Adaptive Hierarchical Density-Based Spatial Clustering Algorithm for Streaming Applications

Clustering algorithms are commonly used in the mining of static data. Some examples include data for relationships between variables and segmentation into components. The use a clustering algorithm real-time is much less common. This due to variety factors, including algorithm’s high computation cost. In other words, may be impractical or near-real-time implementation. Furthermore, necessitate tuning hyperparameters order fit dataset. this paper, we approach moving points using our proposed Adaptive Hierarchical Density-Based Spatial Applications with Noise (HDBSCAN) algorithm, which an implementation adaptive building minimum spanning tree. We switch Boruvka Prim as means build tree, one most expensive components HDBSCAN. HDBSCAN yields improvement execution time by 5.31% without depreciating accuracy algorithm. motivation research stems from desire cluster on video. Cameras monitor crowds improve public safety. can identify potential risks overcrowding movements groups people understanding flow crowds. Surveillance equipment combined deep learning assist addressing issue detecting objects, these items real generate information about clusters.