Werner Bailer
ABSTRACT
In order to compactly represent a set of data, its medoid (the element with minimum summed distance to all other elements) is a useful choice. This has applications in clustering, compression and visualisation of data. In multimedia data, the set of data is often sampled as a sequence in time or space, such as a video shot or views of a scene. The exact calculation of the medoid may be costly, especially if the distance function between elements is not trivial. While approximation methods for medoid selection exist, we show in this work that they do not perform well on sequences of images. We thus propose a novel algorithm for efficiently selecting an approximate medoid of a temporal sequence and assess its performance on two large-scale video data sets.
- 2015. Call for Proposals for Compact Descriptors for Video Analysis (CDVA) -- Search and Retrieval. Technical Report ISO/IEC JTC1/SC29/WG11/N15339.Google Scholar
- 2015. ISO/IEC 15938-13, Information technology -- Multimedia content description interface -- Part 13: Compact descriptors for visual search. (2015).Google Scholar
- Julia Handl and Joshua Knowles. 2005. Multiobjective clustering around medoids. In 2005 IEEE Congress on Evolutionary Computation, Vol. 1. IEEE, 632--639. Google ScholarCross Ref
- Jie Lin, Ling-Yu Duan, Yaping Huang, Siwei Luo, Tiejun Huang, and Wen Gao. 2014. Rate-adaptive compact fisher codes for mobile visual search. IEEE Signal Processing Letters 21, 2 (2014), 195--198. Google ScholarCross Ref
- David G Lowe. 2004. Distinctive image features from scale-invariant keypoints. International journal of computer vision 60, 2 (2004), 91--110. Google ScholarDigital Library
- Kyriakos Mouratidis, Dimitris Papadias, and Spiros Papadimitriou. 2005. Medoid queries in large spatial databases. In International Symposium on Spatial and Temporal Databases. Springer, 55--72. Google ScholarDigital Library
- R. T. Ng and Jiawei Han. 2002. CLARANS: a method for clustering objects for spatial data mining. IEEE Transactions on Knowledge and Data Engineering 14, 5 (Sep 2002), 1003--1016. https://doi.org/10.1109/TKDE.2002.1033770 Google ScholarDigital Library
- Paul Over, George Awad, Martial Michel, Jonathan Fiscus, Wessel Kraaij, Alan F. Smeaton, Georges Quénot, and Roeland Ordelman. 2015. TRECVID 2015 -- An Overview of the Goals, Tasks, Data, Evaluation Mechanisms and Metrics. In Proceedings of TRECVID 2015. NIST, USA.Google Scholar
- Adriano Arantes Paterlini, Mario A Nascimento, and Caetano Traina Jr. 2011. Using pivots to speed-up k-medoids clustering. Journal of Information and Data Management 2, 2 (2011), 221.Google Scholar
- Georg Peters, Martin Lampart, and Richard Weber. 2008. Evolutionary Rough k-Medoid Clustering. Springer Berlin Heidelberg, Berlin, Heidelberg, 289--306. https://doi.org/10.1007/978-3-540-85064-9_13Google Scholar
Index Terms
- Efficient Approximate Medoids of Temporal Sequences
Recommendations
PTAS for p-Means q-Medoids r-Given Clustering Problem
Mathematical Optimization Theory and Operations ResearchAbstractWe consider the problem of clustering (partitioning) a given set of d-dimensional Euclidean vectors into a fixed number of clusters with different center types so that the total dispersion would be minimal, where by dispersion we mean the sum of ...
An improved K-medoids algorithm based on step increasing and optimizing medoids
The proposed clustering algorithm improves performance and preserves efficiency.We propose a candidate medoids subset to optimize the clustering medoids.We propose increasing the medoid methods in a step-wise fashion.Results report better performances ...
An efficient hybrid clustering algorithm for molecular sequences classification
ACM-SE 44: Proceedings of the 44th annual Southeast regional conferenceThe k-means clustering and hierarchical agglomerative clustering algorithms are two popular methods to partition data into groups. The k-means clustering algorithm heavily favors spherical clusters and does not deal with noise adequately. To overcome ...
Comments