# Parallelization of Machine Learning Algorithms Respectively on Single Machine and Spark

**Authors:** Jiajun Shen  
**arXiv:** [2206.07090](https://arxiv.org/abs/2206.07090)  
**Date:** 2022-05-08  
**Summary page:** https://www.summarizepaper.com/en/paper/2206.07090v2/

## Key points

- Paper focuses on parallelization of machine learning algorithms for analyzing large datasets
- Big data technologies have made extracting useful information from massive amounts of data a critical problem
- Applying machine learning algorithms to analyze such data can be time-consuming and inefficient on single machines
- Researchers conducted research on parallelizing classic machine learning algorithms on single machines and the Spark platform
- Aim was to compare runtime and efficiency of traditional machine learning algorithms with their parallelized counterparts on both platforms
- Results showed significant improvements in runtime and efficiency when using Spark's distributed computing capabilities compared to single machines
- Research highlights importance of parallelization in enhancing performance of machine learning algorithms with big data.

## Summary

This paper focuses on the parallelization of machine learning algorithms to address the challenges posed by analyzing large datasets using traditional single machines. With the rapid development of big data technologies, extracting useful information from massive amounts of data has become a critical problem. However, applying machine learning algorithms to analyze such data can be time-consuming and inefficient on single machines. To overcome these issues, the researchers conducted research on parallelizing several classic machine learning algorithms on both single machines and the Spark platform, which is a popular big data processing framework. The aim was to compare the runtime and efficiency of traditional machine learning algorithms with their parallelized counterparts on both platforms. The results of this study demonstrated significant improvements in the runtime and efficiency when leveraging the capabilities of distributed computing offered by Spark compared to running them on a single machine. Overall, this research highlights the importance of parallelization in enhancing the performance of machine learning algorithms when dealing with big data and contributes to advancing our understanding of how to extract valuable insights from massive datasets efficiently and effectively.