On using prototype reduction schemes to optimize dissimilarity-based classification

Kim, Sang‐Woon; Oommen, B. John

doi:10.1016/j.patcog.2007.03.006

Cited by 21 publications

(26 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, Lozano et al [13] employed prototype optimization methods often applied in vector spaces, such as editing and condensing, for constructing more general dissimilarity-based classifiers. Kim and Oommen [8] used the well-known condensed nearest neighbor rule [14] to reduce the original training set before computing the dissimilarity-based classifiers on the entire data. Other new methods have been evolved to be applied in the dissimilarity space, such as Kcentres, Edicon, ModeSeek, Featsel and a genetic algorithm [6,9].…”

Section: Prototype Selection Methodsmentioning

confidence: 99%

See 1 more Smart Citation

One-Sided Prototype Selection on Class Imbalanced Dissimilarity Matrices

Millán-Giraldo

García

Sánchez

2012

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. In the dissimilarity representation paradigm, several prototype selection methods have been used to cope with the topic of how to select a small representation set for generating a low-dimensional dissimilarity space. In addition, these methods have also been used to reduce the size of the dissimilarity matrix. However, these approaches assume a relatively balanced class distribution, which is grossly violated in many real-life problems. Often, the ratios of prior probabilities between classes are extremely skewed. In this paper, we study the use of renowned prototype selection methods adapted to the case of learning from an imbalanced dissimilarity matrix. More specifically, we propose the use of these methods to under-sample the majority class in the dissimilarity space. The experimental results demonstrate that the one-sided selection strategy performs better than the classical prototype selection methods applied over all classes.

show abstract

Section: Prototype Selection Methodsmentioning

confidence: 99%

“…In this context, prototype selection constitutes one of the most active research lines, which has primarily been addressed in two ways: (i) finding a small representation set capable of generating a low-dimensional dissimilarity space [4,6,7], and (ii) reducing the original dissimilarity matrix [8,9].…”

Section: D(t T ) To D(t R)mentioning

confidence: 99%

One-Sided Prototype Selection on Class Imbalanced Dissimilarity Matrices

Millán-Giraldo

García

Sánchez

2012

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…Inside the family of supervised classifiers, we can find the nearest neighbour (NN) rule method [1,2] that predicts the class of a new prototype by computing a similarity [3,4] measure between it and all prototypes from the training set, called the k-nearest neighbours (k-NN) classifier. Recent studies show that k-NN classifier could be improved by employing numerous procedures.…”

Section: Introductionmentioning

confidence: 99%

A memetic algorithm for evolutionary prototype selection: A scaling up approach

García

Cano

Herrera

2008

Pattern Recognition

169

View full text Add to dashboard Cite

Prototype selection problem consists of reducing the size of databases by removing samples that are considered noisy or not influential on nearest neighbour classification tasks. Evolutionary algorithms have been used recently for prototype selection showing good results. However, due to the complexity of this problem when the size of the databases increases, the behaviour of evolutionary algorithms could deteriorate considerably because of a lack of convergence. This additional problem is known as the scaling up problem.Memetic algorithms are approaches for heuristic searches in optimization problems that combine a population-based algorithm with a local search. In this paper, we propose a model of memetic algorithm that incorporates an ad hoc local search specifically designed for optimizing the properties of prototype selection problem with the aim of tackling the scaling up problem. In order to check its performance, we have carried out an empirical study including a comparison between our proposal and previous evolutionary and non-evolutionary approaches studied in the literature.The results have been contrasted with the use of non-parametric statistical procedures and show that our approach outperforms previously studied methods, especially when the database scales up. ᭧

show abstract

“…The major questions we encountered when designing DBCs are summarized as follows: (1) how to select prototypes; (2) how to measure dissimilarities between object samples; and (3) how to design classifiers in the dissimilarity space. Several strategies have been used to explore these questions [3], [4], [5]. The details of the strategies are omitted here, but we now attempt to explain the first question in the present paper.…”

Section: Introductionmentioning

confidence: 99%

“…In these methods, a training set, T , is pruned to yield a set of representative prototypes, Y , where, without loss of generality, |Y | ≤ |T |. On the other hand, by invoking a prototype reduction scheme (PRS), Kim and Oommen [5] also obtained a representative subset, Y , which is utilized by the DBC. Aside from using PRSs, Kim and Oommen simultaneously proposed the use of the Mahalanobis distance as the dissimilarity-measurement criterion.…”

Section: Introductionmentioning

confidence: 99%

On Using Dimensionality Reduction Schemes to Optimize Dissimilarity-Based Classifiers

Kim

Gao

2008

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Abstract. The aim of this paper is to present a strategy by which a new philosophy for pattern classification pertaining to dissimilarity-based classifiers (DBCs) can be efficiently implemented. Proposed by Duin and his co-authors, DBCs are a way of defining classifiers among classes; they are not based on the feature measurements of individual patterns, but rather on a suitable dissimilarity measure among the patterns. The problem with this strategy is that we need to select a representative set of data that is both compact and capable of representing the entire data set. However, it is difficult to find the optimal number of prototypes and, furthermore, selecting prototype stage may potentially lose some useful information for discrimination. To avoid these problems, in this paper, we propose an alternative approach where we use all available samples from the training set as prototypes and subsequently apply dimensionality reduction schemes. That is, we prefer not to directly select the representative prototypes from the training samples; rather, we use a dimensionality reduction scheme after computing the dissimilarity matrix with the entire training samples. Our experimental results demonstrate that the proposed mechanism can improve the classification accuracy of conventional approaches for two real-life benchmark databases.

show abstract

On using prototype reduction schemes to optimize dissimilarity-based classification

Cited by 21 publications

References 17 publications

One-Sided Prototype Selection on Class Imbalanced Dissimilarity Matrices

One-Sided Prototype Selection on Class Imbalanced Dissimilarity Matrices

A memetic algorithm for evolutionary prototype selection: A scaling up approach

On Using Dimensionality Reduction Schemes to Optimize Dissimilarity-Based Classifiers

Contact Info

Product

Resources

About