Abstract
Leveraging the capabilities of Knowledge Distillation (KD) strategies, we devise a strategy to fight the recent retraction of face recognition datasets. Given a pretrained Teacher model trained on a real dataset, we show that carefully utilising synthetic datasets, or a mix between real and synthetic datasets to distil knowledge from this teacher to smaller students can yield surprising results. In this sense, we trained 33 different models with and without KD, on different datasets, with different architectures and losses. And our findings are consistent, using KD leads to performance gains across all ethnicities and decreased bias. In addition, it helps to mitigate the performance gap between real and synthetic datasets. This approach addresses the limitations of synthetic data training, improving both the accuracy and fairness of face recognition models.

Abstract
The advent of 5th Generation (5G) networks has introduced the strategy of network slicing as a paradigm shift, enabling the provision of services with distinct Quality of Service (QoS) requirements. The 5th Generation New Radio (5G NR) standard complies with the use cases Enhanced Mobile Broadband (eMBB), Ultra-Reliable Low Latency Communications (URLLC), and Massive Machine Type Communications (mMTC), which demand a dynamic adaptation of network slicing to meet the diverse traffic needs. This dynamic adaptation presents both a critical challenge and a significant opportunity to improve 5G network efficiency. This paper proposes a Deep Reinforcement Learning (DRL) agent that performs dynamic resource allocation in 5G wireless network slicing according to traffic requirements of the 5G use cases within two scenarios: eMBB with URLLC and eMBB with mMTC. The DRL agent evaluates the performance of different decoding schemes such as Orthogonal Multiple Access (OMA), Non-Orthogonal Multiple Access (NOMA), and Rate Splitting Multiple Access (RSMA) and applies the best decoding scheme in these scenarios under different network conditions. The DRL agent has been tested to maximize the sum rate in scenario eMBB with URLLC and to maximize the number of successfully decoded devices in scenario eMBB with mMTC, both with different combinations of number of devices, power gains and number of allocated frequencies. The results show that the DRL agent dynamically chooses the best decoding scheme and presents an efficiency in maximizing the sum rate and the decoded devices between 84% and 100% for both scenarios evaluated.

Abstract
The current era where living demands an accelerated digital transition mainly focused on encouraging a smarter, healthier, and more sustainable mobility, in all its dimensions – a must concern for the young generations. The convergence through several digital services and APP can be an attitudes and perception changer within the group of academic mobility users’, promoting a more sustainable and better mobility choices that impact on the academic user’s mobility routines. Thus, encouraging a global shift to shared and active mobility services and systems bringing significant contributions to environmental sustainability and, also, to users’ health. The Academic Mobility as a Service (AMaaS) provide a digital service with mobility alternatives to support the academic population geographically located in different faculty campuses and Higher Education Institutions (HEI). The AMaaS applied to a restrict group is helpful to test innovative transport solutions and its high cybersecurity vulnerabilities. Despite the shortage of AMaaS case studies and the lack of security reference, it is imperative that a cybersecurity by design is planned and included in AMaaS design. In this paper AMaaS critical cybersecurity challenges, and potential risks are discussed and AMaaS Security by Design framework is described. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

Abstract
The lower refractive index sensitivity (RIS) of plasmonic nanoparticles (NP) in comparison to their plasmonic thin films counterparts hindered their wide adoption for wavelength-based sensor designs, wasting the NP characteristic field locality. In this context, high aspect-ratio colloidal core-shell Ag@Au nanorods (NRs) are demonstrated to operate effectively at telecommunication wavelengths, showing RIS of 1720 nm/RIU at 1350 nm (O-band) and 2325 nm/RIU at 1550 nm (L-band), representing a five-fold improvement compared to similar Au NRs operating at equivalent wavelengths. Also, these NRs combine the superior optical performance of Ag with the Au chemical stability and biocompatibility. Next, using a side-polished optical fiber, we detected glyphosate, achieving a detection limit improvement from 724 to 85 mg/L by shifting from the O to the C/L optical bands. This work combines the significant scalability and cost-effective advantages of colloidal NPs with enhanced RIS, showing a promising approach suitable for both point-of-care and long-range sensing applications at superior performance than comparable thin film-based sensors in either environmental monitoring and other fields.

Abstract
This project focuses on the development and implementation of a Mini Learning Factory (Mini LF) 5.0, aligned with the principles of Industry 5.0, Cyber-Physical Systems (CPS), and Open Design. Industry 5.0 emphasizes human-centric innovation, fostering collaboration between humans and machines while promoting sustainability. CPS facilitates the integration of the physical and digital realms, enabling more agile and flexible production processes. Open Design plays a pivotal role by encouraging collaborative participation, transparency, and the democratization of knowledge, which leads to more personalized and sustainable solutions in product and service design. The research adopts the Design Science Research (DSR) methodology, involving problem identification, artifact development, evaluation, and iterative improvement. The goal is to create a replicable, low-cost training environment that equips students with practical skills in line with Industry 5.0's requirements. The Mini LF 5.0 also aims to explore new methods for human-machine interaction, collaborative communication, and sustainable production, while ensuring the technical and financial viability of the project for wider adoption. © 2025 The Authors.

Abstract
In hyperspectral remote sensing, it is common to perform direct analysis of reflectance signals to identify key absorption features, and to apply techniques like the Spectral Angle Mapper to compare spectra and generate a similarity score. In this paper, we introduce the first application of the Continuous Wavelet Transform (CWT) in the context of hyperspectral remote sensing of marine plastic litter. First, we use the CWT to decompose plastic litter reflectance spectra from publicly available datasets and analyze its structure from the perspective of its frequency content at different wavelengths. Then, we propose a matching technique based on the cosine similarity of the magnitude gradients of the CWTs, named CWT Gradient Matching (CWTGM). Our results show that the CWT can be used to identify features which may otherwise prove difficult to analyze, and may also be useful in guiding sensor design. We also demonstrate that the CWTGM technique may be a viable option to measure similarity based on the frequency content of spectral reflectance signals. © 2025 Marine Technology Society.