Bài báo quốc tế
Lĩnh vực nổi bật
Trí tuệ nhân tạo
1.069
Internet vạn vật
23
An toàn bảo mật
11
Blockchain
3
Tìm kiếm theo
Năm xuất bản
Lĩnh vực bài báo
Kho tri thức
/
Bài báo quốc tế
Sắp xếp theo
Liên quan
CONDITIONAL MALWARE IMAGE GENERATION USING WGAN-GP FOR DATA AUGMENTATION
Nguyễn Hoa CươngThe efficacy of vision-based deep learning models in malware classification is frequently hindered by data
scarcity and severe class imbalance. To address this critical challenge, this paper proposes the
implementation of a Conditional Wasserstein Generative Adversarial Network with Gradient Penalty
(CWGAN-GP) to synthesize high-fidelity, two-dimensional malware representations. We conditioned the
network on five distinct classes: benign, spyware, trojan, virus, and worm. Designed to generate 224x224
RGB images, the model was trained on a structured subset of malware datasets. Empirical results over 60
training epochs demonstrate highly stable convergence and the effective elimination of mode collapse, a
common flaw in standard GANs. Following the robust training phase, the model successfully generated a
completely balanced dataset comprising 10,000 synthetic images (2,000 samples per class). This reliable
data augmentation strategy provides a vital foundation for mitigating class imbalance, thereby improving
the predictive accuracy and generalization capability of downstream deep learning-based malware detection
systems.
Năm:2026
|
Chủ đề: Khoa học kỹ thuật và công nghệ khác
|
Nhà xuất bản: Journal of Theoretical and Applied Information Technology
Design of Compact and High-Efficiency Microwave Rectifier with Triple-band Operation for Wireless Power Transfer
Đặng Hoài BắcIn this paper, a novel design for triple-band microwave rectifier is proposed with compact size and high power conversion efficiency, where step-lines were employed to suppress triple-band second harmonics of for efficiency improvement, and triple-band operation was obtained by applying a cross-shape fundamental matching network for the first time. The notable point in this proposal compared to the published research is that the structures are optimized when the harmonics suppression network uses only two transmission lines connected in series, while the triple-band fundamental matching network only requires four transmission lines to realize. Furthermore, the parameters of these transmission lines are determined specifically by the closed-form equations built on transmission line theory. This proposed methodology was verified by a fabricated rectifier based on a voltage doubler configuration using Schottky diode BAT15-03W with the size of 9 cm2. The measurement results show that the peaks of power conversion efficiency are 81.48%, 79.4%, and 73.3% at 0.91 GHz, 2.41 GHz, and 5.61 GHz, respectively. Compared to other publications about triple-band microwave rectifiers, the rectifier designed with the proposed methodologies has superior power conversion efficiency and electrical size.
Năm:2025
|
Chủ đề: Kỹ thuật điện tử và viễn thông
|
Nhà xuất bản: IEEE Access
MIST: A Multilingual Dataset and Benchmark for Fine-Grained Audio Inpainting Tampering Localization
Vũ Sơn TùngPartial speech inpainting, replacing a few words within a genuine utterance via voice cloning to alter its meaning, is an emerging audio deepfake threat. Unlike fully synthesized speech, the manipulated content constitutes only 2–7% of the utterance, and existing benchmarks are limited to single-region tampering with utterance-level labels. To enable systematic study of this problem, we introduce MIST, a large-scale multilingual dataset spanning six languages with 1–3 independently inpainted word-level segments per utterance, generated through LLM-guided semantic replacement and neural voice cloning, totaling approximately 497k tampered utterances with precise temporal annotations. To establish a benchmark for this task, we further introduce ISA, a backbone-agnostic coarse-to-fine localization framework that recovers all tampered regions without prior knowledge of their count; and SF1@τ, a segment-level F1 metric based on temporal IoU matching that jointly evaluates region counting and boundary precision. Zero-shot experiments confirm that utterance-level classifiers trained on fully synthesized speech fail on MIST (SF1@0.5 ≤ 1.7% across all backbones), while fine-tuning on MIST annotations dramatically improves performance: the Wav2Vec2-AASIST backbone improves from 1.2% to 29.2%, and the strongest backbone (WavLM-AASIST) reaches SF1@0.5 = 33.5%, demonstrating that the primary bottleneck is training data rather than inference methodology. The dataset, code, and evaluation toolkit are publicly available.
