Search Results (1,512)

Feature selection is an important task in medical applications, given that the dimensionality and numerosity of such datasets are very high. In these cases, the time parameter also becomes important, along with classification accuracy, in estimating the performance of a learning model. This approach proposes intelligent agent teams that are capable of automatically discovering the best time to build models while keeping the general accuracy at the highest levels. For computing attributes’ relevance for the classification process, several techniques were used: Wrapper Evaluation, Information Gain, gain ratio, correlation, and Relief Attribute Evaluator. One of our contributions is the Threshold Agent, which evaluates the attributes as class attributes and considers the relevance of the attributes returned by the Wrapper method. This agent selects the strongest attributes (above a threshold value) and returns a subset that is learnt by the next attribute evaluation method within the Feature Selection Agent. The proposed agents discovered that an optimum subset composed of 20 attributes (out of 133 attributes of the initial dataset) leads to accuracy rates equal to the ones registered on the entire dataset, meaning 98%, using the Naive Bayes learning model, while improving the time taken to build the model from 0.1 s to 0.03 s. For the proposed dataset, Naïve Bayes outperformed other classification techniques, such as J48, Random Forest, and Dl4MlpClassifier. The proposed agents also integrated the best discovered model into a chatbot that performs medical diagnoses based on the symptoms collected from users. Full article

As electric vehicles (EVs) gain popularity, the existing public charging infrastructure is struggling to keep pace with the rapidly growing demand for the immediate charging needs of EVs. V2V power trading has gradually attracted widespread attention and development. EVs need to transmit sensitive information, such as transaction plans, through communication entities in the Internet of Vehicles (IoV). This could lead to leaks of sensitive information, thereby threatening the fairness of transactions. In addition, due to the differences in the cryptographic systems of entities, communication between entities faces challenges. Therefore, a privacy-preserving scheme for V2V double auction power trading based on heterogeneous signcryption and IoV is proposed. Firstly, a heterogeneous signcryption algorithm is designed to realize secure communication from certificateless cryptography to identity-based cryptography. Secondly, the scheme employs a pseudonym mechanism to protect the real identities of EVs. Furthermore, a verification algorithm is designed to verify the information sent by EVs and ensure the traceability and revocation of malicious EVs. The theoretical analysis shows that the proposed scheme could serve common security functions, and the experiment demonstrates that the proposed scheme reduces communication costs by about 14.56% and the computational cost of aggregate decryption by 80.51% compared with other schemes in recent years. Full article

In new wireless ecosystems, simultaneous wireless information and power transfer (SWIPT) and cooperative non-orthogonal multiple access (CNOMA) together make a potential design model. These systems enhance spectral efficiency (SE), energy efficiency (EE), and data interchange reliability by combining energy harvesting (EH), superposition coding (SC), and relay-assisted transmission. Despite this, CNOMA’s energy efficiency is still constrained by the fact that relay nodes servicing multiple users require a significant amount of power. Most previous studies look at performance as if imperfect successive interference cancellation (SIC) were possible. To solve these problems, this study presents a multiuser SWIPT-enabled cooperative wavelet NOMA (CWNOMA) framework that reduces imperfect SIC, inter-symbol interference (ISI), and inter-user interference. SWIPT-CWNOMA enhances overall energy efficiency (EE), keeps relays functional, and maintains data transmission strong for users by obtaining energy from received signals. The proposed architecture is evaluated against traditional CNOMA and orthogonal multiple access (OMA) in both perfect and imperfect scenarios with SIC. The authors derive closed-form formulas for EE, signal-to-interference-plus-noise ratio (SINR), and achievable rate to support the analysis. Residual error because of imperfect SIC for near users shows lower values in a varying range of SNR. Across 0–30 dB SNR, SWIPT-CWNOMA achieves, on average, $1.4$ times higher energy efficiency, approximately $4.7$ lower BER, and $1.9$ times higher achievable rate than OFDMA, which establishes SWIPT-CWNOMA as a promising candidate for next-generation energy-efficient wireless networks. Full article

