Search Results (71)

As voice cloning technology rapidly advances, the risk of personal voices being misused by malicious actors for fraud or other illegal activities has significantly increased, making the collection of speech data increasingly challenging. To address this issue, this study proposes a data augmentation method based on XText-to-Speech (XTTS) synthesis to tackle the challenges of small-sample, multi-class speech recognition, using profanity as a case study to achieve high-accuracy keyword recognition. Two models were therefore evaluated: a CNN model (Proposed-I) and a CNN-Transformer hybrid model (Proposed-II). Proposed-I leverages local feature extraction, improving accuracy on a real human speech (RHS) test set from 55.35% without augmentation to 80.36% with XTTS-enhanced data. Proposed-II integrates CNN’s local feature extraction with Transformer’s long-range dependency modeling, further boosting test set accuracy to 88.90% while reducing the parameter count by approximately 41%, significantly enhancing computational efficiency. Compared to a previously proposed incremental architecture, the Proposed-II model achieves an 8.49% higher accuracy while reducing parameters by about 98.81% and MACs by about 98.97%, demonstrating exceptional resource efficiency. By utilizing XTTS and public corpora to generate a novel keyword speech dataset, this study enhances sample diversity and reduces reliance on large-scale original speech data. Experimental analysis reveals that an optimal synthetic-to-real speech ratio of 1:5 significantly improves the overall system accuracy, effectively addressing data scarcity. Additionally, the Proposed-I and Proposed-II models achieve accuracies of 97.54% and 98.66%, respectively, in distinguishing real from synthetic speech, demonstrating their strong potential for speech security and anti-spoofing applications. Full article

This study presents a hybrid sensor placement methodology that combines criterion-based candidate selection with advanced optimization algorithms. Four established selection criteria—modal kinetic energy (MKE), modal strain energy (MSE), modal assurance criterion (MAC) sensitivity, and mutual information (MI)—are used to evaluate DOF sensitivity and generate candidate pools. These are followed by one of four optimization algorithms—greedy, genetic algorithm (GA), particle swarm optimization (PSO), or simulated annealing (SA)—to identify the optimal subset of sensor locations. A key feature of the proposed approach is the incorporation of constraint dynamics using the Udwadia–Kalaba (U–K) generalized inverse formulation, which enables the accurate expansion of structural responses from sparse sensor data. The framework assumes a noise-free environment during the initial sensor design phase, but robustness is verified through extensive Monte Carlo simulations under multiple noise levels in a numerical experiment. This combined methodology offers an effective and flexible solution for data-driven sensor deployment in structural health monitoring. To clarify the rationale for using the Udwadia–Kalaba (U–K) generalized inverse, we note that unlike conventional pseudo-inverses, the U–K method incorporates physical constraints derived from partial mode shapes. This allows a more accurate and physically consistent reconstruction of unmeasured responses, particularly under sparse sensing. To clarify the benefit of using the U–K generalized inverse over conventional pseudo-inverses, we emphasize that the U–K method allows the incorporation of physical constraints derived from partial mode shapes directly into the reconstruction process. This leads to a constrained dynamic solution that not only reflects the known structural behavior but also improves numerical conditioning, particularly in underdetermined or ill-posed cases. Unlike conventional Moore–Penrose pseudo-inverses, which yield purely algebraic solutions without physical insight, the U–K formulation ensures that reconstructed responses adhere to dynamic compatibility, thereby reducing artifacts caused by sparse measurements or noise. Compared to unconstrained least-squares solutions, the U–K approach improves stability and interpretability in practical SHM scenarios. Full article

Due to the dynamic and harsh underwater environment, which involves a long propagation delay, high bit error rate, and limited bandwidth, it is challenging to achieve reliable communication in underwater wireless sensor networks (UWSNs) and network support applications, like environmental monitoring and natural disaster prediction, which require energy efficiency and low latency. To tackle these challenges, we introduce AC-RL-based power control (ACRLPC), a novel hybrid MAC protocol that can efficiently integrate Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA)-based MAC and Time Division Multiple Access (TDMA) with Actor–Critic Reinforcement Learning (AC-RL). The proposed framework employs adaptive strategies, utilizing adaptive power control and intelligent access methods, which adjust to fluctuating conditions on the network. Harsh and dynamic underwater environment performance evaluations of the proposed scheme confirm a significant outperformance of ACRLPC compared to the current protocols of FDU-MAC, TCH-MAC, and UW-ALOHA-QM in all major performance measures, like energy consumption, throughput, accuracy, latency, and computational complexity. The ACRLPC is an ultra-energy-efficient protocol since it provides higher-grade power efficiency by maximizing the throughput and limiting the latency. Its overcoming of computational complexity makes it an approach that greatly relaxes the processing requirement, especially in the case of large, scalable underwater deployments. The unique hybrid architecture that is proposed effectively combines the best of both worlds, leveraging TDMA for reliable access, and the flexibility of CSMA/CA serves as a robust and holistic mechanism that meets the desired enablers of the system. Full article

