A review of instance selection methods

Abstract

In supervised learning, a training set providing previously known information is used to classify new instances. Commonly, several instances are stored in the training set but some of them are not useful for classifying therefore it is possible to get acceptable classification rates ignoring non useful cases; this process is known as instance selection. Through instance selection the training set is reduced which allows reducing runtimes in the classification and/or training stages of classifiers. This work is focused on presenting a survey of the main instance selection methods reported in the literature.

Selecting training sets for support vector machines: a review

“…There exist generic training set selection techniques [also referred to as instance selection algorithms in the literature (Olvera-López et al 2010)], and those designed for other classifiers [k-nearest neighbors (Angiulli 2005), neural networks (Reeves and Taylor 1998), and many other (Hernandez-Leal et al 2013;Wenyuan et al 2013)], but-due to the specific characteristics of the SVM training process and operation-the majority of SVM training set selection algorithms are crafted for this classifier. In this review, we summarize the state-of-the-art algorithms for selecting SVM training data from large datasets.…”

A Conceptual Framework for Data Quality in Knowledge Discovery Tasks (FDQ-KDT): A Proposal

“… Redundancy: as the name implies is the redundant information such as duplicate instances and derived attributes of others that contain the same information [26], [27]. In Table 3 are shown the approaches to solve the problems related to amount of data and redundancy.…”

“…[26], [27] It is worth mentioning that construction of reports (step of SEMMA methodology) is taking into account in all phases of FDQ-KDT.…”

“…Several approaches exist to tackle the issues of data quality in outliers [22], noise [18], inconsistency [23], incompleteness [24], redundancy [26], [27], amount of data [28]- [30], heterogeneity [14], and timeliness [25]. Nevertheless the results to date not consider resolve the issues in ensemble.…”

A Fast Hybrid k-NN Classifier Based on Homogeneous Clusters

“…DRTs [18,5,20,9,6,2,13] reduce the computational cost of classification by building a small representative set of the initial training data, called the Condensing Set (CS). The idea behind DRTs is to apply the k-NN classifier over S. Ougiaroglou is supported by a scholarship from the Greek Scholarships Foundations (I.K.Y.…”