Search Results (123)

Customer churn prediction has become crucial for businesses, yet it poses significant challenges regarding privacy preservation and prediction accuracy. In this paper, we address two fundamental questions: (1) How can customer churn be effectively predicted while ensuring robust privacy protection of sensitive data? (2) How can large language models enhance churn prediction accuracy while maintaining data privacy? To address these questions, we propose FairRAG, a robust architecture that combines differential privacy, retrieval-augmented generation, and LLMs. Our approach leverages OPT-125M as the core language model along with a sentence transformer for semantic similarity matching while incorporating differential privacy mechanisms to generate synthetic training data. We evaluate FairRAG on two diverse datasets: Bank Churn and Telco Churn. The results demonstrate significant improvements over both traditional machine learning approaches and standalone LLMs, achieving accuracy improvements of up to 11% on the Bank Churn dataset and 12% on the Telco Churn dataset. These improvements were maintained when using differentially private synthetic data, thus indicating robust privacy and accuracy trade-offs. Full article

It is fundamental, yet challenging, to accurately predict water levels at hydrological stations located along the banks of an open channel river due to the complex interactions between different hydraulic structures. This paper presents a novel application for short-term multivariate prediction applied to hydrological variables based on a multivariate NARX model coupled to a nonlinear recursive Ensemble Kalman Filter (EnKF). The proposed approach is designed for two hydrological stations of the Atrato river in Colombia, where the variables, water level, water flow, and water precipitation, are correlated using a NARX model based on neural networks. The NARX model is designed to consider the complex dynamics of the hydrological variables and their corresponding cross-correlations. The short-term two-day water-level forecast is designed with a fourth-order NARX model. It is observed that the NARX model coupled with EnKF improves the robustness of the proposed approach in terms of external disturbances. Furthermore, the proposed approach is validated by subjecting the NARX–EnKF coupled model to five levels of additive white noise. The proposed approach employs metric regressions to evaluate the proposed model by means of the Root Mean Squared Error (RMSE) and the Nash–Sutcliffe model efficiency (NSE) coefficient. Full article

Adequate and diverse diets are essential for children’s physical and cognitive development, yet food insecurity and malnutrition continue to threaten this fundamental right, which remains a pressing concern in many resource-poor settings. This study investigated food and nutrition security in Early Childhood Development (ECD) centres in Makhanda, South Africa, through a community-based participatory research approach. Using a mixed-methods approach combining questionnaire interviews, focus group discussions, direct observations, and community asset mapping across eight ECD centres enrolling 307 children aged 0–5 years, the study engaged ECD facilitators and analysed dietary practices across these centres. Results indicated that financial constraints severely affect the quality and diversity of food provided at the centres, thus undermining the ability to provide nutritionally adequate meals. The average amount spent on food per child per month at the centres was R90 ± R25 (South African Rand). Although three meals were generally offered daily, cost-driven dietary substitutions with cheaper, less diverse alternatives, often at the expense of nutritional value, were common. Despite guidance from Department of Health dieticians, financial limitations contributed to suboptimal feeding practices, with diets dominated by grains and starchy foods, with limited access to and rare consumption of protein-rich foods, dairy, and vitamin A-rich fruits and vegetables. ECD facilitators noted insufficient parental contributions and low engagement in supporting centre operations and child nutrition provision, indicating a gap in awareness and limited nutrition knowledge regarding optimal infant and young child feeding (IYCF) practices. The findings emphasise the need for sustainable, multi-level and community-led interventions, including food gardening, creating ECD centre food banks, parental nutrition education programmes, and enhanced financial literacy among ECD facilitators. Strengthening local food systems and establishing collaborative partnerships with communities and policymakers are essential to improve the nutritional environment in ECD settings. Similarly, enhanced government support mechanisms and policy-level reforms are critical to ensure that children in resource-poor areas receive adequate nutrition. Future research should focus on scalable, locally anchored models for sustainable child nutrition interventions that are contextually grounded, community-driven, and should strengthen the resilience of ECD centres in South Africa. Full article