A professional development program (PDP) combining direct and indirect promotion of self-regulated learning (SRL) was conducted with secondary school teachers in two parts. In the first part, the teachers were encouraged to promote student cognitive engagement through the inclusion of more interactive and constructive compared to passive and active lesson tasks in their teaching. In the second part, the teachers were provided with information which emphasized the importance of the direct promotion of SRL knowledge and strategies. The teachers were provided with excerpts from videos of classroom instruction to analyze and reflect upon. The results were based on an analysis of the talk and action of the teachers from videoed observations of their own classrooms before the PDP (Round 1), after the first part (Round 2), and after the second part (Round 3). The PDP influenced the teachers’ indirect promotion of SRL through the inclusion of more interactive and constructive and fewer passive and active lesson tasks in their teaching. Direct SRL promotion was also influenced although to a lesser extent, through the teachers’ inclusion of more motivational, metacognitive support statements to students to encourage them to keep on trying, as well as more explicit strategy promotion and reference to the benefits of SRL strategies. Full article

This paper proposes a novel intelligent switching control strategy for a five-bar lower limb exoskeleton. First, during the support phase, terminal sliding mode control (TSMC) is employed to ensure robust stability and high-torque amplification capabilities. Then, during the swing phase, a hybrid controller combining proportional-integral-derivative (PID) control and the adaptive neuro-fuzzy inference system (ANFIS) is implemented to generate natural and compliant leg movements. Finally, to achieve smooth transitions between phases, an intelligent switching algorithm based on multi-sensor information fusion is proposed. Simulation results demonstrate that the proposed strategy keeps trajectory tracking errors below 0.05 across all gait phases and achieves stable torque amplification ratios ranging from 1:6 to 1:10. This performance significantly reduces the user’s physical exertion. These findings validate the effectiveness of this control framework in improving the stability and comfort of human–machine interaction. Full article

Few-Shot Class-Incremental Learning (FSCIL) aims to continually learn novel classes from limited data while retaining knowledge of previously learned classes. To mitigate catastrophic forgetting, most approaches pre-train a powerful backbone on the base session and keep it frozen during incremental sessions. Within this framework, existing studies primarily focus on representation learning in FSCIL, particularly Self-Supervised Contrastive Learning (SSCL), to enhance the transferability of representations and thereby boost model generalization to novel classes. However, they face a trade-off dilemma: improving transferability comes at the expense of discriminability, precluding simultaneous high performance on both base and novel classes. To address this issue, we propose BR-FSCIL, a representation learning framework for the FSCIL scenario. In the pre-training stage, we first design a Hierarchical Contrastive Learning (HierCon) algorithm. HierCon leverages label information to model hierarchical relationships among features. In contrast to SSCL, it maintains strong discriminability when promoting transferability. Second, to further improve the model’s performance on novel classes, an Alignment Modulation (AM) loss is proposed that explicitly facilitates learning of knowledge shared across classes from an inter-class perspective. Building upon the hierarchical discriminative structure established by HierCon, it additionally improves the model’s adaptability to novel classes. Through optimization at both intra-class and inter-class levels, the representations learned by BR-FSCIL achieve a balance between discriminability and transferability. Extensive experiments on mini-ImageNet, CIFAR100, and CUB200 demonstrate the effectiveness of our method, which achieves final session accuracies of 53.83%, 53.04%, and 62.60%, respectively. Full article

