Prototype generation based on MOPSO

Prototype generation based on MOPSO

For access to this article, please select a purchase option:

Buy chapter PDF
(plus tax if applicable)

IET members benefit from discounts to all IET publications and free access to E&T Magazine. If you are an IET member, log in to your account and the discounts will automatically be applied.

Learn more about IET membership 

Recommend Title Publication to library

You must fill out fields marked with: *

Librarian details
Your details
Why are you recommending this title?
Select reason:
Swarm Intelligence - Volume 3: Applications — Recommend this title to your library

Thank you

Your recommendation has been sent to your librarian.

When classifying a test instance, the nearest neighbor classifier will consume lots of time because it needs to search the whole training set for the instance's nearest neighbors. Prototype generation is a widely used approach to improve its time efficiency which generates a small set of prototypes to classify a test instance instead of using the whole training set. This paper applies the particle swarm optimization (PSO) to prototype generation and presents two novel methods to improve the classifier's performance. A fitness function named error rank is proposed to enhance the nearest neighbor classifier's generalization ability. In order to keep the classifier from overfitting the training set, this paper proposes the multiobjective optimization strategy which divides the whole training set into several subsets and regards the performance criterion on each subset as an objective function of the multiobjective PSO. The multiobjective optimization strategy pursues the performance over multiple subsets simultaneously, resulting in better generalization ability. Experimental results over 31 UCI datasets and 59 additional datasets show that the proposed algorithm achieves state-of-the-art performance.

Chapter Contents:

  • Abstract
  • 1.1 Introduction
  • 1.2 Related work
  • 1.3 Particle swarm optimization and its application to prototype generation
  • 1.3.1 Framework of PSO
  • 1.3.2 Multiobjective PSO
  • 1.3.3 PSO for prototype generation
  • 1.4 Error rank
  • 1.4.1 Motivation of error rank
  • 1.4.2 Definition of error rank
  • Rank distance normalization
  • Emphasizing instances with small rank distances
  • 1.5 Multiobjective optimization strategy for learning
  • 1.6 Experiments
  • 1.6.1 Experimental setup
  • 1.6.2 Comparison algorithms
  • 1.6.3 Hypothesis test
  • 1.6.4 Experimental results
  • Experimental analysis of error rank
  • Experimental analysis of the multiobjective optimization strategy
  • Experimental comparison with existing prototype generation algorithms
  • Training time consumption
  • 1.6.5 Comparison with 28 prototype generation algorithms on the 59 datasets offered by Triguero et al.
  • 1.7 Conclusions
  • References

Inspec keywords: pattern classification; particle swarm optimisation

Other keywords: training set; nearest neighbor classifier; particle swarm optimization; multiobjective optimization strategy; test instance; fitness function named error rank; prototype generation

Subjects: Optimisation techniques; Data handling techniques

Preview this chapter:
Zoom in

Prototype generation based on MOPSO, Page 1 of 2

| /docserver/preview/fulltext/books/ce/pbce119h/PBCE119H_ch1-1.gif /docserver/preview/fulltext/books/ce/pbce119h/PBCE119H_ch1-2.gif

Related content

This is a required field
Please enter a valid email address