International Journal of Science and Research (IJSR)

International Journal of Science and Research (IJSR)
Call for Papers | Fully Refereed | Open Access | Double Blind Peer Reviewed

ISSN: 2319-7064

Downloads: 2 | Views: 41 | Weekly Hits: ⮙2 | Monthly Hits: ⮙2

Informative Article | Computer Science and Information Technology | India | Volume 10 Issue 3, March 2021

A Review on Continuous Integration and Continuous Deployment (CI/CD) for Machine Learning

Ankur Mahida [9]

Abstract: Agile Continuous Integration and Continuous Deployment (CI/CD) is a collection of practices that help capitalize on automation by facilitating automatic software development, testing, and deployment processes. From its initial days in virtual desktop infrastructure software engineering, CI/CD has evolved and is widely utilized for its numerous advantages of streamlining workflows, enhanced collaboration, and quality of the finished software. Today, the ML field has been gaining widespread attention in CI/CD practices to tackle the challenges inherent in iterative procedures of model designing, the complexity of the data preparation, and the necessity of continuous monitoring and retraining. This overview gives a detailed analysis of the application of the CI/CD pattern in the ML area, considering how the practices could be replicated to improve the entire ML process: from data pre Via auto - provisioning different stages of the ML cycle through CI/CD pipelines, organizations can enjoy a high degree of consistent - ness and reproducibility across various datasets, environments, and teams, simultaneously want shorter development cycles, increased collaboration, and better model quality and reliability. The review stresses the possible perks of adopting CI/CD practices in the field of ML, including shorter time - to - market for ML solutions, better collaboration and interplay between data scientists, engineers, and domain experts, higher quality and reliability of ML models in production environments, and improved scalability and replicability of ML operations. Furthermore, the document addresses the obstacles that CI/CD encounters in ML, ranging from data pipeline automation to versioning, continuous integration, and tests, automated model training and evaluation, deployment and monitoring, collaboration and documentation, as well as the inclusion of security measures and governance into CI/CD pipelines.

Keywords: Continuous Integration, Continuous Deployment, Machine Learning, Automation, DevOps, CI/CD Pipeline, Model Deployment, Model Monitoring

Edition: Volume 10 Issue 3, March 2021,

Pages: 1967 - 1970

How to Download this Article?

Type Your Valid Email Address below to Receive the Article PDF Link

Verification Code will appear in 2 Seconds ... Wait