Vehicular Ad Hoc Network (VANET) is a fundamental component of the intelligent transportation systems (ITS), providing critical road information to users. However, the volatility of VANETs creates significant vulnerabilities from malicious actors. Thus, verifying joining entities is crucial to maintaining the VANET’s communication security. Authentication delays must stay below 100 ms to meet VANET requirements, posing a major challenge for security. Our previous research introduced a Kerberos–Blockchain (KBC) authentication system that contains two main components separately: Authentication Server (AS) and Ticket Granting Server (TGS). However, this KBC architecture required an additional server to accommodate increasing vehicle volumes in urban environments, leading to higher infrastructure costs. This paper presents an integrated authentication server that merges AS and TGS into a Combined Server (CBS) while retaining blockchain security. We evaluate it using OMNeT++ with SUMO for traffic simulation and Ganache for blockchain implementation. Results show that CBS removes the need for an extra server while keeping authentication delays under 100 ms. It also improves throughput by $104\%$ and reduces signaling overhead by $45\%$ compared to KBC. By optimizing authentication without compromising security, the integrated server greatly enhances the cost-effectiveness and efficiency of VANET systems. Full article

Background: There is varied practice with Disease-Modifying Treatment (DMT) for Multiple Sclerosis worldwide. We evaluated early DMT initiation within a real-world population for long-term outcomes. Method: The Scottish Multiple Sclerosis Register (SMSR) identified participants diagnosed with Relapsing Remitting Multiple Sclerosis (RRMS) in 2010/2011. We compared two groups of propensity-matched participants at diagnosis, who went on to receive either early treatment (<12 months from diagnosis) or late/never treated. Participants underwent detailed clinicoradiological evaluation and patient-reported outcome measures 11–13 years post-diagnosis. The primary outcome was mean Expanded Disability Status Scale (EDSS). Results: The SMSR identified 298 participants. A total of 141 had complete retrospective clinical data and 81 agreed to participate, with 32 successfully matched (16 pairs). Median time on DMT was 10.8 years (range 0.4–12.5) for those treated early and 4.0 (0–11.5) years for the late/never-treated group. A total of 7/16 (44%) never received a DMT of those not treated early. All early-treated participants commenced first-line DMT (5/16 subsequently escalated to second-line DMTs). Of those treated later (9/16), 7/9 participants (78%) commenced first-line and 2/9 s-line DMT. There were no serious adverse events identified with any DMT. There was no significant difference in the primary outcome, with mean EDSS 3.93 in the late/never-treated group vs. 4.53 in the early-treated group at 11–13 years post-diagnosis (p = 0.57). There was no significant difference in median change in EDSS from the time of diagnosis to prospective assessment between early and late/never-treated groups. Patient Reported Outcome Measurement Information System (PROMIS) scores for cognition favoured no early treatment (p = 0.02), whilst satisfaction with treatment choice favoured early treatment (p = 0.03). Conclusions: Our cohort did not show clear benefit with early DMT in RRMS, contrasting with other larger studies, with no significant differences between early and late/never-treated patients on clinicoradiological outcomes. Possible explanations include confounding by variables not included in matching and group allocation based on diagnosis date rather than first clinical symptom. Most participants were treated with injectable DMTs, not in keeping with current practice. A prospective, long-term follow-up deep phenotyping study would help characterise benefits of early DMT use, but this is clearly challenging in practice. Full article

Quantum key distribution (QKD) networks promise information-theoretic security for multiple nodes by leveraging the fundamental laws of quantum mechanics. In practice, QKD networks require dedicated routing protocols to coordinate secure key distribution among distributed nodes. However, most existing routing protocols operate under the assumption that all relay nodes are honest and fully trustworthy, an assumption that may not hold in realistic scenarios. Malicious nodes may tamper with routing updates, causing inconsistent key-state views or divergent routing plans across the network. Such inconsistencies increase routing failure rates and lead to severe wastage of valuable secret keys. To address these challenges, we propose a distributed routing framework that combines two key components: (i) Causal Consistency Key-State Update, which prevents malicious nodes from propagating inconsistent key states and routing plans; and (ii) Trust-Aware Multi-path Flow Optimization, which incorporates trust metrics derived from discrepancies in reported states into the path-selection objective, penalizing suspicious links and filtering fabricated demands. Across 50-node topologies with up to 30% malicious relays and under all three attack modes, our protocol sustains a high demand completion ratio (DCR) (mean $0.90$, range $0.81$–$0.98$) while keeping key utilization low ($16.6$ keys per demand), decisively outperforming the baselines—Multi-Path Planned (DCR $0.48$, $30.8$ keys per demand) and OSPF (DCR $\le 0.12$, 296 keys per demand; max 1601). These results highlight that our framework balances reliability and efficiency, providing a practical and resilient foundation for secure QKD networking in adversarial environments. Full article