Năm:2026
|
Chủ đề: Kỹ thuật thông tin
|
Nhà xuất bản: IEEE Access
Comparative Study of Nickel Foam and Carbon Paper Substrates on the Electrochemical Performance of CoFe2O4 Electrodes
Nguyễn Thị Thuý HằngThe choice of current collector substrate plays a critical role in determining the electrochemical performance of supercapacitor electrodes. In this study, cobalt ferrite (CoFe2O4) nanoparticles were synthesized via a solution combustion method and employed as active materials for supercapacitor electrodes. Two types of conductive substrates—nickel foam and carbon paper—were systematically investigated to evaluate their influence on the electrochemical behavior of CoFe2O4-based electrodes. Electrodes were fabricated using identical slurry composition and deposition protocols to ensure fair comparison. Electrochemical measurements including cyclic voltammetry (CV), galvanostatic charge–discharge (GCD), and electrochemical impedance spectroscopy (EIS) were conducted in a three-electrode configuration. The results reveal that electrodes supported on nickel foam exhibited approximately 20% higher specific capacitance compared to those using carbon paper, along with improved charge transfer kinetics and reduced internal resistance. This enhancement is attributed to the three-dimensional porous
structure and superior electrical conductivity of the nickel foam, which facilitate efficient electron transport and electrolyte accessibility. These findings highlight the importance of substrate selection in optimizing the performance of transition metal oxide-based supercapacitor electrodes and provide guidance for the rational design of advanced energy storage systems.
Năm:2026
|
Chủ đề: Trí tuệ nhân tạo
|
Nhà xuất bản:
NF-DCL: Enhancing video anomaly detection with synthetic normal features and Debiased Contrastive Learning
Nguyễn Thu NgaVideo anomaly detection (VAD) is a challenging task, as it involves detecting anomalies at the frame level in a video stream with video-level supervision. The majority of existing weakly-supervised VAD methodologies, which employ Multiple-Instance Learning (MIL) along with a top-
ranking loss, often exhibit high false alarm rates for normal frames within abnormal videos. This pitfall comes from two main factors: the limited discriminatory capacity of the top-
ranking loss and the inefficiency of MIL in aggregating information from normal video segments. In contrast, semi-supervised VAD approaches that rely solely on normal video data during training often enhance the prediction of normal frames within abnormal videos, but at the expense of accuracy in detecting abnormal frames. To enhance the advantages of both approaches, we propose a two-stage method, namely Normalizing Flow-Debiased Contrastive Learning (NF-DCL). In the first stage, a Normalizing Flow model is employed to learn the distribution of normal feature representations and generate diverse synthetic normal samples. In the second stage, these synthesized features guide a Debiased Contrastive Loss integrated with MIL and top-
ranking loss, enhancing feature discrimination while alleviating label ambiguity and biased learning. Extensive experiments on UCF-Crime, ShanghaiTech, and XD-Violence demonstrate that NF-DCL consistently improves performance and significantly reduces false alarms, outperforming existing state-of-the-art methods.
