Robust clustering using a kNN mode seeking ensemble

Abstract

In this paper we present a new algorithm for parameter-free clustering by mode seeking. Mode seeking, especially in the form of the mean shift algorithm, is a widely used strategy for clustering data, but at the same time prone to poor performance if the parameters are not chosen correctly. We propose to form a clustering ensemble consisting of repeated and bootstrapped runs of the recent kNN mode seeking algorithm, an algorithm which is faster than ordinary mean shift and more suited for high dimensional data. This creates a robust mode seeking clustering algorithm with respect to the choice of hyper-parameters and high dimensional input spaces, while at the same inheriting all other strengths of mode seeking in general. We demonstrate promising results on a number of synthetic and real data sets.

Prediction of daily global solar radiation and air temperature using six machine learning algorithms; a case of 27 European countries

“…This variation may be due to the effect of humans on nature. These results confirm research conducted by de Larsen et al (Larsen et al, 2020) in 2020, and results of other studies detailed (Maillo et al, n.d.;Saikia et al, n.d.;Erdal and Aytug, 2016;Min et al, 2003;Nordhaug et al, 2018;Pardis et al, 2019).…”

“…the class that is predicted by the largest number of individual classifiers is selected as the prediction) (Ho, n.d.). In the case of regression, the aggregation is performed by performing the average of the individual predictions (Maillo et al, n.d.;Saikia et al, n.d.;Erdal and Aytug, 2016;Nordhaug et al, 2018).…”

Trajectory feature extraction and multi‐criteria k nearest neighbour based job‐to‐crowd matching for the crowdshipping last mile delivery

“…Garcia and Maureira propose a kNN quantum cuckoo search algorithm for solving the multi-dimensional knapsack problem [9]. Myhre et al propose a robust clustering method, which uses a kNN-seeking model to ensemble the clusters [10]. Different from the conventional kNN method, their method can divide the cluster not only based on the density of the data but also considering the statistical characteristic (mean shift) to further divide the cluster into smaller clusters with higher member similarity.…”

Prediction of natural gas consumption in residential and commercial buildings; a case of 5 big cities located in the United States

“…the class that is predicted by the largest number of individual classifiers is selected as the prediction) [28]. In the case of regression, the aggregation is performed by performing the average of the individual predictions [29][30].…”