Chinese, a tonal language with inherent homophonic ambiguity, poses significant challenges for semantic disambiguation in natural language processing (NLP), hindering applications like speech recognition, dialog systems, and assistive technologies. Traditional static disambiguation methods suffer from poor adaptability in dynamic environments and low-frequency scenarios, limiting their real-world utility. To address these limitations, we propose BLAF—a novel MacBERT-BiLSTM Hybrid Architecture—that synergizes global semantic understanding with local sequential dependencies through dynamic multimodal feature fusion. This framework incorporates innovative mechanisms for the principled weighting of heterogeneous features, effective alignment of representations, and sensor-augmented cross-modal learning to enhance robustness, particularly in noisy environments. Employing a staged optimization strategy, BLAF achieves state-of-the-art performance on the SIGHAN 2015 (data fine-tuning and supplementation): 93.37% accuracy and 93.25% F1 score, surpassing pure BERT by 15.74% in accuracy. Ablation studies confirm the critical contributions of the integrated components. Furthermore, the sensor-augmented module significantly improves robustness under noise (speech SNR to 18.6 dB at 75 dB noise, 12.7% reduction in word error rates). By bridging gaps among tonal phonetics, contextual semantics, and computational efficiency, BLAF establishes a scalable paradigm for robust Chinese homophone disambiguation in industrial NLP applications. This work advances cognitive intelligence in Chinese NLP and provides a blueprint for adaptive disambiguation in resource-constrained and dynamic scenarios. Full article

This study presents a hybrid computational investigation into the radiation shielding behavior of Fe-enriched Al-based alloys (Al-Fe-Mo-Si-Zr) for potential use in nuclear applications. Four alloy compositions with varying Fe contents (7.21, 6.35, 5.47, and 4.58 wt%) were analyzed using a combination of Monte Carlo simulations, machine learning (ML) predictions based on multilayer perceptrons (MLPs), EpiXS, and SRIM-based charged particle transport modeling. Key photon interaction parameters—including mass attenuation coefficient (MAC), half-value layer (HVL), buildup factors, and effective atomic number (Z_eff)—were calculated across a wide energy range (0.015–15 MeV). Results showed that the 7.21Fe alloy exhibited a maximum MAC of 12 cm²/g at low energies and an HVL of 0.19 cm at 0.02 MeV, indicating improved gamma attenuation with increasing Fe content. The ML model accurately predicted MAC values in agreement with Monte Carlo and XCOM data, validating the applicability of AI-assisted modeling in material evaluation. SRIM calculations demonstrated enhanced charged particle shielding: the projected range of 10 MeV protons decreased from ~55 µm (low Fe) to ~50 µm (high Fe), while alpha particle penetration reduced accordingly. In terms of fast neutron attenuation, the 7.21Fe alloy reached a maximum removal cross-section (Σ_R) of 0.08164 cm⁻¹, showing performance comparable to conventional materials like concrete. Overall, the results confirm that Fe-rich Al-based alloys offer a desirable balance of lightweight design, structural stability, and dual-function radiation shielding, making them strong candidates for next-generation protective systems in high-radiation environments. Full article

Machine-to-machine (M2M) communications are becoming the most important factor shaping network traffic. However, traditional controls developed for human-generated traffic are not able to cope with new demands. Thus, hybrid MAC protocols have been proposed to make use of the combined advantages of contention and reservation. Most of them are based on a contention stage (where a variant of CSMA/CA or ALOHA is used) followed by a reservation stage (e.g., TDMA or FDMA). In this paper, we introduce 2CA-R², a hybrid MAC protocol for M2M communications intended to be used in the device domain. What distinguishes this proposal is that the contention stage is controlled by a conflict–resolution algorithm known as Adaptive-2C. The protocol was evaluated using a model based on a Markov chain and computer simulations. Its performance was compared with DCF, the MAC technique used in IEEE802.11 standards. Our results show significant improvements over DCF in various metrics of network performance across different traffic situations. We also evaluated the time the protocol takes to resolve an access conflict, and we observed substantial improvements in the number of stations that can be served with the same network resource (in some cases, around a 40% improvement). Full article

