular task. Various learning methods for nonstationary environments are derived. Two experiments with unknown time delays illustrate the approach. A by-product of this work is the demonstration that a system consisting of two feedforward networks can solve tasks that only dynamic recurrent networks were supposed to solve.

AB - A novel gradient-based system for processing sequential time-varying inputs and outputs is described. With the method it is possible to train a system with time-varying inputs and outputs to use its dynamic links for temporarily binding variable contents to variable names as long as it is necessary for solving a particular task. Various learning methods for nonstationary environments are derived. Two experiments with unknown time delays illustrate the approach. A by-product of this work is the demonstration that a system consisting of two feedforward networks can solve tasks that only dynamic recurrent networks were supposed to solve.

UR - https://ieeexplore.ieee.org/document/170693/

UR - http://www.scopus.com/inward/record.url?scp=0026287754&partnerID=8YFLogxK

U2 - 10.1109/ijcnn.1991.170693

DO - 10.1109/ijcnn.1991.170693

M3 - Conference contribution

SN - 0780302273

SP - 2075

EP - 2079

BT - 1991 IEEE International Joint Conference on Neural Networks - IJCNN '91

PB - Publ by IEEEPiscataway

ER -

Learning temporary variable binding with dynamic links

Abstract

Access to Document

Other files and links

Fingerprint

Cite this