Carleton University
Conceptualizing, Applying and Evaluating SecMLOps: A Paradigm for Embedding Security into the ML Lifecycle
Abstract
dc:description.abstractAs machine learning (ML) systems become increasingly integrated into critical infrastructure and decision-making processes, ensuring their security and reliability is of paramount importance. However, traditional DevOps practices often fall short in addressing the unique security challenges posed by ML systems. This thesis introduces SecMLOps, a novel paradigm that explicitly integrates security considerations throughout the MLOps lifecycle to develop more secure, reliable, and trustworthy ML systems. The research begins with a comprehensive analysis of the DevOps landscape using a multivocal literature review methodology, providing a structured understanding of the various specialized DevOps variants and their relationships. Building upon these insights, the SecMLOps paradigm is conceptualized, examining its key aspects of people, technology, processes, governance, and compliance. To operationalize SecMLOps, a comprehensive framework is developed, integrating critical security activities into the MLOps workflow and assigning specific roles and responsibilities across the lifecycle stages. The effectiveness of the SecMLOps framework is evaluated through a real-world use case on a state-of-the-art pedestrian detection system, demonstrating its robustness against prevalent security threats such as data poisoning and adversarial examples. This thesis critically assesses the practical impact and challenges of adopting SecMLOps, considering factors such as organizational transformation, process integration, and resource implications. The applicability and generalizability of the framework across different domains are discussed, emphasizing its inherent adaptability while acknowledging the need for domain-specific considerations. By providing a structured approach to integrating security into MLOps and validating its effectiveness, this thesis advances the state of the art in secure ML development and deployment. The SecMLOps paradigm empowers organizations to proactively address security challenges, ensuring the integrity and trustworthiness of their ML systems in the face of evolving threats. As ML continues to expand into critical domains, the contributions of this thesis serve as a foundation for future research and practice in developing secure and reliable ML systems.
Degree
thesis:*- Name thesis:degree_name
- Doctor of Philosophy (Ph.D.)
- Level thesis:degree_level
- Doctoral
- Discipline thesis:degree_discipline
- Engineering, Electrical and Computer
- Grantor dc:publisher
- Carleton University
- Year dc:date.issued
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Zhang, Xinrui
Subjects
dc:subject × 2Rights
dc:rights- Statement dc:rights
-
- Copyright © 2025 the author(s). Theses may be used for non-commercial research, educational, or related academic purposes only. Such uses include personal study, distribution to students, research and scholarship. Theses may only be shared by linking to the Carleton University Institutional Repository and no part may be copied without proper attribution to the author; no part may be used for commercial purposes directly or indirectly via a for-profit platform; no adaptation or derivative works are permitted without consent from the copyright owner.
- Language dc:language.iso
- en