In collaboration with NCC Montenegro, NCC Bosnia and Herzegovina, and NCC Türkiye, HPC Serbia successfully organized the 1.5-day online training event AI in Action for SMEs on 2–3 March 2026 as part of the EuroCC4SEE initiative.
The event brought together regional expertise to support SMEs, start-ups, and researchers in adopting practical Artificial Intelligence solutions. Participants explored real-world AI applications, including Time Series Forecasting, Explainable AI, and Data Anonymization, through a combination of theoretical sessions and hands-on HPC demonstrations.
The program addressed key aspects of AI implementation, trustworthiness, and data protection, while highlighting how High-Performance Computing (HPC) enables faster experimentation, scalable model training, and advanced AI use cases relevant to business and research environments. The interactive format encouraged cross-border exchange and practical discussion on how SMEs can move from AI awareness to concrete implementation.
The training confirmed strong regional interest in HPC-powered AI solutions and demonstrated the value of coordinated action within the EuroCC4SEE network.
The agenda is available at https://indico.truba.gov.tr/event/250/.
Time-Series Forecasting with AI & Machine Learning
In this session, Milutin Pavicevic from the University of Donja Gorica introduced participants to the fundamentals of time-series data and the main challenges involved in analyzing and predicting values that evolve over time. The presentation explained how AI and machine learning methods can be applied to identify patterns, trends, and seasonality in sequential data, enabling more accurate forecasting across a wide range of real-world applications. By combining theoretical concepts with practical insights, the session highlighted the growing importance of data-driven forecasting techniques in research, industry, and decision-making processes.
Demo Session on Time Series
During the demo session, Gamze Erdoğan from the ITU Informatics Institute provided a hands-on introduction to time-series analysis, guiding participants through practical techniques for working with data that evolves over time. The session demonstrated how to explore, preprocess, and visualize time-dependent datasets, highlighting key concepts such as trend detection, seasonality, and anomaly identification. Through live examples and interactive workflows, participants gained insight into common tools and methods used for forecasting and modeling, as well as best practices for handling real-world data challenges.
Explainability in AI: Theoretical, Computational, and Security Perspectives
In this session, Gamze Erdoğan, on behalf of Dr. Enver Özdemir from the ITU Informatics Institute, presented how and why modern AI systems make decisions, focusing on methods that make these processes transparent and interpretable. The session presented key theoretical foundations of explainable AI (XAI), alongside computational techniques used to extract explanations from complex models such as deep neural networks.
Explainable AI Demo on HPC
This session, delivered by Dr. Valentina Janev, Miloš Nenadović, and Dejan Paunović from the Mihajlo Pupin Institute on behalf of the IntelliLung Consortium, presents the implementation of an AI-driven decision support system (AI-DSS) within the IntelliLung Project. The session highlights how "ethics-by-design" and explainable AI methodologies are applied to develop a trustworthy AI-DSS for use in intensive care units, supporting mechanical ventilation decisions.
Applied Data Anonymization Techniques
In the session Applied Data Anonymization Techniques, Lemana Spahić from Verlab Institute provided an overview of key data protection challenges in modern data-driven environments, emphasizing the risks associated with handling sensitive and personal data. The session introduced core concepts of anonymization, including practical techniques for reducing re-identification risks while preserving data utility. Through real-world examples drawn from her professional experience, Dr. Spahić illustrated common pitfalls, best practices, and the balance between privacy protection and analytical value, offering participants actionable insights into implementing effective anonymization strategies in research and industry contexts.
Text Anonymization: a Practical Demo
In this session, Dr. Bojana Bašaragin from the Institute for Artificial Intelligence Research and Development of Serbia delivered a hands-on demonstration of techniques for anonymizing textual data. The session focused on identifying and removing or transforming sensitive information within unstructured text, such as personal names, locations, and identifiers, while maintaining the usability of the data for analysis.
