Related Experiment Video
Updated: Jul 9, 2025
09:17
Digital Spatial Profiling for Characterization of the Microenvironment in Adult-Type Diffusely Infiltrating Glioma
Published on: September 13, 2022
2.3K
A fast parallelized DBSCAN algorithm based on OpenMp for detection of criminals on streaming services
Lesia Mochurad1, Andrii Sydor1, Oleh Ratinskiy1
1Department of Artificial Intelligence, Lviv Polytechnic National University, Lviv, Ukraine.
Frontiers in Big Data
|November 29, 2023
Summary
This study introduces a parallel DBSCAN algorithm to efficiently analyze streaming service data, significantly speeding up clustering without sacrificing accuracy. The method shows high scalability on multicore systems, offering valuable applications in marketing and fraud detection.
Area of Science:
- Computer Science
- Data Science
- Algorithm Analysis
Background:
- Streaming services like Twitch are increasingly popular, generating vast amounts of user data.
- Analyzing this data efficiently is crucial for understanding user behavior and identifying patterns.
Purpose of the Study:
- To develop and evaluate a parallel DBSCAN algorithm for faster data clustering.
- To improve the efficiency of analyzing medium-sized datasets from streaming platforms.
Main Methods:
- Implementation of a parallel DBSCAN algorithm utilizing OpenMP for parallel computing.
- Avoidance of redundant neighbor search calculations for improved performance.
- Validation of clustering quality using the silhouette value.
Main Results:
- The parallel DBSCAN algorithm achieved significant speed-up on medium-sized datasets.
- Acceleration rate correlated with the number of available CPU cores.
- High efficiency, approaching one, was observed.
Conclusions:
- The proposed parallel DBSCAN algorithm offers a scalable and efficient solution for clustering streaming data.
- The algorithm demonstrates potential applications in diverse fields including marketing, cybersecurity, and fraud detection.