The increasing prevalence and dissemination of multidrug-resistant bacteria represent a serious concern for public health. Aeromonas caviae is a pathogenic microorganism that causes a wide spectrum of diseases in fish and humans and is often associated with aquatic environments and isolated from foods and animals. Here, we present the isolation and characterization of the V15^T strain isolated from a drinking water storage tank in Rio de Janeiro, Brazil. The V15^T strain has a genome length of 4,443,347 bp with an average G + C content of 61.78% and a total of 4028 open reading frames. Its genome harbors eight types of antibiotic resistance genes (ARGs) involving resistance to beta-lactamases, macrolides, and quinolones. The presence of bla_MOX-6, bla_OXA-427/bla_OXA-504, and mutations in parC were detected. In addition, other ARGs (macA, macB, opmH, and qnrA) and multidrug efflux pumps (such as MdtL), along with several resistance determinants and 106 genes encoding virulence factors, including adherence (polar and lateral flagella), secretion (T2SS, T6SS), toxin (hlyA), and stress adaptation (katG) systems, were observed. The genome sequence reported here provides insights into antibiotic resistance, biofilm formation, evolution, and virulence in Aeromonas strains, highlighting the need for more public health attention and the further monitoring of drinking water systems. Also, the results of physiological and phylogenetic data, average nucleotide identity (ANI) calculation, and digital DNA–DNA hybridization (dDDH) analysis support the inclusion of the strain V15^T in the genus Aeromonas as a new subspecies with the proposed name Aeromonas caviae subsp. aquatica subsp. nov. (V15^T = P53320^T). This study highlights the genomic plasticity and pathogenic potential of Aeromonas within household drinking water systems, calling for the revision of water treatment protocols to address biofilm-mediated resistance and the implementation of routine genomic surveillance to mitigate public health risks. Full article

The science of biogeography was described through mathematical equations in 1967 by Robert MacArthur and Edward Wilson. In 2008, Dan Simon presented an algorithm called biogeography-based optimization, or BBO, which used some of the principles and definitions described in MacArthur and Wilson’s book. The objectives of this work were to study the behavior of the BBO method when it is hybridized with other evolutionary search methods and to analyze the effect of its application to some examples of mechanical engineering systems. The operators considered in the hybridization study are genetic recombination (crossover) and local search, aiming to overcome the limitations and difficulties that arise when using the original BBO. The results of the original BBO were promising in the context of a global search. However, there is a diversity problem that does not allow for good quality increments in the final phase of the evolutionary process. The additional modifications included, such as the concept of blending in migration, the cycle of mutations and the replacement of the worst solutions by injection of new ones, all show positive effects on the method’s performance. However, the biggest increase happened with the implementation of the hybridization processes. Crossover improved the speed and diversity of the population in some cases, while local search helped the algorithm in later generations, allowing it to quickly reach the optimum point. With this mentioned, it is important to note that the best results were all obtained with the fully modified algorithm. Statistical tests were implemented to validate the significance of changes due to modifications included in the original proposal of BBO. Full article

Banana streak GF virus (BSGFV) is the extremely dangerous monopartite badnavirus (genus, Badnavirus; family, Caulimoviridae) of banana (Musa acuminata AAA Group) that imposes a serious threat to global banana production. The BSGFV causes a devastating pandemic in banana crops, transmitted by deadly insect pest mealybug vectors and replicated through an RNA intermediate. The BSGFV is a reverse-transcribing DNA virus that has a monopartite open circular double-stranded DNA (dsDNA) genome with a length of 7325 bp. RNA interference (RNAi) is a natural mechanism that has revolutionized the target gene regulation of various organisms to combat virus infection. The current study aims to locate the potential target binding sites of banana-encoded microRNAs (mac-miRNAs) on the BSGFV-dsDNA-encoded mRNAs based on three algorithms, RNA22, RNAhybrid and TAPIR. Mature banana (2n = 3x = 33) miRNAs (n = 32) were selected and hybridized to the BSGFV genome (MN296502). Among the 32 targeted mature locus-derived mac-miRNAs investigated, two banana mac-miRNA homologs (mac-miR162a and mac-miR172b) were identified as promising naturally occurring biomolecules to have binding affinity at nucleotide positions 5502 and 9 of the BSGFV genome. The in silico banana-genome-encoded mac-miRNA/mbg-miRNA-regulatory network was developed with the BSGFV—ORFs using Circos software (version 0.69-9) to identify potential therapeutic target proteins. Therefore, the current work provides useful biological material and opens a new range of opportunities for generating BSGFV-resistant banana plants through the genetic manipulation of the selected miRNAs. Full article

