Complex survey data are derived by sampling the population of interest for the study based on a complex sampling design, implementing techniques such as stratification or clustering. In order to compensate for the unequal selection probabilities derived from the implementation of complex sampling design techniques, a sampling weight is assigned to each individual, which indicates the number of units this represents in the population. This kind of data is usually used for the same purposes as traditional simple random samples, including the development of prediction models, among others. However, due to the special characteristics of the data collection process, traditional statistical techniques are usually not valid in this context, and complex survey data need to be treated in particular ways. The goal of this talk is to present new methodological proposals that account for complex sampling designs in the development process of prediction models. In particular, a new design-based variable selection technique that accounts for the complex sampling structure has been proposed based on LASSO regression models. In addition, new design-based estimators have been proposed for estimating the discrimination ability of logistic regression models for dichotomous outcomes fitted to complex survey data. The validity of both proposals has been evaluated through simulation studies. Finally, these proposals have been implemented into the svyVarSel and svyROC R packages, available on CRAN.

Ensuring fairness in AI is a complex challenge that requires bridging ethical principles with practical implementation. This talk explores key concepts in algorithmic fairness, from defining bias and fairness in predictive machine learning to assessing stigmas in large language models (LLMs). Using real-world examples, we illustrate the sociotechnical nature of fairness and the need for case-specific approaches.

As LLMs reshape the landscape, we examine new challenges, including how bias manifests differently across languages, methods for measuring these disparities, and strategies for mitigation. Finally, we advocate for an ethics-by-design mindset, emphasizing continuous monitoring, multidisciplinary collaboration, and proactive governance to ensure fairness remains at the core of AI development.

Load forecasting is essential for the efficient, reliable, and cost-effective management of power systems. Load forecasting performance can be improved by learning the similarities among multiple entities (e.g., regions, buildings). Techniques based on multi-task learning obtain predictions by leveraging consumption patterns from the historical load demand of multiple entities and their relationships. However, existing techniques cannot effectively assess inherent uncertainties in load demand or cannot account for dynamic changes in consumption patterns. This talk proposes a multi-task learning technique for online and probabilistic load forecasting. This technique provides accurate probabilistic predictions for loads of multiple entities by leveraging their dynamic similarities. The method's performance is evaluated using datasets that register the load demand of multiple entities and contain diverse and dynamic consumption patterns. The experimental results show that the proposed method can significantly enhance the effectiveness of current multi-task learning approaches across a wide variety of load consumption scenarios.

 In this talk we will look at the interplay between Shannon Classical Information Theory and Quantum Information.

The concepts of Classical  Compression and Channel Coding will be related with their Quantum counterparts.

El uso de herramientas de inteligencia artificial (IA) en todos los campos de la actividad humana plantea cuestiones cruciales sobre la responsabilidad ética profesional. ¿En qué medida las técnicas de IA representan un desafío ético adicional comparado con herramientas tradicionales de cómputo? La charla se centra en la problemática de la explicabilidad y la comprensión del trabajo con IA mediante procesos de «caja negra»: ¿cómo un ingeniero, un médico, un abogado, puede asumir responsabilidad ética si no puede explicar racionalmente su trabajo, tanto a sí mismo como a los demás? Se demuestra así necesaria una nueva comprensión de la ética en el diseño y uso de las tecnologías de inteligencia artificial.