The exponential growth of digital image data has given rise to the need of efficient content management and retrieval tools. Currently, there is a lack of tools for processing the collected unlabeled data in a schematic manner. K-means is one of the most widely used clustering methods and has been applied in a variety of fields, one of them being image sorting. Although a useful tool for image management, the K-means method is heavily influenced by initializations, the most important one being the need to know the number of clusters a priori. A number of different methods have been proposed for identifying the correct number of clusters for K-means, one of them being the variance ratio criterion (VRC). Despite its popularity, the VRC method comes with two very important shortcomings: it only yields good results when the data dimensionality is low and it does not scale well for a high number of clusters, making it very difficult to use in computer vision applications. We propose an extension to the VRC method that works for increased cluster number and high-dimensionality data sets and therefore is fit for image data sets.
BerkhinP. A survey of clustering data mining techniques. In: JKoganCNicholasMTeboulle, et al. (eds) Grouping multidimensional data. Berlin: Springer2006, pp. 25–71.
3.
KassRERafteryAE.Bayes factors. J Am Stat Assoc1995; 90(430): 773–795.
4.
BozdoganH.Model selection and Akaike’s information criterion (AIC): the general theory and its analytical extensions. Psychometrika1987; 52(3): 345–370.
5.
DunnJC.A fuzzy relative of the ISODATA process and its use in detecting compact well-separated clusters. J Cybernetics1973; 3(3): 32–57.
RousseeuwPJ.Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. J Comp Appl Math1987; 20: 53–65.
8.
TibshiraniRWaltherGHastieT.Estimating the number of clusters in a data set via the gap statistic. J R Stat Soc B2001; 63(2): 411–423.
9.
YaoJCaoXZhaoQ, et al. Robust subspace clustering via penalized mixture of Gaussians. Neurocomputing2018; 278: 4–11.
10.
GengXGengYMuS, et al. An improved K-means algorithm based on fuzzy metrics. IEEE Access2020; 8: 217416–217424.
11.
AwadFHHamadMM.Improved K-means clustering algorithm for big data based on distributed smartphoneneural engine processor. Electronics2022; 11(6): 883.
12.
LeiJQinTPengB, et al. Reducing background induced domain shift for adaptive person re-identification. IEEE Trans Ind Inform2022; 19(6): 7377–7388.
13.
LinGZhengZChenL, et al. Multi-modal 3D shape clustering with dual contrastive learning. Appl Sci2022; 12(15): 7384.
14.
IshtiaqAAhmedSKhanMF, et al. Intelligent clustering using moth flame optimizer for vehicular ad hoc networks. Int J Distrib Sens Netw2019; 15(1): 155014771882446.
15.
YaariRHuppertADattnerI.Data-driven clustering of infectious disease incidence into age groups. Stat Methods Med Res2022; 31(12): 2486–2499.
16.
KingMLHeringASAguilarOM.Building predictive models of counterinsurgent deaths using robust clustering and regression. J Def Model Simul2016; 13(4): 449–465.
17.
CalińskiTHarabaszJ.A dendrite method for cluster analysis. Commun Stat Theor Methods1974; 3(1): 1–27.
18.
McQueenJB. Some methods of classification and analysis of multivariate observations. In: Proceedings of the fifth Berkeley symposium on mathematical statistics and probability, Berkeley, CA, 21 June–18 July 1965 and 27 December 1965–7 January 1966, pp. 281–297. Berkeley, CA: University of California Press.
19.
HartiganJAWongMA.Algorithm AS 136: a K-means clustering algorithm. J R Stat Soc C1979; 28(1): 100–108.
20.
PellegDMooreAW.X-means: extending K-means with efficient estimation of the number of clusters. ICML2000; 1: 727–734.
IshiokaT. An expansion of X-means for automatically determining the optimal number of clusters. In: Proceedings of the fourth IASTED international conference computational intelligence, Calgary, AB, Canada, 4–6 July 2005.
23.
PellegDMooreA.Accelerating exact K-means algorithms with geometric reasoning. In: Proceedings of the 5th ACM SIGKDD international conference on knowledge discovery and data mining, San Diego, CA, 15–18 August 1999, pp. 277–281. New York: Association for Computing Machinery.
24.
XieFXuY.Bayesian repulsive Gaussian mixture model. J Am Stat Assoc2020; 115(529): 187–203.
25.
PetraliaFRaoVDunsonD.Repulsive mixtures. Adv Neural Inf Process Syst2012; 25.
26.
ChengZZhangHTChenMZ, et al. Aggregation pattern transitions by slightly varying the attractive/repulsive function. PLoS ONE2011; 6(7): e22123.
27.
Kenyon-DeanKCianfloneAPage-CacciaL, et al. Clustering-oriented representation learning with attractive-repulsive loss. arXiv [Preprint] arXiv:1812.07627, 2018.
28.
EdwardsAWFCavalli-SforzaLL. A method for cluster analysis. Biometrics1965; 21: 362–375.
29.
KästerTWendtVSagererG.Comparing clustering methods for database categorization in image retrieval. In: BMichaelisGKrell (eds) Joint pattern recognition symposium. Berlin: Springer, 2003, pp. 228–235.
30.
AddagarlaSKAmalanathanA.Probabilistic unsupervised machine learning approach for a similar image recommender system for E-Commerce. Symmetry2020; 12(11): 1783.
31.
KermaniSSamadzadehaghdamNEtehadTavakolM.Automatic color segmentation of breast infrared images using a Gaussian mixture model. Optik2015; 126(21): 3288–3294.
32.
McCallumANigamKUngarLH. Efficient clustering of high-dimensional data sets with application to reference matching. In: Proceedings of the 6th ACM SIGKDD international conference on knowledge discovery and data mining, Boston, MA, 20–23 August 2000, pp. 169–178. New York: Association for computing machinery.
33.
MardiaKV.Multi-dimensional multivariate Gaussian Markov random fields with application to image processing. J Multivariate Anal1988; 24(2): 265–284.
34.
VidalRMaYSastrySS.Principal component analysis. In: YMaSSastryRVidal (eds) Generalized principal component analysis. New York: Springer, 2016, pp. 25–62.
35.
BerryMWBrowneM.Understanding search engines: mathematical modeling and text retrieval. Philadelphia, PA: SIAM, 2005.
36.
BerryMWDumaisSTO’BrienGW.Using linear algebra for intelligent information retrieval. SIAM Rev1995; 37(4): 573–595.
37.
AgrawalRFaloutsosCSwamiA.Efficient similarity search in sequence databases. In: International conference on foundations of data organization and algorithms, Chicago, IL, 13–15 October 1993, pp. 69–84. New York: Springer.
38.
KeoghEKeoghEMehrotraS, et al. Locally adaptive dimensionality reduction for indexing large time series databases. In: Proceedings of the 2001 ACM SIGMOD international conference on management of data, Santa Barbara, CA, 21–24 May 2001, pp. 151–162. New York: Association for Computing Machinery.
39.
JiaWSunMLianJ, et al. Feature dimensionality reduction: a review. Complex Intell Syst2022; 8(3): 2663–2693.
40.
SalehBElgammalA.Large-scale classification of fine-art paintings: learning the right metric on the right feature. arXiv [Preprint]. arXiv:1505.00855, 2015.
41.
IandolaFNHanSMoskewiczMW, et al. SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and¡ 0.5 MB model size. arXiv [Preprint]. arXiv:1602.07360, 2016.
42.
DengJDongWSocherR, et al. ImageNet: a large-scale hierarchical image database. In: IEEE conference on computer vision and pattern recognition, Miami, FL, 20–25June2009, pp. 248–255. New York: IEEE.
