PROJECT TITLE :

ESA-Stream: Efficient Self-Adaptive Online Data Stream Clustering

ABSTRACT:

A wide variety of big data applications generate an enormous amount of streaming data that is high-dimensional, real-time, and constantly changing. The ability to cluster this kind of data streams in a way that is both effective and efficient is essential for these applications. Despite the fact that there are well-known data stream clustering algorithms that are founded on the widely used online-offline framework, these algorithms are still confronted with a number of significant obstacles. The following important questions have not yet been answered to everyone's satisfaction: How can dimensionality reduction be performed in an online environment that is dynamic in a way that is both effective and efficient? How can we enable the clustering algorithm to carry out processing in real time and in its entirety online? How can algorithm parameters be made to learn in a self-supervised or self-adaptive manner so that they can better deal with high-speed streams of evolving data? In this paper, we focus on addressing these challenges by proposing a fully online data stream clustering algorithm that we call ESA-Stream. This algorithm has the ability to learn parameters online dynamically in a self-adaptive manner, speed up dimensionality reduction, and cluster data streams effectively and efficiently in an environment that is both online and dynamic. Experiments conducted on a wide variety of synthetic and real-world data streams demonstrate that ESA-Stream performs significantly better than state-of-the-art baselines in terms of both its effectiveness and its efficiency.