Interactive correlation clustering

Geerts, Floris; Ndindi, Reuben

doi:10.1109/dsaa.2014.7058069

Cited by 3 publications

(5 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, we provide a thorough analysis of the algorithm for correctness, which was not present in [11] and obtain an approximation guarantee matching our previous algorithm, in the worst case, and improves on our previous algorithm in most cases. These analyses are more challenging than for the algorithm given in [11]. Furthermore, we have revised and included additional experiments on real and synthetic datasets (Sect.…”

Section: Introductionmentioning

confidence: 86%

“…This work extends [11] by dealing with both deletions and relabeling of edges (Sect. 3) and by describing how our algorithm for the Bounded Correlation problem can be used in the context of a user interaction process (Sect.…”

Section: Introductionmentioning

confidence: 99%

“…6). More importantly, the approximation algorithm reported in [11] is substantially improved. In particular, our new algorithm BMulticut now implements an "adaptive" optimization strategy in which the underlying integer program (and its relaxation) is updated between two region growing steps provided that this is expected to lead to a better solution (Sect.…”

Section: Introductionmentioning

confidence: 99%

“…7). Improvements on the quality of approximate solutions have been observed in comparison to the algorithm given in [11]. Organization of the paper.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Bounded correlation clustering

Geerts

Ndindi

2016

Int J Data Sci Anal

Self Cite

View full text Add to dashboard Cite

Correlation clustering is to partition a set of objects into clusters such that the number of false positives and negatives is minimized. In this paper, we combine correlation clustering and user interaction. More specifically, we allow the user to control the quality of the clustering by providing error bounds on the number of false positives and negatives. If no clustering exists that satisfies these bounds, a set of edges is returned for user inspection such that the deletion or relabeling of these edges guarantees the existence of a clustering consistent with the error bounds. A user, however, may reject the deletion or relabeling of certain edges and ask for an alternative set of edges to be provided. If no such set of edges exists, a minimal change to the error bounds should be provided, after which the interactive process continues. The focus of this paper is on the algorithmic challenges involved in returning a minimal set of edges to the user. More specifically, we formalize the Bounded Correlation Clustering problem and show that it is intractable. Therefore, we propose an approximation algorithm based on the well-known region growing technique. We experimentally validate the efficiency and accuracy of the approximation algorithm.

show abstract

Section: Introductionmentioning

confidence: 86%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…7). Improvements on the quality of approximate solutions have been observed in comparison to the algorithm given in [11]. Organization of the paper.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Bounded correlation clustering

Geerts

Ndindi

2016

Int J Data Sci Anal

Self Cite

View full text Add to dashboard Cite

show abstract

“…The must/cannot links are specified when the user drags and drops images into new clusters [66], and in Wang and Davidson [107], violation of clustering is considered unsuitable, which may incur quality penalty. In Vikram and Dasgupta [103], the user specifies constraint triplets such as ({a,b},c), meaning "cluster should contain a and b but not c." In Geerts and Ndindi [46], the must/cannot links are specified by letting the user update a graph, resulting in deleting or re-labeling the edges until the user is satisfied. Most often, this functionality is provided directly through the user interface.…”

Section: Add Constraintsmentioning

confidence: 99%

Interactive Clustering

et al. 2020

View full text Add to dashboard Cite

In this survey, 105 papers related to interactive clustering were reviewed according to seven perspectives: (1) on what level is the interaction happening, (2) which interactive operations are involved, (3) how user feedback is incorporated, (4) how interactive clustering is evaluated, (5) which data and (6) which clustering methods have been used, and (7) what outlined challenges there are. This article serves as a comprehensive overview of the field and outlines the state of the art within the area as well as identifies challenges and future research needs.

show abstract