Although both noise suppression and speech restoration are fundamental to speech enhancement, many Deep neural network (DNN)-based approaches tend to focus disproportionately on one, often overlooking the importance of their joint handling. In this study, we propose a dual-path architecture designed to balance noise suppression and speech restoration. The main path consists of an encoder and two specialized decoders: one dedicated to estimating the clean speech spectrum and the other to predicting a noise suppression mask. To reinforce the joint modeling of noise suppression and speech restoration, we introduce an auxiliary refinement path. This path consists of a separate encoder–decoder structure and is designed to further refine the enhanced speech by incorporating complementary information, learned independently from the main path. By using this dual-path architecture, the model better preserves fine speech details while reducing residual noise. Experimental results on the VoiceBank + DEMAND dataset show that our model surpasses conventional methods across multiple evaluation metrics in the causal setup. Specifically, it achieves a PESQ score of 3.33, reflecting improved speech quality, and a CSIG score of 4.48, indicating enhanced intelligibility. Furthermore, it demonstrates superior noise suppression, achieving an SNRseg of 10.44 and a CBAK score of 3.75. Full article

Objectives: Codon usage bias is a fundamental feature of gene expression that can influence evolutionary processes and genetic diversity. This study aimed to investigate the mitochondrial codon usage characteristics and their driving forces in three Medicago species: Medicago polymorpha, Medicago sativa, and Medicago truncatula. Methods: The complete mitochondrial genome sequences of the three species were downloaded from GenBank, and 21 shared coding sequences were screened. Codon usage patterns were analyzed using CodonW 1.4.2 and CUSP software. Key parameters, including the relative synonymous codon usage (RSCU), effective number of codons (ENC), codon adaptation index (CAI), codon bias index (CBI), and frequency of optimal codons (Fop), were calculated. Phylogenetic trees and RSCU clustering maps were constructed to explore evolutionary relationships. Results: The GC contents of the mitochondrial genomes followed the order of GC1 > GC2 > GC3. ENC values averaged above 35, while CAI, CBI, and Fop values ranged from 0.160 to 0.161, −0.078 to −0.076, and 0.362 to 0.363, respectively, indicating a weak preference for codons ending with A/U. Correlation and neutrality analyses suggested that codon usage bias was influenced by both mutation pressure and natural selection, with natural selection being the dominant factor. Fifteen optimal codons, predominantly ending with A/U, were identified. Phylogenetic analysis confirmed the close relationship among the three Medicago species, consistent with traditional taxonomy, whereas the RSCU clustering did not align with the phylogenetic relationships. Conclusions: This study provides insights into the mitochondrial codon usage patterns and their evolutionary determinants in Medicago species, highlighting the predominant role of natural selection in shaping codon usage bias. The findings offer a foundation for comparative genomic studies and evolutionary analyses and may be beneficial for improving genetic engineering and breeding programs of Medicago species. Full article

Convolutional Neural Networks (CNNs) are fundamental to modern AI applications but often suffer from significant memory bottlenecks due to non-contiguous access patterns during convolution operations. Although previous work has optimized data layouts at the software level, hardware-level solutions for multi-memory accelerators remain underexplored. In this paper, we propose a hardware-level approach to mitigate memory row conflicts in FPGA-based CNN accelerators. Specifically, we introduce a dynamic DDR controller generated using Vivado 2019.1, which optimizes feature map allocation across memory banks and operates in conjunction with a multi-memory architecture to enable parallel access. Our method reduces row conflicts by up to 21% and improves throughput by 17% on the KCU1500 FPGA, with validation across YOLOv2, VGG16, and AlexNet. The key innovation lies in the layer-specific address mapping strategy and hardware-software co-design, providing a scalable and efficient solution for CNN inference across both edge and cloud platforms. Full article

