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

Bayes-optimality motivated linear and multilayered perceptron-based dimensionality reduction.

R Lotlikar¹, R Kothari

¹Artificial Neural Systems Laboratory, Department of Electrical and Computer Engineering and Computer Science, University of Cincinnati, Cincinnati, OH 45221-0030, USA.

IEEE Transactions on Neural Networks

|February 6, 2008

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

Particle composition and morphology over urban environment (New Delhi): Plausible effects on wheat leaves.

Environmental research·2021

Same author

Recurrent erythema nodosum leprosum associated with Mycobacterium indicus pranii vaccine in a case of leprosy: a rare paradox.

Journal of the European Academy of Dermatology and Venereology : JEADV·2021

Same author

Use of polypropylene mesh in contaminated and dirty strangulated hernias: short-term results.

Hernia : the journal of hernias and abdominal wall surgery·2018

Same author

Rechallenge temozolomide in glioma: A case series from India.

Indian journal of cancer·2017

Same author

Patterns of care and outcomes for second-line targeted therapy in metastatic renal cell carcinomas: A registry based analysis.

Indian journal of cancer·2017

Same author

Phycoremediation of textile wastewater by unicellular microalga Chlorella pyrenoidosa.

Cellular and molecular biology (Noisy-le-Grand, France)·2014

Same journal

Universal perceptron and DNA-like learning algorithm for binary neural networks: LSBF and PBF implementations.

IEEE transactions on neural networks·2013

Same journal

Guest editorial: special section on white box nonlinear prediction models.

IEEE transactions on neural networks·2011

Same journal

Data-based fault-tolerant control of high-speed trains with traction/braking notch nonlinearities and actuator failures.

IEEE transactions on neural networks·2011

Same journal

Guest editorial: special section on data-based control, modeling, and optimization.

IEEE transactions on neural networks·2011

Same journal

Neural network-based multiple robot simultaneous localization and mapping.

IEEE transactions on neural networks·2011

Same journal

Data-driven model-free adaptive control for a class of MIMO nonlinear discrete-time systems.

IEEE transactions on neural networks·2011

See all related articles

This study introduces new dimensionality reduction methods optimizing classification accuracy by approximating Bayes error. The proposed nonlinear method, using a multilayered perceptron, significantly outperforms linear approaches and Fisher's linear discriminant.

Area of Science:

Machine Learning
Pattern Recognition
Data Science

Background:

Dimensionality reduction maps high-dimensional data to lower dimensions, improving classification reliability and performance by removing redundant information.
Current methods often rely on scatter matrices or data statistics, which do not directly correlate with classification accuracy.
Bayes error is the ideal optimality criterion for classification, but it is analytically challenging to express.

Purpose of the Study:

To develop novel dimensionality reduction techniques optimized for classification accuracy.
To formulate both linear and nonlinear dimensionality reduction methods based on an approximation of Bayes error.
To evaluate the performance of these new methods against existing techniques.

Main Methods:

Related Experiment Videos

Proposed an optimality criterion based on an approximation of Bayes error.
Formulated a linear dimensionality reduction method.
Developed a nonlinear dimensionality reduction method utilizing a multilayered perceptron to generate the lower-dimensional representation.

Main Results:

The nonlinear method demonstrated superiority over the linear method, effectively unfolding nonlinear manifolds.
The proposed nonlinear method provided significantly better lower-dimensional representations for classification compared to Fisher's linear discriminant (FLD) and other nonlinear methods.
The approach successfully addressed the limitations of existing dimensionality reduction techniques that do not directly optimize for classification performance.

Conclusions:

The developed dimensionality reduction methods, particularly the nonlinear approach, offer improved performance for classification tasks.
Approximating Bayes error provides a more effective optimality criterion for dimensionality reduction than traditional scatter matrix methods.
The nonlinear multilayered perceptron-based method represents a significant advancement in creating reliable lower-dimensional data representations for classification.