0000000000171766
AUTHOR
José Salvador Sánchez
Learning vector quantization with alternative distance criteria
An adaptive algorithm for training of a nearest neighbour (NN) classifier is developed in this paper. This learning rule has some similarity to the well-known LVQ method, but uses the nearest centroid neighbourhood concept to estimate optimal locations of the codebook vectors. The aim of this approach is to improve the performance of the standard LVQ algorithms when using a very small codebook. The behaviour of the learning technique proposed here is experimentally compared to those of the plain k-NN decision rule and the LVQ algorithms.
From the nearest neighbour rule to decision trees
This paper proposes an algorithm to design a tree-like classifier whose result is equivalent to that achieved by the classical Nearest Neighbour rule. The procedure consists of a particular decomposition of a d-dimensional feature space into a set of convex regions with prototypes from just one class. Some experimental results over synthetic and real databases are provided in order to illustrate the applicability of the method.
Improving the k-NCN classification rule through heuristic modifications
Abstract This paper presents an empirical investigation of the recently proposed k-Nearest Centroid Neighbours ( k -NCN) classification rule along with two heuristic modifications of it. These alternatives make use of both proximity and geometrical distribution of the prototypes in the training set in order to estimate the class label of a given sample. The experimental results show that both alternatives give significantly better classification rates than the k -Nearest Neighbours rule, basically due to the properties of the plain k -NCN technique.
Using proximity and spatial homogeneity in neighbourhood-based classifiers
In this paper, a set of neighbourhood-based classifiers are jointly used in order to select a more reliable neighbourhood of a given sample and take an appropriate decision about its class membership. The approaches introduced here make use of two concepts: proximity and symmetric placement of the samples.
On the use of neighbourhood-based non-parametric classifiers
Alternative non-parametric classification schemes, which come from the use of different definitions of neighbourhood, are introduced. In particular, the Nearest Centroid Neighbourhood along with the neighbourhood relation derived from the Gabriel Graph and the Relative Neighbourhood Graph are used to define the corresponding (k-)Nearest Neighbour-like classifiers. Experimental results are reported to compare the performance of the approaches proposed here to the one obtained with the k-Nearest Neighbours rule.
Editing prototypes in the finite sample size case using alternative neighborhoods
The recently introduced concept of Nearest Centroid Neighborhood is applied to discard outliers and prototypes 111 class overlapping regions in order to improve the performance of the Nearest Neighbor rule through an editing procedure, This approach is related to graph based editing algorithms which also define alternative neighborhoods in terms of geornetric relations, Classical editing algorithms are compared to these alternative editing schemes using several synthetic and real data problems. The empirical results show that, the proposed editing algorithm constitutes a good trade-off among performance and computational burden.
Restricted Decontamination for the Imbalanced Training Sample Problem
The problem of imbalanced training data in supervised methods is currently receiving growing attention. Imbalanced data means that one class is much more represented than the others in the training sample. It has been observed that this situation, which arises in several practical domains, may produce an important deterioration of the classification accuracy, in particular with patterns belonging to the less represented classes. In the present paper, we report experimental results that point at the convenience of correctly downsizing the majority class while simultaneously increasing the size of the minority one in order to balance both classes. This is obtained by applying a modification o…
Prototype selection for the nearest neighbour rule through proximity graphs
Abstract In this paper, the Gabriel and Relative Neighbourhood graphs are used to select a suitable subset of prototypes for the Nearest Neighbour rule. Experiments and results are reported showing the effectiveness of the method and comparing its performance to those obtained by classical techniques.