The Number Theoretic Transform (NTT) is a cornerstone for efficient polynomial multiplication, which is fundamental to lattice-based cryptographic algorithms such as CRYSTALS-Kyber—a leading candidate in post-quantum cryptography (PQC). However, existing NTT accelerators often rely on integer multiplier-based modular reduction techniques, such as Barrett or Montgomery reduction, which introduce significant computational overhead and hardware resource consumption. These accelerators also lack optimization in unified architectures for forward (FNTT) and inverse (INTT) transformations. Addressing these research gaps, this paper introduces a novel, high-speed NTT accelerator tailored specifically for CRYSTALS-Kyber. The proposed design employs an innovative shift-add modular reduction mechanism, eliminating the need for integer multipliers, thereby reducing critical path delay and enhancing circuit frequency. A unified pipelined butterfly unit, capable of performing FNTT and INTT operations through Cooley–Tukey and Gentleman–Sande configurations, is integrated into the architecture. Additionally, a highly efficient data handling mechanism based on Register banks supports seamless memory access, ensuring continuous and parallel processing. The complete architecture, implemented in Verilog HDL, has been evaluated on FPGA platforms (Virtex-5, Virtex-6, and Virtex-7). Post place-and-route results demonstrate a maximum operating frequency of 261 MHz on Virtex-7, achieving a throughput of 290.69 Kbps—1.45× and 1.24× higher than its performance on Virtex-5 and Virtex-6, respectively. Furthermore, the design boasts an impressive throughput-per-slice metric of 111.63, underscoring its resource efficiency. With a 1.27× reduction in computation time compared to state-of-the-art single butterfly unit-based NTT accelerators, this work establishes a new benchmark in advancing secure and scalable cryptographic hardware solutions. Full article

Self-love is a fundamental psychological construct cultivated throughout human history. In Confucian culture, it is considered the ultimate Ren, while in ancient Greek thought, it serves as the center from which love radiates. Previous qualitative research identified five dimensions of self-love, but these lacked quantitative validation. This study developed the Chinese Adult Self-Love Scale (SLS) based on prior qualitative findings, constructing an initial item bank of 90 statements. The first study assessed the item relevance and clarity, resulting in a preliminary 68-item scale. Exploratory factor analysis (EFA) of 456 participants refined it to a 22-item scale with five indicators. The second study, with 929 participants, examined its reliability and validity. Cronbach’s α exceeded 0.75, and the test–retest reliability after six weeks was 0.66. Confirmatory factor analysis (CFA) supported the scale’s validity (CFI = 0.91, TFI = 0.90, RMSEA = 0.048). This study highlights self-love’s inheritance and development among Chinese adults, as well as its cross-cultural commonalities. It provides a valid, reliable tool for measuring self-love and offers a theoretical foundation for future cross-cultural research. While limitations exist, the findings suggest promising directions for further exploration. Full article

Understanding the time dynamics of systemic risk in banking networks is crucial for preventing financial crises and ensuring economic stability. This paper aims to quantify key transition times in the evolution of distress within a banking system using a mathematical framework. We investigate the dynamics of systemic risk in a hypothetical, homogeneous banking network using the Undistressed–Exposed–Distressed–Recovered (UEDR) model. The UEDR model, inspired by compartmental epidemic frameworks, captures how financial distress propagates and recedes through interactions between banks. It is selected because of its tractability and its ability to distinguish between different stages of bank vulnerability. We focus on two critical times, denoted as $t_1$ and $t_2$, which play a fundamental role in understanding the behavior of the distressed compartment (representing the number of distressed banks) over time. The time $t_1$ represents the first instance of a decrease in the number of distressed banks, indicating the containment of systemic risk. On the other hand, the time $t_2$ marks the onset when the number of undistressed banks falls below a specified threshold, signifying the restoration of financial stability. We examine these time dependencies by considering the initial conditions of the UEDR model and assess their characteristics using partial differential equations. We establish the continuity, smoothness, and uniqueness of solutions for $t_1$ and $t_2$, along with their corresponding boundary conditions. Furthermore, we provide explicit representation formulas for $t_1$ and $t_2$, allowing for precise estimation when the initial population compartments are large. Our results provide practical insights for financial regulators and policymakers in determining time-sensitive interventions for mitigating systemic risk and accelerating recovery in banking systems. The findings highlight how mathematical modeling can inform real-time risk management strategies in financial networks. Full article

