Abstract

The K-nearest neighbour classifier is very effective and simple non-parametric technique in pattern classification; however, it only considers the distance closeness, but not the geometricalplacement of the k neighbors. Also, its classification performance is highly influenced by the neighborhood size k and existing outliers. In this paper, we propose a new local mean based k-harmonic nearest centroid neighbor (LMKHNCN) classifier in orderto consider both distance-based proximity, as well as spatial distribution of k neighbors. In our method, firstly the k nearest centroid neighbors in each class are found which are used to find k different local mean vectors, and then employed to compute their harmonic mean distance to the query sample. Lastly, the query sample is assigned to the class with minimum harmonic mean distance. The experimental results based on twenty-six real-world datasets shows that the proposed LMKHNCN classifier achieves lower error rates, particularly in small sample-size situations, and that it is less sensitive to parameter k when compared to therelated four KNN-based classifiers.

Highlights

  • KNN [1] is a traditional non-parametric, and most famous, technique among machine learning algorithms [2,3,4]

  • As revealed by the other comparative methods in almost all twenty-six real-world data sets. This is because the classification results, the proposed local mean based k-harmonic nearest centroid neighbor (LMKHNCN) classifier performs better than the other comparative concept of local centroid mean vector and harmonic mean distance similarity used in the proposed methods in almost all twenty-six real-world data sets

  • This is because the concept of local centroid method makes it focus on more reliable local mean vectors with smaller distances to the unknown mean vector and harmonic mean distance similarity used in the proposed method makes it focus samples in each class

Read more

Summary

Introduction

KNN [1] is a traditional non-parametric, and most famous, technique among machine learning algorithms [2,3,4]. The first matter is that KNN classification performance is affected by existing outliers, especially in small training sample-size situations [22]. This implies that one has to pay attention in selecting a suitable value for neighborhood size k [23]. To overcome the influence of outliers, a local mean-based k nearest neighbor (LMKNN) classifier has been introduced in [3]. As LMKNN shows significant performance in response to existing outliers, its concept further applies to distance metric learning [24], group-based classification [6], and discriminant analysis [25].

Objectives
Methods
Results
Conclusion
Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call