Abstract
The development of deep learning algorithms has extensively empowered humanity's task automatization capacity. However, the huge improvement in the performance of these models is highly correlated with their increasing level of complexity, limiting their usefulness in human-oriented applications, which are usually deployed in resource-constrained devices. This led to the development of compression techniques that drastically reduce the computational and memory costs of deep learning models without significant performance degradation. These compressed models are especially essential when implementing multi-model fusion solutions where multiple models are required to operate simultaneously. This paper aims to systematize the current literature on this topic by presenting a comprehensive survey of model compression techniques in biometrics applications, namely quantization, knowledge distillation and pruning. We conduct a critical analysis of the comparative value of these techniques, focusing on their advantages and disadvantages and presenting suggestions for future work directions that can potentially improve the current methods. Additionally, we discuss and analyze the link between model bias and model compression, highlighting the need to direct compression research toward model fairness in future works.

Abstract
Lung cancer stands as the most prevalent and deadliest type of cancer, with adenocarcinoma being the most common subtype. Computed Tomography (CT) is widely used for detecting tumours and their phenotype characteristics, for an early and accurate diagnosis that impacts patient outcomes. Machine learning algorithms have already shown the potential to recognize patterns in CT scans to classify the cancer subtype. In this work, two distinct pipelines were employed to perform binary classification between adenocarcinoma and non-adenocarcinoma. Firstly, radiomic features were classified by Random Forest and eXtreme Gradient Boosting classifiers. Next, a deep learning approach, based on a Residual Neural Network and a Transformer-based architecture, was utilised. Both 2D and 3D CT data were initially explored, with the Lung-PET-CT-Dx dataset being employed for training and the NSCLC-Radiomics and NSCLC-Radiogenomics datasets used for external evaluation. Overall, the 3D models outperformed the 2D ones, with the best result being achieved by the Hybrid Vision Transformer, with an AUC of 0.869 and a balanced accuracy of 0.816 on the internal test set. However, a lack of generalization capability was observed across all models, with the performances decreasing on the external test sets, a limitation that should be studied and addressed in future work.

Abstract
Deep learning techniques have demonstrated significant potential in computer-assisted diagnosis based on medical imaging. However, their integration into clinical workflows remains limited, largely due to concerns about interpretability. To address this challenge, we propose Efficient-Proto-Caps, a lightweight and inherently interpretable model that combines capsule networks with prototype learning for lung nodule characterization. Additionally, an innovative Davies-Bouldin Index with multiple centroids per cluster is employed as a loss function to promote clustering of lung nodule visual attribute representations. When evaluated on the LIDC-IDRI dataset, the most widely recognized benchmark for lung cancer prediction, our model achieved an overall accuracy of 89.7 % in predicting lung nodule malignancy and associated visual attributes. This performance is statistically comparable to that of the baseline model, while utilizing a backbone with only approximately 2 % of the parameters of the baseline model’s backbone. State-of-the-art models achieved better performance in lung nodule malignancy prediction; however, our approach relies on multiclass malignancy predictions and provides a decision rationale aligned with globally accepted clinical guidelines. These results underscore the potential of our approach, as the integration of lightweight and less complex designs into accurate and inherently interpretable models represents a significant advancement toward more transparent and clinically viable computer-assisted diagnostic systems. Furthermore, these findings highlight the model’s potential for broader applicability, extending beyond medicine to other domains where final classifications are grounded in concept-based or example-based attributes. © 2013 IEEE.

Abstract

Abstract
The rapid proliferation of Internet of Things (IoT) systems, encompassing a wide range of devices and sensors with limited battery life, has highlighted the critical need for energy-efficient solutions to extend the operational lifespan of these battery-powered devices. One effective strategy for reducing energy consumption is minimizing the number and size of retransmitted packets in case of communication errors. Among the potential solutions, Incremental Redundancy Hybrid Automatic Repeat reQuest (IR-HARQ) communication schemes have emerged as particularly compelling options by adopting the best aspects of error control, namely, automatic repetition and variable redundancy. This work addresses the challenge by developing a simulator capable of executing and analysing several (H)ARQ schemes using different channel models, such as the Additive White Gaussian Noise (AWGN) and Gilbert-Elliott (GE) models. The primary objective is to compare their performance across multiple metrics, enabling a thorough evaluation of their capabilities. The results indicate that IR-HARQ outperforms alternative methods, especially in the presence of burst errors. Furthermore, its potential for further adaptation and enhancement opens up new ways for optimizing energy consumption and extending the lifespan of battery-powered IoT devices.

Abstract
The Sub-THz and millimeter-wave bands have gained popularity, with the expectation that they will host the next generation of wireless communication systems. Furthermore, research on beam-steering characteristics provided by Programmable Electromagnetic Surfaces, such as Reflective Intelligent Surfaces (RISs), has garnered considerable attention as an enabling technology for 6G communications. Due to size limitations, RISs face challenges related to power consumption in the reconfigurable elements and their integration with unit cells operating at high frequencies. This paper discusses the design of a 1-bit reconfigurable unit cell at the D-band using non-volatile technology to minimize static power consumption. Simulation results show that the proposed unit cell performs well with a reflection loss of less than 1.3 dB in both reconfigurable states across a frequency band from 120 to 170 GHz. Moreover, the phase difference between the two states is maintained at 180? ± 20?, with an operational bandwidth of approximately 16 GHz. The beamforming capabilities, with steering angles from -60? to 60?, of the 12×12 RIS, utilizing the proposed unit cell, have been demonstrated in terms of controlling the main beam radiation precisely to various angles with consistent performance at frequencies of 147 GHz, 152 GHz, and 152.5 GHz. © 2024 IEEE.