Background: During the COVID-19 pandemic, transgender and gender-diverse (TGD) people and people with disabilities in India faced disproportionate barriers to accessing vaccination services. Building on previous studies, this study explored the experiences of COVID-19 vaccine access in these two marginalized communities, using the WHO Behavioral and Social Drivers (BeSD) framework. Methods: Keeping community-based participatory methods (CBPR) at heart, we conducted a survey adapted from the BeSD COVID-19 survey tool. The survey was adapted using insights from a prior study, a literature review, stakeholder consultations, and discussions with a community leadership group (CLG) and an advisory board (AdB). Participants were recruited through transgender, gender-diverse, and disability rights networks. Data were analyzed descriptively, using percent analysis, and psychometrically, using exploratory factor analysis on polychoric correlations. Results: The adapted BeSD survey tool showed a high 0.85 (p < 0.05) internal consistency and criterion validity. Moreover, it showed a high willingness to be vaccinated (for ease of access to other services and community responsibility); however, systemic barriers hindered vaccination access. TGD people and people with disabilities faced multiple barriers in being vaccinated. The TGD community reported documentation mismatches and mistrust in health systems. People with disabilities reported mobility challenges, escort dependence, financial challenges, and variable accessibility at vaccination sites. Both groups faced digital exclusion, received inadequate information that did not address their specific needs, and experienced inconsistent implementation of inclusive policies. Community-led facilitation led to more uptake. Conclusions: Vaccine willingness alone is insufficient to ensure that vaccines reach everyone. Addressing trust deficits, infrastructural barriers, and digital exclusions requires diligent attention and commitment from the government to mitigate broader challenges faced by TGD people and people with disabilities. Full article

Over one million international students from 207 countries study in the United States to pursue their academic goals. Transitioning to an unfamiliar country presents numerous challenges, and existing support structures often fail to fully support international students. Pet companionship may support students in alleviating homesickness and enhancing mental well-being. However, there is a lack of research exploring the experience of international students in the U.S. living with pets and what unique barriers they face. This quantitative survey recruited 662 international students to explore why they may or may not choose to live with pets while they are in the U.S. and the challenges they face regarding having pets while studying abroad. Participants reported barriers such as financial and housing restrictions, as well as concerns about pet care during travel or vacations and uncertainty about their future plans, which deter them from committing to long-term pet ownership. However, most of the participants who had experience living with pets or planned to have a pet believed that the benefits of having a pet outweighed the challenges. More than 60% of the participants were committed to keeping their pets permanently, even if they needed to move back to their home country or to another foreign country. While results are limited to a non-representative sample of international students, this research provides insights that may inform how to enrich support systems for both international students and animal welfare by highlighting the unique challenges and benefits of human–animal interactions for international students. Full article

