EPIA 2026

Track Description

The AITD track focuses on the scientific and technical challenges of building artificial intelligence systems that operate on time-dependent data. This includes time series, data streams, and other sequential or temporal structures where order, dynamics, and evolution over time are central to modelling and decision-making. The scope spans both supervised and unsupervised settings (forecasting, regression, activity recognition, classification, clustering, anomaly detection), as well as the design of learning algorithms that can cope with non-stationarity, concept drift, sparse or irregular sampling, and missing values. Emphasis is also placed on methods that are interpretable, robust, privacy-aware, and that provide reliable uncertainty quantification. The track also welcomes work on spatio-temporal modelling, temporal pattern and motif discovery, and applications across domains such as finance, healthcare, energy, and industry.

Topics of Interest

• Time series analysis, forecasting, extrinsic regression, classification, clustering
• Online learning and data stream mining
• Detection and adaptation to Concept Drift
• Spatio-temporal data analysis and modeling
• Temporal pattern discovery and motif mining
• Time series with sparse or irregular sampling, or missing values
• Anomaly and outlier detection in temporal data
• Explainable AI (XAI) for temporal models
• Stress testing and robustness in time-dependent data
• Privacy-preserving approaches for learning from temporal data
• Uncertainty quantification in time-dependent data
• Learning from sequential data
• Applications of time series and data streams, such as in Finance, Healthcare, Energy, or Industry 4.0

Track Chairs

• Vitor Cerqueira - University of Coimbra
• Paulo Salgado G. de Mattos Neto - Universidade Federal de Pernambuco
• Moisés Santos - Universidade do Porto