<div class="csl-bib-body">
<div class="csl-entry">Iglesias Vazquez, F. (2025, October). <i>Machine Learning for Complex, Time-Dependent Applications: Representative Cases</i> [Presentation]. Seminario de Excelencia (IRNASA CSIC), Salamanca, Spain.</div>
</div>
-
dc.identifier.uri
http://hdl.handle.net/20.500.12708/220439
-
dc.description.abstract
We explore the use of Machine Learning in complex, time-dependent applications across three research domains: cybersecurity, environmental science, and medicine. Through representative case studies, it highlights how methodological rigor, data understanding and visualization, and tailored analytical design are essential to deriving meaningful results.
In cybersecurity, we address the identification and online detection of attacks in network traffic, discussing challenges in labeling, non-stationarity, and data imbalance. In environmental science, the focus shifts to the development of Digital Twins for regional natural resources and, specifically, for the prediction of groundwater levels, illustrating the integration of heterogeneous data sources and the modeling of dynamic environmental systems. In medicine, the use of persistent homology provides a novel perspective on microbial community diversity, interaction and dynamics, linking topological patterns to biological relevance.
Across all cases, the discussion follows a structured path: problem definition and data characterization, exploration and visualization, and the design of algorithms and pipelines adapted to the nature of the data and research goals. The presentation also touches on the design of large-scale data integration platforms supporting interdisciplinary research. Finally, it critically examines the hype surrounding Deep Learning and LLMs, emphasizing when these approaches are justified, and when more classical or interpretable methods of machine learning remain the most effective choice.
en
dc.language.iso
en
-
dc.subject
data analysis
en
dc.subject
artificial intelligence
en
dc.subject
time-dependent applications
en
dc.title
Machine Learning for Complex, Time-Dependent Applications: Representative Cases