This study examines the use and functions of discourse markers (DMs) in nonnative English speech produced by Arab English speakers. Four DMs (and, but, so, y’know) are analysed based on two theoretical frameworks: Schiffrin’s (1987) framework of functions of DMs and Schourup’s (1999) characterisation of DMs. Semi-structured interviews were conducted with 10 Arab English speakers. The findings show clear patterns in the participants’ use of DMs. The marker and is mainly used for its ideational function, that is, for connecting ideas, events, and positions to keep the discourse together, while pragmatic functions, such as continuing an action or turn organisation, are less represented. But is mainly used for its ideational function, indicating contrastive ideas; less frequently, it is used for such pragmatic functions as returning to a topic, making disclaimers, reclaims, or showing functional contrasts. In contrast, so demonstrates a broader range of functions; while it occasionally marks results at the ideational level, it is extensively utilised pragmatically for marking claims, compliance, requests, and topic transitions, as well as managing turn initiation and adjacency pairs. y’know is used for such pragmatic functions as organising shared knowledge, signalling significant information or disapproval in stories, and appealing. Overall, the findings in the study suggest that, in the narrative register, when using DMs, Arab English speakers rely chiefly on their ideational functions. In contrast, their pragmatic functions are used much less, except for so and y’know, which also show more diversified functions. Full article

Infertility is one of the major farming constraints facing smallholder dairy cattle farming in Tanzania. Despite its impact, there is limited information on how farmers understand and manage it. The present study aimed to assess farmers’ knowledge, attitudes and practices related to dairy cattle infertility. A cross-sectional survey was conducted using a structured questionnaire involving 301 farmers across six major dairy-farming regions: Tanga, Arusha, Kilimanjaro, Mbeya, Morogoro and Njombe. Overall, 95% of respondents reported encountering infertility on their farms. Farmers were asked to identify signs of infertility from the list of 10 (8 correct and 2 distractors); the median score for correct identification was 7 (range 2–10). The most recognised sign was return to oestrus after insemination (94%). Most farmers correctly identified low milk yield and mastitis as not being signs of infertility. The main reported causes included poor nutrition/housing (93%), livestock diseases (89%), poor record keeping (85%), and poor oestrus detection (83%). Nearly all (98%) viewed infertility as a serious issue, predominantly naming repeat breeding (95%) and failure to produce a calf/year (90%). Management strategies included seeking veterinary services (94%), slaughter (69%), sell to other farmers (23%) and self-treatment (16%). Our findings highlight widespread awareness of infertility while pointing out gaps in management, which reinforces the need for improved farmer education and support services. Full article

The rapid growth of the semiconductor industry has created a bottleneck in which traditional manual methods for designing fabrication plants (fabs) cannot keep pace with their high complexity and short technological lifecycles. This problem stems from the critical mismatch between a fab’s multiyear construction timeline and the rapidly shrinking lifecycle of the advanced chips it is built to produce. To address this challenge, the present study proposes a methodological framework that uses dynamic generative design within a Building Information Modelling (BIM) environment. This approach applies algorithms to generalized models to generate and evaluate numerous potential design solutions automatically. For facility layouts, the framework produces plans that balance spatial efficiency, material flow, and stringent cleanroom protocols. For complex utility systems, it moves beyond simple clash detection to proactively generate resource-efficient, clash-free routing paths that consider both constructability and long-term maintainability. The primary contribution of this study is a standardized, data-agnostic design process that enhances design quality without requiring sensitive project data, establishing a robust foundation for future Digital Twin integration. Full article

This paper presents a novel submersible seaweed cultivation infrastructure designed to enhance seaweed growth through deep cycling. The system consists of a square grid of ropes for growing seaweed, supported by buoys, mooring lines, and innovative SubTractors—movable buoys that enable controlled submersion. The grid ropes are stabilized by four SubTractors, an array of small buoys, intermediate sinker weights and mooring lines anchored to the seabed. The SubTractors facilitate dynamic positioning, allowing the seaweed rope grid to be submerged below the thermocline—at depths of 100 m or more—where nutrient-rich deep water accelerates seaweed growth in offshore sites with low surface nutrient levels. Small buoys attached to the grid provide buoyancy, keeping the seaweed rope grid planar and near the surface to optimize photosynthesis when not submerged. This paper first describes the seaweed cultivation infrastructure, then develops a hydroelastic model of the proposed cultivation system, followed by a hydroelastic analysis under varying wave and current conditions. The results provide insights into the system’s dynamic behaviour, informing engineering design and structural optimization. Full article