Tsunamis are devastating natural phenomena that cause extensive damage to both human life and infrastructure. To mitigate such impacts, tsunami early warning systems have been deployed globally. South Korea has also initiated a project to install a tsunami warning system to monitor its surrounding seas. To ensure reliable warning decisions, various types of data must be combined, but efficiently transmitting heterogeneous data poses a challenge due to the unique characteristics of underwater acoustic communication. Therefore, this paper proposes a Hybrid Duplex Medium Access Control (HDMAC) protocol designed for a tsunami warning system, with a specific focus on heterogeneous data transmission. HDMAC efficiently handles both seismic and environmental data by utilizing hybrid duplexing, which combines frequency duplex for seismic data with time duplex for environmental data. The protocol addresses the distinct transmission requirements for each data type by optimizing channel utilization through a group Automatic Repeat request (ARQ) scheme and packet size adjustment. Theoretical analysis predicts that HDMAC can achieve a channel utilization of up to 0.91 in smaller networks and 0.64 in larger networks. HDMAC is validated through simulations, and the simulation results closely match these predictions. The simulation results demonstrate the efficiency of HDMAC in supporting real-time submarine earthquake monitoring systems. Full article

In the framework of the space-air-ground-ocean integrated network, the underwater acoustic sensor network (UASN) plays a pivotal role. The design of media access control (MAC) protocols is essential for the UASN to ensure efficient and reliable data transmission. From the perspective of differentiated services in the UASN, a service-aware and scheduling-based hybrid MAC protocol, named the SSH-MAC protocol, is proposed in this paper. In the SSH-MAC protocol, the centralized scheduling strategy is adopted by sensor nodes with environmental monitoring service, and the distributed scheduling strategy is adopted by sensor nodes with target detection service. Considering the time-varying data generation rate of sensor nodes, the sink node will switch the scheduling mode of sensor nodes based on the specific control packet and adjust the resource allocation ratio between centralized scheduling and distributed scheduling. Simulation results show that the performance of the SSH-MAC protocol, in terms of utilization, end-to-end delay, packet delivery ratio, energy consumption, and payload efficiency, is good. Full article

Wireless sensor networks (WSNs) based on the IEEE 802.15.4 standard have important applications in many fields, such as the Internet of Things and smart cities because of their low energy consumption. Hybrid carrier sense multiple access/time division multiple access (CSMA/TDMA) is the key technique to reduce energy consumption in the standard, but it also increases packet delay and reduces network throughput. Although the cluster tree topology is a typical topology defined by the IEEE 802.15.4 standard, there are few efficient medium access control (MAC) protocols specifically for this type of topology. To this end, we present an improved hybrid CSMA/TDMA MAC protocol based on a sharable slot algorithm for WSNs with cluster tree topology, called sharable slot-based MAC (SS-MAC). By designing the operating mechanism and frame structure, improving the hybrid CSMA/TDMA and channel-hopping techniques of IEEE 802.15.4 MAC, and introducing a sharable slot algorithm to wake up tree nodes asynchronously as well as a short address strategy to identify member nodes, the proposed protocol improves packet delay and throughput under the premise of low collision and low node energy consumption. Moreover, we derive mathematical expressions of the parameters of the sharable slot algorithm and evaluate the energy consumption, throughput and packet delay of the SS-MAC based on the queue modeling of packet arrivals. Numerical simulations verify that the proposed MAC protocol outperforms the other three existing MAC protocols, namely, IEEE 802.15.4 MAC, SSMA and LELLMAC, in terms of energy consumption, throughput and packet delay. Full article