Năm:2026
|
Chủ đề: Kỹ thuật thông tin
|
Nhà xuất bản: Computer Vision and Image Understanding
Synthesis, structure, and spectroscopic properties of Dy3+-doped ZnSe quantum dots: Judd–Ofelt analysis, energy transfer, and thermal stability
Nguyễn Xuân CaDy³⁺-doped ZnSe quantum dots (QDs) were successfully synthesized via a high-temperature wetchemical method using 1-octadecene as the reaction medium. Structural, morphological, and optical studies have been used to investigate optical properties and interactions between the substrate and the dopant. X-ray diffraction confirmed the formation of single-phase cubic ZnSe with a slight shift of diffraction peaks toward lower angles, indicating lattice expansion due to the substitution of Zn²⁺ by larger Dy³⁺ ions. TEM revealed nearly spherical QDs with particle sizes of 5–7 nm. Photoluminescence spectra exhibit characteristic Dy³⁺ emission bands corresponding to the ⁴F₉/₂ → ⁶HJ (J = 15/2, 13/2, 11/2,
9/2) transitions, with dominant yellow emission at ~582 nm. The emission intensity increases with Dy³⁺ concentration and reaches a maximum at 2 mol% before decreasing due to concentration quenching. Spectral overlap between the ZnSe host emission and Dy³⁺ excitation bands confirms an efficient hostsensitized energy transfer mechanism. Judd–Ofelt (JO) theory was applied to determine the optical intensity parameters (Ω₂, Ω₄, Ω₆) and evaluate the local environment around Dy³⁺ ions in the ZnSe crystal lattice. The obtained trend (Ω₂ > Ω₆ > Ω₄) indicates a moderately asymmetric coordination around Dy³⁺ ions. In addition, the temperature-dependent photoluminescence properties of ZnSe:Dy³⁺ QDs were investigated in the temperature range of 200–350 K, revealing good spectral stability with an activation energy of ~42.6 meV. The lifetime decreases as Dy³⁺ concentration increases according to the Inokuti–Hirayama law, suggesting that dipole-dipole interaction is the primary energy transfer mechanism. These results provide new insights into the spectroscopy and energy transfer of Dy³⁺ ions
in ZnSe QDs.
Năm:2026
|
Chủ đề: Vật lý
|
Nhà xuất bản: Ceramics Internationalf
Social networking sites engagement and purchase intention toward foreign brands: The mediating roles of multicultural experience and foreign brand love
Đặng Văn ThạcVirtual interactions and engagement on social networking sites (SNSs) enable consumers to experience multi- culturalism and interact with foreign brands. This study examines a unique model grounded in social exchange theory and emotion theory to demonstrate how SNSs engagement influences purchase intention toward foreign brands. Utilizing data from 693 Facebook users in an emerging market, partial least squares structural equation modeling (PLS-SEM) reveals that SNSs engagement enhances purchase intention toward foreign brands, with multicultural experience and foreign brand love mediating this relationship. This study contributes novel insights into the mechanisms underlying the link between consumers' SNSs engagement and their intention to purchase foreign brands. The findings encourage foreign brands and social media managers to strengthen their SNS presence to drive consumer emotions and purchases globally.
Năm:2026
|
Chủ đề: Trí tuệ nhân tạo
|
Nhà xuất bản: Acta Psychologica
Study on the structural, optical and magnetic properties of cobalt-doped CdSSe quantum dots
Nguyễn Thị Thu HoànCo-doped CdSSe alloy quantum dots (QDs) were successfully synthesized via a wet chemical hot-injection method, with Co²⁺ doping concentrations ranging from 1 to 10% at. In addition to the intrinsic band gap tunability of CdSSe QDs, Co incorporation introduces magnetic functionality, enabling the development of diluted magnetic semiconductor nanostructures for optoelectronic and spintronic applications. The effects of sulfur/selenium ratio and cobalt doping on the structural, optical, photoluminescence, and magnetic properties of CdSSe QDs were systematically investigated. X-ray diffraction results confirm that both undoped and Co-doped CdSSe QDs crystallize in the cubic zinc-blende structure, with no secondary phases detected. The lattice constant decreases with increasing Co concentration due to the substitution of smaller Co²⁺ ions for Cd²⁺ ions in the host lattice. Optical absorption and photoluminescence measurements reveal that the emission wavelength of CdSₓSe1-x QDs can be effectively tuned across the visible region by adjusting the S/Se ratio. Upon Co²⁺ doping, a pronounced blue shift of both absorption and photoluminescence peaks is observed, accompanied by an increase in band gap energy, indicating strong modification of the electronic structure induced by Co-related energy levels. This behavior is attributed to the substitution of Cd²⁺ by smaller Co²⁺ ions, which induces compressive lattice strain and shifts the conduction band edge to higher energy. Time-resolved photoluminescence analysis shows a decrease in carrier lifetime with increasing Co concentration, attributed to enhanced non-radiative recombination via Co²⁺-induced trap states. Magnetic measurements demonstrate that Co-doped CdSSe QDs exhibit weak room-temperature ferromagnetism coexisting with diamagnetic behavior, with saturation magnetization increasing up to 5% Co doping and decreasing at higher concentrations due to the onset of antiferromagnetic interactions. These results demonstrate that Co doping is an effective method for simultaneously tuning the optical and magnetic properties of CdSSe QDs, making them promising candidates for optoelectronic and spintronic applications.