This paper extends the research on cascaded-resonator (CR)-based filter banks introduced in previous studies. These IIR filter banks are online adaptive, making them highly suitable for spectral decomposition and signal analysis. The multiple-resonator-based structure offers an efficient design with low side lobes and high stopband attenuation. While earlier works focused on the fundamental structure and principles of these filter banks, the efficiency of their design was not thoroughly explored. In this study, thanks to the full periodicity of the frequency response, significant improvements in the modeling of the characteristic polynomial (the denominator of the transfer function) and preprocessing filters are introduced, resulting in an enhanced sparsity and a computational efficiency. Additionally, the previously employed linear programming algorithm for solving semi-infinite problems including a large number of linear constraints is replaced by more advanced quadratic programming (QP) or linear least-squares (LLS) optimization methods. These changes lead to a much faster and more powerful design process, even for filter banks with a larger number of resonator cascades and/or resonators per cascade. Furthermore, additional enhancements to the design methodology are proposed, further improving the robustness and applicability of the filters. These advancements enable the creation of highly efficient filter banks capable of handling complex and dynamic spectral analysis tasks in real time. Full article

Chemically identical chlorophyll (Chl) molecules undergo conformational changes when they are embedded in a protein matrix. The conformational changes will modulate their absorption spectra to meet the need for programmed excitation energy transfer or electron transfer. To interpret spectroscopic data using the knowledge of pigment–protein interactions requires a single pigment embedded in one polypeptide matrix. Unfortunately, most of the known photosynthetic systems contain a set of multiple pigments in each protein subunit. This makes it complicated to interpret spectroscopic data using structural data due to the potential overlapping spectra of two or more pigments. Chl–protein interactions have not been systematically studied to answer three fundamental questions: (i) What are the structural characteristics and commonly shared substructures of different types of Chl molecules (e.g., Chl a, b, c, d, and f)? (ii) How many structural groups can Chl molecules be divided into and how are different structural groups influenced by their surrounding environments? (iii) What are the structural characteristics of pigment surrounding environments? Having no clear answers to the unresolved questions is probably due to a lack of computational methods for quantifying conformational changes in individual Chls and individual surrounding amino acids. The first version of the Triangular Spatial Relationship (TSR)-based method was developed for comparing protein 3D structures. The input data for the TSR-based method are experimentally determined 3D structures from the Protein Data Bank (PDB). In this study, we take advantage of the 3D structures of Chl-binding proteins deposited in the PDB and the TSR-based method to systematically investigate the 3D structures of various types of Chls and their protein environments. The key contributions of this study can be summarized as follows: (i) Specific structural characteristics of Chl d and f were identified and are defined using the TSR keys. (ii) Two and three clusters were found for various types of Chls and Chls a, respectively. The signature structures for distinguishing their corresponding two and three clusters were identified. (iii) Histidine residues were used as an example for revealing structural characteristics of Chl-binding sites. This study provides evidence for the three unresolved questions and builds a structural foundation through quantifying Chl conformations as well as structures of their embedded protein environments for future mechanistic understanding of relationships between Chl–protein interactions and their corresponding spectroscopic data. Full article