During the coronavirus disease (COVID-19) pandemic healthcare workers (HCWs) acquired immunity by vaccination or exposure to multiple variants of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Our study is a comparative analysis between subgroups of HCWs constructed based on the number of SARS-CoV-2 infections, vaccination, and the dominant variant of SARS-CoV-2 in the population. We collected and analyzed data using the χ² test and density incidence of reinfections in Microsoft Excel for Mac, Version 16.84, and MedCalc^®, 22.026. Of the 829 HCWs, 70.1% (581) had only one SARS-CoV-2 infection and 29.9% (248) had two infections. Of the subjects with two infections, 77.4% (192) worked in high-risk departments and 93.2% (231) of the second infections were registered during Omicron dominance. The density incidence of reinfections was higher in HCWs vaccinated with the primary schedule than those vaccinated with the first booster, and the incidence ratio was 2.8 (95% CI: 1.2; 6.7). The probability of reinfection was five times lower (95% CI: 2.9; 9.2) in HCWs vaccinated with the primary schedule if the first infection was acquired during Omicron dominance. The subjects vaccinated with the first booster had a density incidence of reinfection three times lower (95% CI: 1.9; 5.8) if the first infection was during Omicron. The incidence ratio in subgroups constructed based on characteristics such as gender, age group, job category, and department also registered significant differences in density incidence. The history of SARS-CoV-2 infection by variant is important when interpreting and understanding public health data and the results of studies related to vaccine efficacy for hybrid immunity subgroup populations. Full article

Wireless and power line communication hybrid relay technology can realize complementary advantages and comprehensively improve the communication coverage and performance of power Internet of Things. In order to study the mechanism of the physical layer and Media Access Control (MAC) layer algorithm that affects the performance of hybrid relay systems, the cross-layer performance modeling, optimization, and simulation analysis are carried out for the non-orthogonal multiple access (NOMA) technology. Firstly, a two-hop NOMA media access control protocol is designed based on the CSMA algorithm. Considering the effects of non-ideal channel transmission at the physical layer and competitive access at the MAC layer on the system performance, a cross-layer performance analysis model of hybrid wireless and power line communication relay system under NOMA is established. Finally, a cross-layer optimization model based on multi-objective programming is established for the hybrid relay system. By analyzing the relationship between transmitting power and performance index, the joint optimization of transmitting power and power distribution factor between users is realized. Simulation results verify the validity and reliability of the proposed cross-layer model. The results show that the hybrid relay algorithm combined with NOMA and CSMA can effectively improve the performance of the system throughput, packet loss probability, and delay. Full article

For processing streaming events from a Dynamic Vision Sensor camera, two types of neural networks can be considered. One are spiking neural networks, where simple spike-based computation is suitable for low-power consumption, but the discontinuity in spikes can make the training complicated in terms of hardware. The other one are digital Complementary Metal Oxide Semiconductor (CMOS)-based neural networks that can be trained directly using the normal backpropagation algorithm. However, the hardware and energy overhead can be significantly large, because all streaming events must be accumulated and converted into histogram data, which requires a large amount of memory such as SRAM. In this paper, to combine the spike-based operation with the normal backpropagation algorithm, memristor–CMOS hybrid circuits are proposed for implementing event-driven neural networks in hardware. The proposed hybrid circuits are composed of input neurons, synaptic crossbars, hidden/output neurons, and a neural network’s controller. Firstly, the input neurons perform preprocessing for the DVS camera’s events. The events are converted to histogram data using very simple memristor-based latches in the input neurons. After preprocessing the events, the converted histogram data are delivered to an ANN implemented using synaptic memristor crossbars. The memristor crossbars can perform low-power Multiply–Accumulate (MAC) calculations according to the memristor’s current–voltage relationship. The hidden and output neurons can convert the crossbar’s column currents to the output voltages according to the Rectified Linear Unit (ReLU) activation function. The neural network’s controller adjusts the MAC calculation frequency according to the workload of the event computation. Moreover, the controller can disable the MAC calculation clock automatically to minimize unnecessary power consumption. The proposed hybrid circuits have been verified by circuit simulation for several event-based datasets such as POKER-DVS and MNIST-DVS. The circuit simulation results indicate that the neural network’s performance proposed in this paper is degraded by as low as 0.5% while saving as much as 79% in power consumption for POKER-DVS. The recognition rate of the proposed scheme is lower by 0.75% compared to the conventional one, for the MNIST-DVS dataset. In spite of this little loss, the power consumption can be reduced by as much as 75% for the proposed scheme. Full article