EMPOWERING DECISION-MAKING AND AUTONOMY: INTEGRATING MACHINE LEARNING INTO MICROSERVICES ARCHITECTURES



Authors

  • Roshan Mahant1, Sumit Bhatnagar2

DOI:

https://doi.org/10.15282/jmes.17.1.2023.10.0759


Keywords:

Machine Learning Operations, Microservices, Machine learning, convolutional neural networks (CNN), k-nearest neighbor (KNN), and deep neural networks (DNN).


Abstract

The emergence of microservices architecture has revolutionized the development of distributed software systems, offering a plethora of advantages such as modularity, scalability, and flexibility. However, along with these benefits come challenges in deployment, coordination, and interaction among microservices. In this paper, we conduct a survey to explore how artificial intelligence (AI) techniques have been leveraged to tackle these challenges. Specifically, we propose encapsulating machine learning (ML) as data-oriented microservices, thereby simplifying the integration of ML capabilities into applications. To exemplify this, convolutional neural networks (CNN), k-nearest neighbor (KNN), and deep neural networks (DNN).By decoupling algorithm implementation from configuration, we conduct an in-depth analysis of their functionalities and depict them as Representational State Transfer (REST) ML service modules that are dynamically linkable. We employ a self-contained module following a service-oriented design for the implementation of neural network training. To evaluate our solution, we compare its performance with direct programming using the Tensor Flow library, showcasing that our methodology significantly streamlines the deployment of machine learning-based data analytics by enabling the reuse and sharing of programs and configurations.



Published

2024-06-22

How to Cite