In 2023, at the Center for Biological Resources (CRB) at the Institute of Genetics and Biophysics (IGB, Naples, Italy) of the National Research Council (CNR), the Breast Cancer Tissues and Organoids Biobank (BCTO BioBank) was founded. This is a new generation Biobank, dedicated to the collection, characterization, storage, and distribution of tissues and their 3D ‘organoid’ patients-derived. Tumor and healthy tissues from breast cancer patients have been collected from surgeons at Monaldi Hospital (Naples, Italy) and used to generate the corresponding tumor and healthy organoids from the same patient. After their establishment in culture, both organoids were characterized for their receptor status on a microfluidic 2-lane OrganoPlate, by immunofluorescence. The resulting data were compared with the expression profile obtained by immunohistochemistry on respective parental tissues. These data allowed us to phenotypically validate the generated organoids and classify them in a dedicated database, where also the clinical data of the corresponding patients were collected. During the six months of activities, we collected and characterized 27 samples. The continuous BCTO BioBank activity is fundamental to generating a high number of samples, for a broader and efficiently elaborated patient stratification at molecular level, biomarker discovery investigations, and for tailored treatment protocols design. Full article

The present work aims to fill the gaps in existing research on the application of machine learning in fraud detection and management in the banking sector. It provides a theoretical perspective on the evolution of algorithms, highlights practical aspects, and derives relevant metrics for evaluating their performance on unbalanced data. In the growing context of artificial intelligence, the adoption of an innovative, systematic approach to studying fraud in banking transactions through advanced machine learning algorithms is completely positive for the overall accuracy and effectiveness of risk management and has really practical and applied significance. The proven methodology (Classification and Regression Trees, Gradient Boosting, and Extreme Gradient Boosting) was tested on nearly 1.5 million in the banking sector, confirming the observations related to the application of fundamental assessments and specialized statistical methods through machine learning algorithms, demonstrating superior discriminatory power compared to classical models. The development provides valuable insights for managers, researchers, and policymakers aiming to strengthen the security and resilience of banking systems in times of evolving financial fraud challenges. Full article

Narrowband active noise control (NANC) systems are extensively used to cancel narrowband noise. Direct, parallel, and direct–parallel NANC systems use nonacoustic sensors to measure rotational speeds, and a bank of signal generators then produces synchronized sinusoidal waveforms as reference signals corresponding to the fundamental frequency of the undesired noise. The performance of direct NANC systems is based on the frequency difference between two adjacent reference input sinusoids. Parallel NANC systems apply several sinewave generators and two-weight adaptive filters in parallel to attenuate these narrowband components. Conventional direct–parallel NANC systems split these sinusoids into several mutually exclusive sets such that the distance between frequencies within sets is maximized. This paper proposes an improved direct–parallel NANC system in which reference sinusoidal signals are separated by amplitude to enhance efficiency and improve noise reduction performance. Several experiments were conducted using a muffler model to verify the performance of the proposed NANC system. Full article

Breastfeeding is fundamental for the development and protection of the newborn, and microorganisms present in breast milk are associated with the development of the infant’s intestinal microbiota. However, there are factors that interfere with breastfeeding, resulting in the need to supply donated milk to milk banks for these children. Even though there is a restriction on medications prescribed for pregnant and breastfeeding women, some antimicrobials are accepted, as long as they are used correctly and as they can increase the selection pressure for resistant bacteria. The microorganisms present in breast milk from a human milk bank were evaluated and the resistance of the isolates to antimicrobials was phenotypically characterized. In total, 184 microbial isolates were identified by mass spectrometry, of 12 bacterial genera and 1 yeast genus. There was a high prevalence of bacteria of the genus Staphylococcus, mainly S. epidermidis (33%). Resistance to antimicrobials varied among species, with a higher percentage of isolates resistant to penicillins and macrolides. Multidrug resistance was identified in 12.6% of 143 isolates. Breast milk contains a wide variety of microorganisms, mainly those of the Staphylococcus and Enterobacter genera. There was a high percentage of resistant isolates, and multidrug resistance in Klebsiella oxytoca (66.7%; 4/6) and S. epidermidis (15.0%; 9/60) isolates, which increases the public health concern. Full article