Ataya: The HUMA Interdisciplinary Seminar Series
13:00-14:00 SAST 

Speaker: Messi Nguelé Thomas (University of Yaoundé I & HUMA Visiting Research Fellow)

Topic: Complex systems are systems composed of many components in interactions and usually generates a large amount. Many complex systems analysis applications are using machine learning algorithms [1,2]. Once trained, the models built by machine learning algorithms can be used to solve many problems related to regression, classification, or clustering. For better inference, models should be trained on large amounts of data; but the greater the amount of data for training, the higher the execution time of the algorithms. The advent of parallel architectures (multi/many-core, GPU) makes it possible to reduce this execution time. However, using them efficiently require to re-think traditional sequential  machine learning algorithms into parallel ones. As multi-core are more and more common, we want to fully exploit them to make complex systems applications to be faster. In this project, we want to parallelize machine learning algorithms that are intrinsically (by definition) sequential. Our goal is to take advantage of parallel architectures that are available even to common user, in order to produce the most efficient machine learning models with the shortest training time. As results, we expect to have: - Parallel model for machine learning algorithms; - A DSLs for machine learning algorithms parallelization; -Algorithms or heuristics that make complex system applications faster.  Currently, we have preliminary results obtained with multi-core in the context of AI4D Africa project.  This project is its extension. It will be done by integrating ethical considerations and seeing if it alters results in terms of execution time or performance metric. 


About the speaker: Dr Messi Nguelé Thomas defended his PhD in Computer Science in 2018 in cotutelle between University Grenoble Alpes and University of Yaoundé I. He is now a Senior Lecturer in Computer Science at the Faculty of Sciences of the University of Yaoundé I. He is Head of the Department of Computer Engineering at ESTLC of the University of Ebolowa. He is the Head of the Computer Science and Applications Laboratory at Ambam. He mainly teaches programming courses (C, Python) and distributed learning courses on parallel architectures. His research work mainly focuses on the parallelization of machine learning algorithms.

He is a recipient of the Africom Award Mobily for Staff (July 2022) and a recipient of the ACTS AI4D Africa funding (August 2022 – August 2024). He was a visiting researcher at University of Grenoble (September 2023 and February 2025). He is affiliated with the Idasco research team (Data Sciences and Complex Systems), in its HiPerDas (High Performance Data Science) branch. He is the President of the ARITeD research association. He is the Education Officer of the Cameroon Computer Science Society and Treasurer of the Cameroon Artificial Intelligence Society. (Personal Website, Google Scholar)
 

References:

[1]- Silva, T. C., & Zhao, L. (2016). Machine learning in complex networks (Vol. 1). Springer International Publishing.

[2]- Qi, D., & Majda, A. J. (2020). Using machine learning to predict extreme events in complex systems. Proceedings of the National Academy of Sciences, 117(1), 52-59.

[3]- Aboul, E. H. (2015). Big Data in Complex Systems-Challenges and Opportunities. Springer International Publishing.

[4]-Soulaimane Guedria. A scalable and component-based deep learning parallelism platform : an application to convolutional neural networks for medical imaging segmentation. Logic in Computer Science [cs.LO]. Université Grenoble Alpes [2020-..]

[5]- https://www.who.int/fr/news-room/fact-sheets/detail/malaria

[6]- Chen, I. Y., Pierson, E., Rose, S., Joshi, S., Ferryman, K., & Ghassemi, M. (2021). Ethical machine learning in healthcare. Annual review of biomedical data science, 4(1), 123-144.

 

How Ataya works: One presenter and their work – in exchange with the audience. Each Ataya session engages with selected work by the presenter (a text, artwork, performance, even food). The presenter introduces their work and grounds the subsequent discussion with the participants. For best engagement, we recommend participants to view the work (made available in advance on our website) before the session. More on the Ataya Series


For any questions: huma@uct.ac.za