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 Videos

Nonlinear mapping networks.

D K Agrafiotis¹, V S Lobanov

¹3-Dimensional Pharmaceuticals Inc, Exton, Pennsylvania 19341, USA. dimitris@3dp.com

Journal of Chemical Information and Computer Sciences

|January 11, 2000

Summary

This summary is machine-generated.

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

Multiobjective optimization of combinatorial libraries.

Molecular diversity·2003

Same author

Multiobjective optimization of combinatorial libraries.

Journal of computer-aided molecular design·2002

Same author

Variable selection for QSAR by artificial ant colony systems.

SAR and QSAR in environmental research·2002

Same author

High-density miniaturized thermal shift assays as a general strategy for drug discovery.

Journal of biomolecular screening·2002

Same author

Combinatorial networks.

Journal of molecular graphics & modelling·2001

Same author

Design and prioritization of plates for high-throughput screening.

Journal of chemical information and computer sciences·2001

Same journal

Future Papers.

Journal of chemical information and computer sciences·2016

Same journal

Future Papers.

Journal of chemical information and computer sciences·2016

Same journal

Future Papers.

Journal of chemical information and computer sciences·2016

Same journal

Future Papers.

Journal of chemical information and computer sciences·2016

Same journal

Future Papers.

Journal of chemical information and computer sciences·2016

Same journal

Future Papers.

Journal of chemical information and computer sciences·2016

See all related articles

A new method combines nonlinear mapping with neural networks to process large datasets. This approach enables dimensionality reduction for big data, overcoming computational limits of traditional techniques.

Area of Science:

Computational statistics
Machine learning
Data mining

Background:

Multidimensional scaling and nonlinear mapping are powerful dimensionality reduction techniques.
Their quadratic complexity limits scalability to large datasets.
Existing methods struggle with high-dimensional, large-scale data analysis.

Purpose of the Study:

To develop a scalable dimensionality reduction technique for large datasets.
To overcome the computational limitations of traditional nonlinear mapping methods.
To enable the application of nonlinear mapping to data mining challenges.

Main Methods:

A novel approach combining nonlinear mapping with feed-forward neural networks.
Utilizes probability sampling and a classical algorithm for initial projection.

Related Experiment Videos

Employs a multilayer neural network trained with back-propagation for learning the nonlinear transform.

Main Results:

The neural network approach allows processing datasets orders of magnitude larger than conventional methods.
Projections generated are virtually indistinguishable from those of traditional approaches.
Demonstrated effectiveness in image processing and combinatorial chemistry applications.

Conclusions:

This hybrid method significantly enhances the scalability of nonlinear mapping.
Neural network encoding of nonlinear transforms makes it applicable to large-scale data mining.
Enables computationally intractable data mining applications to become feasible.