Năm:2026
|
Chủ đề: Vật lý
|
Nhà xuất bản: RSC Advances
Multi-objective Optimization of the 7075 Aluminum Alloy Milling Process: A Comparative Study of MOPSO, NSGA-II, SPEA2, and MOACO
Hoàng SơnThis study investigated the influence of machining parameters and performed a comparative multi-objective optimization of surface roughness (Ra) and Material Removal Rate (MRR) in the milling of 7075 aluminum alloy using Multi-Objective Particle Swarm Optimization (MOPSO), Non-dominated Sorting Genetic Algorithm II (NSGA-II), Strength Pareto Evolutionary Algorithm 2 (SPEA2), and Multi-Objective Ant Colony Optimization (MOACO). Three machining parameters, including spindle speed (S), feed rate (f), and depth of cut (d), were considered. Analysis of Variance (ANOVA) results showed that S and f significantly affect Ra (p < 0.05), contributing 20.06% and 79.68%, respectively, while f and d are the most significant factors influencing MRR (p < 0.05), accounting for 45.01% and 40.11% of the total contribution. A predictive model for Ra developed using the Group Method of Data Handling (GMDH) demonstrated high predictive performance, with R² values of 0.9990 and 0.9962 for the training and validation datasets, respectively. Comparative analysis indicated that NSGA-II produced the most stable solutions, whereas SPEA2 and MOACO exhibited less balanced performance, and MOPSO achieved rapid convergence with relatively dispersed solutions. Experimental validation of Ra and analytical verification of MRR confirmed the reliability of the proposed framework, with mean deviations of 6.5% and 0.37%, respectively. Unlike prior investigations that examined individual algorithms or lacked integrated experimental assessment, this study presents a systematic cross-algorithm evaluation under identical machining conditions. The proposed framework integrates statistical contribution analysis, predictive modeling, and experimental validation, thereby establishing a robust and practically applicable approach for multi-objective milling optimization.
Năm:2026
|
Chủ đề: Khoa học kỹ thuật và công nghệ khác
|
Nhà xuất bản: International Journal of Mechanical Engineering and Robotics Research
Adaptive Cloud–Edge Coordination for Real-Time Phishing URL Detection With Distributed Caching and ONNX-Based Inference
Đàm Minh LịnhPhishing detection systems continue to struggle with real-time responsiveness in distributed
web environments. This paper proposes a cloud–edge coordination framework that integrates browserside Uniform Resource Locator (URL) interception, adaptive cache coordination (ACC), and open
ONNX-accelerated inference. The lightweight hybrid model (DistilBERT–MLP fusion) fuses DistilBERT
embeddings with standardized numerical features through a multilayer perceptron (MLP). Experiments
demonstrate a detection accuracy above 99%, millisecond-level neural inference latency, and end-to-end
detection times below 150 ms in live browser environments. ACC reduces cold-cache initialization from
multisecond delays to microsecond-level cache lookups. Before ACC, a first-time blacklist fetch takes
7–8 s, and subsequent lookups complete in under 1 µs, with total detection latency ranging between
approximately 2 and 110 ms. This behavior demonstrates the effectiveness of ACC in reducing repeated
network queries. The experimental results substantiate three principal contributions: an efficient hybrid
lexical–numeric detector, an ACC-driven mechanism that mitigates cold-start delays, and a cloud–edge
design that strengthens real-time phishing protection at the web scale.
