An iterative method for training multilayer networks with threshold functions.

An iterative method for training multilayer networks with threshold functions. Research Abstract Details 

Research Abstract Table of Contents

Jump to the:

An iterative method for training multilayer networks with threshold functions. Abstract Text:

E M Corwin,A M Logar,W B Oldham,

Concerns the problem of finding weights for feed-forward networks in which threshold functions replace the more common logistic node output function. The advantage of such weights is that the complexity of the hardware implementation of such networks is greatly reduced. If the task to be learned does not change over time, it may be sufficient to find the correct weights for a threshold function network off-line and to transfer these weights to the hardware implementation. This paper provides a mathematical foundation for training a network with standard logistic function nodes and gradually altering the function to allow a mapping to a threshold unit network. The procedure is analogous to taking the limit of the logistic function as the gain parameter goes to infinity. It is demonstrated that, if the error in a trained network is small, a small change in the gain parameter will cause a small change in the network error. The result is that a network that must be implemented with threshold functions can first be trained using a traditional back propagation network using gradient descent, and further trained with progressively steeper logistic functions. In theory, this process could require many repetitions. In simulations, however, the weights have be successfully mapped to a true threshold network after a modest number of slope changes. It is important to emphasize that this method is only applicable to situations for which off-line learning is appropriate.

An iterative method for training multilayer networks with threshold functions. Publishing Authors By Initials

EM Corwin,AM Logar,WB Oldham,

For similar abstracts research abstracts see: abstracts research

PUBMED ID PMID:

MEDLINE DATE:

An iterative method for training multilayer networks with threshold functions. Journal Published:

PUBLICATION TYPE: Journal Article

Journal: IEEE transactions on neural networks / a publicati

VOLUME: 5

Page Numbers: 507-8

Journal Abbreviation:

ISSN: 1045-9227

DAY: 12

MONTH: 02

YEAR: 1994

An iterative method for training multilayer networks with threshold functions. Information

Number of References:

LANGUAGE: eng

NlmUniqueID: 101211035

An iterative method for training multilayer networks with threshold functions. Keywords Mesh Terms:

KEYWORDS:

MESH TERMS:

Chemical & Substance for Abstract: An iterative method for training multilayer networks with threshold functions. Information

Substance Name:

Registry Number:

Grant and Affiliation Information for An iterative method for training multilayer networks with threshold functions.

AFFILIATION: Dept. of Math. and Comput. Sci., South Dakota Sch. of Mines and Technol., Rapid City, SD.

Country: United States

AGENCY:

GRANT:

ACRONYM:

MEDLINETA: IEEE Trans Neural Netw

REFSOURCE:

DATABASENAME:

ACCESSION NUMBER:

Number Hits: 0

An iterative method for training multilayer networks with threshold functions Related Publications

• An iterative method for training multilayer networks with threshold functions.
• Aneurysm of the ductus arteriosus in the neonate.
• Benchmarking a novel ultrasound-CT fusion system for respiratory motion management in radiotherapy: assessment of spatio-temporal characteristics and comparison to 4DCT.
• Biotherapy: the fourth modality of cancer treatment.
• Cardiac volvulus: plain film recognition of an often fatal condition.
• Cervical spine ROM measurements: optimizing the testing protocol by using a 3D ultrasound-based motion analysis system.
• Comorbid personality disorder and treatment use in a community sample of youths: a 20-year follow-up.
• Comparison of Analytical and Numerical Particle Deposition Using Commercial CFD Packages: Impaction and Sedimentation.
• Corrections to "image recovery and segmentation using competitive learning in a layered network.
• Damage-Limited Lifetime of 193-nm Lithography Tools as a Function of System Variables.
• Drosophila aging 2006/2007.
• Editorial: 2008: state of affairs for cancer biotherapy & radiopharmaceuticals.
• Functional imaging in bulk tissue specimens using optical emission tomography: fluorescence preservation during optical clearing.
• High-dose methotrexate in combination chemotherapy for small cell lung cancer.
• INSULIN TREATMENT OF SCHIZOPHRENIA--A CONTROLLED STUDY.
• Image recovery and segmentation using competitive learning in a layered network.
• Inhalation of concentrated ambient particulate matter near a heavily trafficked road stimulates antigen-induced airway responses in mice.
• Insulin treatment of schizophrenia. A three-year follow-up of a controlled study.
• Launching DSM-V.
• Managing broiler litter application rate and grazing to decrease watershed runoff losses.
• Managing broiler litter application rate and grazing to decrease watershed runoff losses.
• Mycetoma of the lung in a cavity that developed from a gunshot wound.
• Preface.
• Quantitative determination of lysozyme-ligand binding in the solution and gas phases by electrospray ionisation mass spectrometry.
• Research and treatment synergism.
• The aetiology of depression and anxiety.
• The brighter side.
• The problem of consistent radiological diagnosis in coalminers' pneumoconiosis; an experimental study.
• The psychophysics of clinical pain.
• The reliability of isometric strength and fatigue measures in patients with knee osteoarthritis.