Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: Jul 7, 2026

Deep Neural Networks for Image-Based Dietary Assessment

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

Handwritten digit recognition by neural networks with single-layer training.

S Knerr¹, L Personnaz, G Dreyfus

¹Ecole Superieure de Phys. et de Chimie Ind. de la Ville de Paris.

IEEE Transactions on Neural Networks

|January 1, 1992

Summary

Single-layer neural network classifiers, using the STEPNET procedure, efficiently solve complex handwritten digit recognition problems. This method achieves performance comparable to more complex networks with appropriate data and learning rules.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Surgery of acute type A dissection: what have we learned during the past 25 years?

Zeitschrift fur Kardiologie·2016

Same author

Epoch-based Entropy for Early Screening of Alzheimer's Disease.

International journal of neural systems·2015

Same author

Finding a place for genomics in health disparities research.

Public health genomics·2012

Same author

Imaging the prosthetic valve sewing ring thrombosis using real-time three-dimensional transoesophageal echocardiography.

Netherlands heart journal : monthly journal of the Netherlands Society of Cardiology and the Netherlands Heart Foundation·2011

Same author

Abrupt change of Antarctic moisture origin at the end of Termination II.

Proceedings of the National Academy of Sciences of the United States of America·2010

Same author

Patient physical characteristics and primary care physician decision making in preconception genetic screening.

Public health genomics·2009

Area of Science:

Computer Science
Artificial Intelligence
Machine Learning

Background:

Neural network classifiers are powerful tools for complex classification tasks.
Real-world applications like handwritten digit recognition present significant challenges.
Existing complex networks can be computationally intensive.

Purpose of the Study:

To introduce and evaluate the STEPNET procedure for efficient neural network classification.
To demonstrate the effectiveness of single-layer training for complex problems.
To compare performance against more complex neural network architectures.

Main Methods:

The STEPNET procedure decomposes classification problems into simpler subproblems.
These subproblems are solved using linear separators.

More Related Videos

Artificial Intelligence-Based System for Detecting Attention Levels in Students

Artificial Intelligence-Based System for Detecting Attention Levels in Students

Published on: December 15, 2023

Related Experiment Videos

Last Updated: Jul 7, 2026

Deep Neural Networks for Image-Based Dietary Assessment

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

Artificial Intelligence-Based System for Detecting Attention Levels in Students

Artificial Intelligence-Based System for Detecting Attention Levels in Students

Published on: December 15, 2023

The study utilizes specific data representations and learning rules for training.

Main Results:

The STEPNET procedure was applied to two large digit databases (European and US Postal Service).
Performance comparable to more complex networks was achieved.
A hardware implementation of the classifier was developed.

Conclusions:

Single-layer neural network classifiers, when trained with the STEPNET procedure, offer an efficient solution for handwritten digit recognition.
The method demonstrates that complex problems can be effectively addressed by decomposing them into simpler, solvable subproblems.
The approach shows promise for both software and hardware implementations in real-world classification scenarios.