Machine Learning Lifecycle Explorer: Leverage the Lifecycle Metadata (onto) Logically

doi:10.1007/978-3-032-03550-9_25

Machine Learning Lifecycle Explorer: Leverage the Lifecycle Metadata (onto) Logically

2026-01-01

1

PRE

AI

Boonserm Kulvatunyou *

DOI：10.1007/978-3-032-03550-9_25

Original

Original request for help

Save

Abstract

En 中文

As machine learning (ML) continues to grow in complexity and importance, effective lifecycle management becomes a critical factor for successful model development and deployment. A key challenge is the management of metadata across the entire ML lifecycle, ensuring reproducibility, performance tracking, and optimal model evaluation. In our previous work, we introduced the Machine Learning Lifecycle Ontology (MLLO), a standardized, extensible, and ML framework-neutral solution for representing metadata throughout the ML process. This paper builds on MLLO by introducing an associated software tool called Machine Learning Lifecycle Explorer (MLLE, pronounced 'milly'). MLLE has been designed to leverage MLLO through a graphical user interface to interact with ML metadata. It consists of two main components: the MLLO Analyzer, which facilitates dataset discovery, model evaluation, and performance tracking, and the MLLO Editor, which enables users to visualize, edit, and create ML pipelines. In this paper, we demonstrate how MLLE utilizes MLLO to facilitate model reuse based on the ML application requirements. The paper concludes by discussing future developments of MLLO and MLLE to support other use cases.

Keywords:

MLOps

Ontology

Machine Learning Lifecycle Management

Machine Learning Lifecycle Explorer: Leverage the Lifecycle Metadata (onto) Logically

Abstract

Journal

Organization

Cited Papers