Abstract
The increased interest and importance of explaining neural networks' predictions, especially in the medical community, associated with the known unreliability of saliency maps, the most common explainability method, has sparked research into other types of explanations. Natural Language Explanations (NLEs) emerge as an alternative, with the advantage of being inherently understandable by humans and the standard way that radiologists explain their diagnoses. We extend upon previous work on NLE generation for multi-label chest X-ray diagnosis by replacing the traditional decoder-only NLE generator with an encoder-decoder architecture. This constitutes a first step towards Reinforcement Learning-free adversarial generation of NLEs when no (or few) ground-truth NLEs are available for training, since the generation is done in the continuous encoder latent space, instead of in the discrete decoder output space. However, in the current scenario, large amounts of annotated examples are still required, which are especially costly to obtain in the medical domain, given that they need to be provided by clinicians. Thus, we explore how the recent developments in Parameter-Efficient Fine-Tuning (PEFT) can be leveraged for this usecase. We compare different PEFT methods and find that integrating the visual information into the NLE generator layers instead of only at the input achieves the best results, even outperforming the fully fine-tuned encoder-decoder-based model, while only training 12% of the model parameters. Additionally, we empirically demonstrate the viability of supervising the NLE generation process on the encoder latent space, thus laying the foundation for RL-free adversarial training in low ground-truth NLE availability regimes. The code is publicly available at https://github.com/icrto/peft-nles.

Abstract

Abstract
The education sector faces unprecedented challenges, from rapidly evolving technologies to diverse learner needs, placing immense pressure on educators to adapt and innovate. Traditional teaching methods need help to keep pace with the demands of modern education, leading to gaps in personalized learning and student engagement. Ethical concerns surrounding AI integration in education remain a significant hurdle, requiring careful navigation and responsible implementation. Integrating Artificial Intelligence in Education: Enhancing Teaching Practices for Future Learning offers a comprehensive solution by exploring how AI can address these challenges and revolutionize education. Through a collection of insightful contributions, it provides practical strategies for integrating AI into teaching practices, empowering educators to personalize learning experiences and enhance student engagement. By examining AI ethics and responsible education, the book equips educators with the knowledge needed to navigate the ethical complexities of AI integration. This book is a practical guide for educators, researchers, policymakers, and practitioners who want to harness the potential of AI in education. It provides a roadmap for leveraging AI technologies to create adaptive learning environments, automate classroom tasks, and enhance instructional design. With a strong focus on practical insights and ethical considerations, this book is a valuable resource for anyone looking to navigate the intersection of AI and education. © 2025 by IGI Global. All rights reserved.

Abstract
This paper presents an approach to address the challenges of manual inspection using multirotor Unmanned Aerial Vehicles (UAV) to detect olive tree flies (Bactrocera oleae). The study employs computer vision techniques based on the You Only Look Once (YOLO) algorithm to detect insects trapped in yellow chromotropic traps. Therefore, this research evaluates the performance of the YOLOv7 algorithm in detecting and quantify olive tree flies using images obtained from two different digital cameras in a controlled environment at different distances and angles. The findings could potentially contribute to the automation of insect pest inspection by UAV-based robotic systems and highlight potential avenues for future advances in this field. In view of the experiments conducted indoors, it was found that the Arducam IMX477 camera acquires images with greater clarity compared to the TelloCam, making it possible to correctly highlight the set of Bactrocera oleae in different prediction models. The presented results in this research demonstrate that with the introduction of data augmentation and auto label techniques on the set of images of Bactrocera oleae, it was possible to arrive at a prediction model whose average detection was 256 Bactrocera oleae in relation to the corresponding ground truth value to 270 Bactrocera oleae.

Abstract
The speed and scale of technological change are raising concerns about the extent to which new technologies will radically transform workplaces. Competition for the best talent is being intensified, and talent management requires new approaches and innovative strategies for developing talent based on corporate culture and its unique properties. By implementing and adopting technology in Human Resources Management (HRM), organizations create a digital employee lifecycle that spans from the initial Hiring Process to encompassing areas such as Performance Management, Learning and Development until the Offboarding, shaping a Talent Management journey. Despite the implementation of technologies being a continuous practice observed in numerous organizations, there are still challenges. The HRM technological market has become massive, and concerns arise about adopting these technologies' costs, practicality, and purpose. Because of that, designing strategies for implementing technologies in HRM, specifically in talent management, is hard to overview. In this context, this document aims to present the necessity and significance in developing a framework that aggregates the implementation process of technologies in talent management supported by Design Science Research (DSR). The holistic perspective of the forthcoming framework consolidates insights into business challenges and their correlation with technology selection, technological capabilities, implementation procedures, as well as anticipated metrics and their impact.

Abstract
This paper proposes to extend the fast-decoupled state estimation formulation to bring its well-known efficiency and benefits to the processing of networks with embedded FACTS devices. The proposed method approaches shunt-, series-, and shunt -series -type devices. The controller parameters are included as new active or reactive state variables, while controlled quantity values are included in the metering scheme of the decoupled approach. From the electrical model adopted for each device, the extended formulation is presented, and a modified fast-decoupled method is devised, seeking to ensure accuracy and impart robustness to the iterative solution. Simulation results conducted throughout the IEEE 30 -bus test system with distinct types of FACTS devices are used to validate and evaluate the performance of the proposed decoupled approaches.