What we’re about
Data Science Portugal (DSPT) is an informal community of AI and machine learning enthusiasts dedicated to sharing knowledge in data science, big data, and related fields. Since 2016, DSPT has fostered collaboration across domains, organizing 70 meetups in five cities and growing to nearly 3.5K members by 2025.
Upcoming events (1)
See all- Meetup #120 - Leveraging Machine Learning for Industrial Process DigitalisationIT - Institute of Telecommunications of Aveiro, Aveiro
Welcome to another DSPT meetup at Aveiro! 🙌
=== SCHEDULE ===
• 18:30 - 18:45: Get together
• 18:45 - 18:55: Welcome message
• 19:00 - 19:40: Talk + Q&A: "Leveraging Machine Learning for Industrial Process Digitalisation – Exploring a Real-World Industrial Use Case" by José Cação
• 19:40 - 20:30: Networking/Coffee Break
• 20:30: Dinner is optional, but might be an excellent networking opportunity! 🍕register here.
A special thanks to our venue IT 👏# See you there!
Abstract: Currently, the manufacturing process of digitalisation is a major focus within the industrial paradigm. Particularly, Machine Learning (ML) is seen as a crucial driver for industrial process automation and optimisation, providing a promising alternative to traditional, less flexible rule-based industrial data processing and analysis methods. This presentation explores a real-world industrial application of integrating ML within industrial fault detection and diagnosis tasks, developed in collaboration with Bosch Termotecnologia S.A. The use case involves an end-of-line testing procedure conducted for all produced boilers, one major bottleneck in this equipment’s production lines. More specifically, an innovative, data-driven methodology to detect and classify potential faults within the spark ignition process, exclusively analysing electrical signal data. The proposed methodology entails a complex data preparation and pre-processing procedure, adapted to the current production infrastructure, combined with a two-step classification approach which (1) separates normal and abnormal signals (binary classification) and (2) classifies the potential type of fault (multilabel classification). Results showcase improvements in up to 80% in detection accuracy over current knowledge-based approaches implemented by the partner, with the two-step approach outperforming traditional single-stage multilabel classification by up to 10%.
Bio:
José Cação holds a master’s degree in Mechanical Engineering from the University of Aveiro (2023, final grade 18/20), having finished his thesis in 2023 with the highest 20/20 grade. The work focused on optimising industrial equipment energy consumption through Deep Learning (DL) approaches. Currently pursuing a Ph.D. in Mechanical Engineering at the University of Aveiro, José Cação focuses his research on industrial data analysis, process digitalisation, and optimisation through Machine Learning (ML) and DL approaches, and the use of Explainable AI to improve model interpretability, trust, and auditability in AI-based decision-making systems.
Throughout his academic career, he has been actively involved in applied research projects with Bosch Termotecnologia S.A., mainly through the Augmanity and ILLIANCE projects – the latter of which he currently contributes to as a researcher. José Cação has authored and co-authored multiple publications, including articles in peer-reviewed journals like Sensors and presentations at leading international conferences, earning distinctions such as the “Best Paper Award” – 5th International Conference on Industry 4.0 and Smart Manufacturing (ISM 2023) and “Best Oral Presentation” at TechMA 2024, at the University of Aveiro.
