The Condensed Nearest Neighbor Rule

By C. Hilborn, D. Lainiotis. 1967

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

A Sequential Algorithm for Training Text Classifiers (1994, methodology): “Uncertainty sampling is similar to the strategy of training on misclassi ed instances [16, 17].”
Top 10 algorithms in data mining (2007, background): “This is known as ‘condensing’ and can greatly speed up the classification of new objects [35].”
An efficient instance selection algorithm for k nearest neighbor regression (2017, methodology): “In order to compare the performance between DISKR and other nstance selection algorithms, five representative algorithms for NN regression have been selected in this study: prototype se- ection using mutual information(PSMI), ensemble of threshold- ased CNN(TE-CNN), ensemble of threshold-based ENN (”
Experiments with a New Boosting Algorithm (1996, methodology): “Table 3 describes the results from experiments with AdaBoost (each experiment was repeated 10 times using different random seeds), the strawman algorithm (each repeated 7 times) , and CNN (7 times).”
Using evolutionary algorithms as instance selection for data reduction in KDD: an experimental study (2003, background): “Cnn [15]—It tries to find a consistent subset, which correctly classifies all of the remaining points in the sample set.”
Heterogeneous Uncertainty Sampling for Supervised Learning (1994, background): “Uncertainty sampling with a single classifier can also be viewed as a variation on the heuristic of training on misclassified instances [15, 33, 35].”
On the Class Imbalance Problem (2008, background): “So CNN is effective only binary classes are of small overlapping.”
A fuzzy K-nearest neighbor algorithm (1985, background): “Since its introduction, the ÄT-NN rule has been studied and improved upon by numerous researchers [13]-[18].”
On building ensembles of stacked denoising auto-encoding classifiers and their further improvement (2018, methodology): “If weights are restricted to {0, 1}, we have a sample selection scheme, which was the form applied in the seminal work of Hart with nearest neighbor classiﬁers ﬁfty years ago [66] .”
Prototype selection for dissimilarity-based classifiers (2006, methodology): “A direct approach to dissimilarities leads to the k-NN method [10].”
Classifier Ensembles for Changing Environments (2004, methodology): “This is the online version of the Hart’s editing algorithm [10].”
A study of the behaviour of linguistic fuzzy rule based classification systems in the framework of imbalanced data-sets (2008, methodology): “• " CNN + Tomek links " : It is similar to the OSS, but the method to find the consistent subset is applied before the Tomek links.”
k-Nearest Neighbour Classifiers (2007, methodology): “Ritter et al. [21] reported improvements on the CNN with their Selective Nearest Neighbour (SNN) which imposes the rule that every case in the training set must be closer to a case of the same class in the edited set than to any other training case of a different class.”
Classification in the Presence of Label Noise: A Survey (2014, background): “For example, the condensed nearest neighbors (CNN) rule [195] builds a subset of training instances that allows classifying correctly all other training instances.”