Affinity Propagation Feature Clustering with Application to Vehicle Detection and Tracking in Road Traffic Surveillance

Yang, Jun; Wang, Yang; Sowmya, Arcot; Zhang, Bang; Xu, Jie; Li, Zhidong

doi:10.1109/avss.2010.40

Cited by 5 publications

(1 citation statement)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Wang et al [8] applied meshing strategy to partition huge datasets, using AP and DP algorithm. He introduced structure similarity and compounded the partitioning centre distance.…”

Section: Applications Of A-pmentioning

confidence: 99%

Efficient Clustering of Water Distribution Network Using Affinity Propagation

Somisetti¹,

Palla²

2020

ISI

View full text Add to dashboard Cite

Clustering on graph data is one of the popular clustering techniques in the field of Graph mining. An accurate and fast clustering algorithm which attains high quality clusters in comparison with traditional data-mining or graph mining clustering-algorithms is Affinitypropagation (A-P). A large part of the real-world data falls under complex networks enhancing its applicability to wide sectors of research in computer science. In this paper, we consider a real-world complex infrastructure network, Water Distribution Network (WDN), consisting of node properties namely elevation, base demand, head, pressure and demand and the edge properties flow, velocity, roughness, length, diameter, friction factor and unit head loss. A-P algorithm initially develops a similarity-matrix by using data points of the real-world dataset. Secondly, responsibility and availability matrices are developed iteratively until constant responsibility and availability values are obtained. This is performed by using vectorized implementation. Finally, criterion matrix is developed as the sum of responsibility and availability to generate cluster centers. The WDN is sparse and complex; A-P gives strong clusters with potential cluster centers with time complexity of O(N 2). In addition, data points of a cluster formed are at same geographical location. An experimental result shows the novelty of A-P on WDN.

show abstract

“…Wang et al [8] applied meshing strategy to partition huge datasets, using AP and DP algorithm. He introduced structure similarity and compounded the partitioning centre distance.…”

Section: Applications Of A-pmentioning

confidence: 99%