Benchmarking in Cluster Analysis: A Study on Spectral Clustering, DBSCAN, and K-Means

Murugesan, Nivedha; Cho, Irene; Tortora, Cristina

doi:10.1007/978-3-030-60104-1_20

Cited by 9 publications

(7 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Spectral clustering is a clustering method with foundations in algebraic graph theory (Jia et al 2014). It has been shown that spectral clustering has better overall performance across different areas of applications (Murugesan et al 2021). Given a graph G, the spectral clustering decomposition of G can be represented by the equation L = D − A , where L is the Laplacian, D is the degree (i.e., a diagonal matrix with the number of edges incident to each node), and A the adjacency matrices of G. Spectral clustering uses, say, the n eigenvectors associated to the n smallest nonzero eigenvalues of L .…”

Section: Spectral Clusteringmentioning

confidence: 99%

Feature extraction with spectral clustering for gene function prediction using hierarchical multi-label classification

et al. 2022

View full text Add to dashboard Cite

Gene annotation addresses the problem of predicting unknown associations between gene and functions (e.g., biological processes) of a specific organism. Despite recent advances, the cost and time demanded by annotation procedures that rely largely on in vivo biological experiments remain prohibitively high. This paper presents a novel in silico approach for to the annotation problem that combines cluster analysis and hierarchical multi-label classification (HMC). The approach uses spectral clustering to extract new features from the gene co-expression network (GCN) and enrich the prediction task. HMC is used to build multiple estimators that consider the hierarchical structure of gene functions. The proposed approach is applied to a case study on Zea mays, one of the most dominant and productive crops in the world. The results illustrate how in silico approaches are key to reduce the time and costs of gene annotation. More specifically, they highlight the importance of: (1) building new features that represent the structure of gene relationships in GCNs to annotate genes; and (2) taking into account the structure of biological processes to obtain consistent predictions.

show abstract

Section: Spectral Clusteringmentioning

confidence: 99%

Feature extraction with spectral clustering for gene function prediction using hierarchical multi-label classification

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Figure 2 shows examples of challenging data sets on which spectral clustering is able to detect the correct clustering structure. For more examples and comparisons, see [5].…”

Section: Background: Spectral Clusteringmentioning

confidence: 99%

“…Although there is not a single clustering technique that is always preferred to the others, spectral clustering has the advantage of detecting clusters of arbitrary shapes, such as non-convex clusters, and of giving good results when clusters overlap. Spectral clustering does not always give the best performance but it is always among the better techniques [5]. One of the main limitations of spectral clustering in practical applications is that it requires continuous data, whereas many data sets have mixed-type data, i.e., continuous and categorical variables.…”

Section: Introductionmentioning

confidence: 99%

Spectral Clustering of Mixed-Type Data

Mbuga

Tortora

2021

Stats

View full text Add to dashboard Cite

Cluster analysis seeks to assign objects with similar characteristics into groups called clusters so that objects within a group are similar to each other and dissimilar to objects in other groups. Spectral clustering has been shown to perform well in different scenarios on continuous data: it can detect convex and non-convex clusters, and can detect overlapping clusters. However, the constraint on continuous data can be limiting in real applications where data are often of mixed-type, i.e., data that contains both continuous and categorical features. This paper looks at extending spectral clustering to mixed-type data. The new method replaces the Euclidean-based similarity distance used in conventional spectral clustering with different dissimilarity measures for continuous and categorical variables. A global dissimilarity measure is than computed using a weighted sum, and a Gaussian kernel is used to convert the dissimilarity matrix into a similarity matrix. The new method includes an automatic tuning of the variable weight and kernel parameter. The performance of spectral clustering in different scenarios is compared with that of two state-of-the-art mixed-type data clustering methods, k-prototypes and KAMILA, using several simulated and real data sets.

show abstract

“…Murugesan et al (Murugesan et al, 2021) and Artés et al (Artés et al, 2019) investigated DBSCAN (Density-Based Spatial Clustering of Applications with Noise) to detect spatial clusters of fire occurrences. DBSCAN (Esther et al, 1996) is a density-based algorithm that separates highdensity from low-density clusters based on data proximity.…”

Section: Introductionmentioning

confidence: 99%

“…DBSCAN uses two tuning parameters: "epsilon'' and "minPts", that determine the radius and minimum number of points in a cluster. This approach effectively helps find high-density fire clusters for benchmarking (Murugesan et al, 2021) and monitoring systems (Artés et al, 2019).…”

Section: Introductionmentioning

confidence: 99%

A Study on Enhanced Spatial Clustering Using Ensemble Dbscan and Umap to Map Fire Zone in Greater Jakarta, Indonesia

Hasana,

Fitrianah

2023

J. Ris. Inform.

View full text Add to dashboard Cite

This research investigated ensemble clustering algorithms and dimensionality reduction for fire zone mapping, specifically DBSCAN + UMAP. We evaluated six clustering methods: DBSCAN, ensemble DBSCAN, DBSCAN + UMAP, ensemble DBSCAN + UMAP, HDBSCAN and Gaussian Mixture Model (GMM). We evaluated our results based on the Silhouette Score and the Davies-Bouldin (DB) index, emphasizing handling irregular cluster shapes, smaller clusters and resolving incompact clusters. Our findings suggested that ensemble DBSCAN + UMAP outperformed five other methods with zero noise clusters indicating clustering results are resistant to outliers, leading to a clearer identification of fire-prone areas, a high Silhouette Score of 0.971, indicating accurate cluster separation of distinct areas of potential fire hazards and an exceptionally low DB Index of 0.05 that indicates compact clusters to identify well-defined and geographically concentrated areas prone to fire hazards. Our findings contribute to the advanced techniques for minimizing the impacts of fires and improving fire hazard assessments in Indonesia.

show abstract

Benchmarking in Cluster Analysis: A Study on Spectral Clustering, DBSCAN, and K-Means

Cited by 9 publications

References 15 publications

Feature extraction with spectral clustering for gene function prediction using hierarchical multi-label classification

Feature extraction with spectral clustering for gene function prediction using hierarchical multi-label classification

Spectral Clustering of Mixed-Type Data

A Study on Enhanced Spatial Clustering Using Ensemble Dbscan and Umap to Map Fire Zone in Greater Jakarta, Indonesia

Contact Info

Product

Resources

About