Despite the increasing availability of vast amounts of data, the challenge of acquiring labeled data persists. This issue is particularly serious in supervised learning scenarios, where labeled data are essential for model training. In addition, the rapid growth in data required by cutting-edge technologies such as deep learning makes the task of labeling large datasets impractical. Active learning methods offer a powerful solution by iteratively selecting the most informative unlabeled instances, thereby reducing the amount of labeled data required. However, active learning faces some limitations with imbalanced datasets, where majority class over-representation can bias sample selection. To address this, combining active learning with data augmentation techniques emerges as a promising strategy. Nonetheless, the best way to combine these techniques is not yet clear. Our research addresses this question by analyzing the effectiveness of combining both active learning and data augmentation techniques under different scenarios. Moreover, we focus on improving the generalization capabilities for minority classes, which tend to be overshadowed by the improvement seen in majority classes. For this purpose, we generate synthetic data using multiple data augmentation methods and evaluate the results considering two active learning strategies across three imbalanced datasets. Our study shows that data augmentation enhances prediction accuracy for minority classes, with approaches based on CTGANs obtaining improvements of nearly 50% in some cases. Moreover, we show that combining data augmentation techniques with active learning can reduce the amount of real data required.

Es prioritario para la Escuela de Ingeniería de Gipuzkoa (EIG) de la Universidad del País Vasco/Euskal Herriko Unibertsitatea (UPV/EHU) reducir la tasa de abandono de los estudiantes de primer curso en los grados de Ingeniería. Para ello, se puso en marcha un Proyecto de Innovación Educativa (PIE) durante el periodo 2015-2017. El PIE propuesto, busca una mejora significativa en las Tasas de Rendimiento (TR), Tasas de Éxito (TE) y Tasas de Abandono (TA) de las asignaturas de primer curso de los grados que se imparten en la EIG, a través de la mejora en la enseñanza y de la información y orientación de nuevo alumnado. El producto a elaborar consiste en un conjunto de propuestas innovadoras para los procesos de enseñanza/aprendizaje (E/A). Ésta comunicación se centra en la encuesta al alumnado de primer curso que se ha llevado a cabo en la EIG. Los resultados preliminares evidencian que los alumnos que llegan a la universidad tienen dificultades para adaptarse a las nuevas dinámicas de trabajo. Están habituados a la sobreinformación, por lo que se sienten perdidos cuando la propia búsqueda de información forma parte del proceso de aprendizaje. [It is critical for the Faculty of Engineering, Gipuzkoa (FEG) of The University of The Basque Country to reduce the Dropout Rate (DR) in the first course of Engineering Grades. To this end, an Educational Improvement Project (EIP) was launched for the period 2015-2017. The proposed EIP seeks a significant improvement in the Yield Rate (YR), Rate of Success (RS) and DR of the first year courses of the degrees that are taught in the FEG, through the improvement in teaching and information and orientation of new students. The product to be elaborated consists of a set of innovative proposals for the processes of Teaching / Learning (T/L). This communication focuses on the questionnaire that has been passed to first-year students in the FEG. The preliminary results show that students have difficulties to adapt to new working dynamics when arriving to university. They are used to the information overload, which makes them get lost when the information search itself is part of the learning process.] Echegaray, Goretti; Barroso, Nora; Laskurain, Iker; Zuza, Kristina; Barragués, José Ignacio