Automation of machine learning pipelines for anomaly detection challenges - Master’s Dissertation in Informatics Engineering

Aug 4, 2026

Upload by Asher

Machine Learning (ML) and Data Science are used to address real-world problems, while rapid technological change introduces challenges across the ML development, deployment cycle, and data cleansing that can reduce accuracy and viability. DevOps practices work for software systems at scale, motivating MLOps, a culture tailored for ML. This thesis studies frameworks and methods to build an automated ML pipeline for ingesting and manipulating large-scale data. A sensor-based system simulates a vehicle interior to collect inputs, while a visual interface enables control and monitoring of the system and its data.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages75

Reading Time3 h 9 min

AccessFree to download

Acknowledgements

Abstract

Keywords

Statement of Integrity

Universidade do Minho Escola de Engenharia Ricardo Rodrigues Martins Automation of machine learning pipelines for anomaly detection challenges December, 2023 Universidade do Minho Escola de Engenharia Ricardo Rodrigues Martins Automation of machine learning pipelines for anomaly detection challenges Master’s Dissertation in Informatics Engineering Work supervised by João Miguel Lobo Fernandes André Leite Ferreira December, 2023 COPYRIGHT AND TERMS OF USE OF THIS WORK BY A THIRD PARTY This is academic work that can be used by third parties as long as internationally accepted rules and good practices regarding copyright and related rights are respected. Accordingly, this work may be used under the license provided below. If the user needs permission to make use of the work under conditions not provided for in the indicated licensing, they should contact the author through the RepositóriUM of Universidade do Minho. License granted to the users of this work Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International CC BY-NC-SA 4.0 [https://creativecommons.org/licenses/by-nc-sa/4.0/deed.en](https://creativecommons.org/licenses/by-nc-sa/4.0/deed.en) STATEMENT OF INTEGRITY I hereby declare having conducted this academic work with integrity. I confirm that I have not used plagiarism or any form of undue use of information or falsification of results along the process leading to its elaboration. I further declare that I have fully acknowledged the Code of Ethical Conduct of the Universidade do Minho. ~~ ~~ , ~~ ~~ (Location) (Date) (Ricardo Rodrigues Martins) To Jacinto Rodrigues, Maria Gonçalves, José Fernando Costa and António Martins. Acknowledgements I would like to begin my acknowledgements by thanking the University of Minho and Bosch Car Multimedia. I have been studying with the University of Minho since 2015, before switching paths to the Integrated Master’s Degree in Informatics Engineering. This course provided me the foundations on software engineering I needed for my professional life and this thesis. Bosch Car Multimedia taking me in for a curricular internship program in 2020, where I was provided with everything I needed to develop both my work and myself as a software engineer, from highly specialized hardware and software, to valuable expertise from its collaborators. From the University of Minho, I would like to thank mainly Professor João Miguel Lobo Fernandes, for taking me under his supervision. His time and support have been essential for developing the Model Training Engine and creating this thesis. From Bosch Car Multimedia, I would like to thank mainly André Leite Ferreira for accepting me for the curricular internship program where the work was created for this thesis, as well as the support for writing it. This opened the door to the career path I have taken, and for which I am grateful. Also from Bosch Car Multimedia, I would like to thank Eliseu Vieira for his important role in my acceptance for the internship; João Fernandes, Bruno Faria and Afonso Costa for taking their time to help me, providing valuable insights not only for this thesis, but also on how to become a better professional; the Olimpo Team and other 2020 interns, João Sá, Luís Coelho, Bruno Nascimento, Rita Pereira, Gabriel Santos, Carolina Gonçalves and Afonso Sousa, for helping and joining me in the adventure of being an intern. On a personal side, thank you to my parents, who gave me everything they could so I had all the means I needed to choose and realise my own future. Every personal success I’ve had or may have, could only be achieved through the many, many sacrifices they have made throughout their life of honesty, love, and hard work. Thank you to my sister, Daniela, who helped me keep a level head through my whole academic journey, and always pushed me to do better. Thank you to my girlfriend, Ticiana, for being endlessly patient, supportive and caring th

FAQs

What problem does the thesis focus on in ML systems?

The thesis targets challenges across the ML development and deployment cycle, including data cleansing, which can impact model accuracy and the viability of ML software systems.

What is MLOps and why is it central to the work?

MLOps is a development culture derived from DevOps principles, aimed at unifying the ML system development cycle while enabling automation, monitoring, and continuous integration/delivery.

How does the proposed solution obtain data and support system control?

A sensor system simulates the interior of a vehicle to gather data for the pipeline, and a visual interface is built to control and manage the overall system and its data.

Aug 4, 2026

Upload by Asher