Năm:2026
|
Chủ đề: Khoa học máy tính và thông tin
|
Nhà xuất bản: IEEE Access
On performance evaluation of NOMA-aided SIMO multi-hop schemes using energy harvesting and fountain coding based information accumulation
Ngô Hoàng ẤnOn performance evaluation of NOMA-aided SIMO multi-hop schemes using energy harvesting and fountain coding based information accumulation
Năm:2026
|
Chủ đề: Kỹ thuật điện tử và viễn thông
|
Nhà xuất bản: Plos One
Enhancing the Secrecy Performance and Secure Energy Efficiency of NOMA Systems Using RIS With Artificial Jamming
Võ Tá TýEnhancing the Secrecy Performance and Secure Energy Efficiency of NOMA Systems Using RIS With Artificial Jamming
Năm:2026
|
Chủ đề: Kỹ thuật điện tử và viễn thông
|
Nhà xuất bản: International Journal of Communication Systems
Secrecy performance analysis and optimization in untrusted UAV-aided PDNOMA communication with energy harvesting and hardware impairments
Lê Thị AnhSecrecy performance analysis and optimization in untrusted UAV-aided PDNOMA communication with energy harvesting and hardware
impairments
Năm:2026
|
Chủ đề: Kỹ thuật điện tử và viễn thông
|
Nhà xuất bản: Physical Communication
Large SiO2 Sizes Influencing Light Scattering Properties in Original White LED Structure
Pham Hong CongPreviously, SiO2 particles (1-10 wt%, <10 𝜇m) were proposed to augment the light flux and correlated color temperature (CCT) deviation in conventional blue-excited LEDs. This research paper explores the application of larger SiO2 particle sizes within a conventional LED model. SiO2 particles with diameters ranging from 1 𝜇m to 20 𝜇m were analyzed for their scattering properties usingtheStobermethod,Mie-scattering-basedMATLABprogramandLightToolssoftwaretosimulateopticalattributesofaphosphor
transmuted WLED apparatus with XRD, SEM, and so on. SiO2 was integrated into the yellow phosphor YAG:Ce layer to induce
scattering phenomena within the active layer. Through our investigation, the concentrations of SiO2 and YAG:Ce phosphor were
consistently maintained at 5 and 10 wt.%, respectively. Our findings focus on assessing backward scattering with larger SiO2 sizes.
Modulating the sizes of SiO2 spheres reduced CCT deviation and enhanced both flux output and color rendering performance. These
results emphasize the potential utility of larger SiO2 particles in enhancing white LED performance and facilitate further investigation
to optimize their useability in production processes
Năm:2026
|
Chủ đề: Kỹ thuật điện
|
Nhà xuất bản: Science and Technology Indonesia
A novel entropy Autoencoder-Synchronized Hashing Semi-supervised network for robust Android malware identification
Nguyễn Huy TrungAndroid malware is growing rapidly, and modern variants increasingly use obfuscation and code-disrupting techniques that evade traditional detectors. These transformations can hide or alter malicious characteristics, making accurate identification difficult. To address this, we propose the Entropy Autoencoder-Synchronized Hashing Semi-supervised Network (EASH-SemiNet), a novel framework integrating semi-supervised learning, an entropy-based autoencoder, a synchronized hashing mechanism, and hash matching. This combination provides robust and adaptive malware detection while significantly reducing reliance on labeled malicious samples. Unlike traditional entropy-based methods, which often suffer from high false-positive rates, EASH-SemiNet leverages synchronized hashing and semi-supervised learning to achieve superior detection accuracy while minimizing reliance on labeled malware data. Our approach successfully detects malware variants, obfuscation, and code-altering tactics using entropy-based features and the synchronized hashing mechanism. Furthermore, the integrated hash-matching strategy efficiently reduces the computational burden imposed by known threats. Thus, EASH-SemiNet offers an effective, efficient, and adaptable solution to the challenges posed by evolving Android malware and limited labeled data.
Năm:2026
|
Chủ đề: Khoa học máy tính và thông tin
|
Nhà xuất bản: Array
Uplink Cooperative NOMA with Downlink Energy Harvesting in PPP-Based Ultra-Dense Networks
Sinh Cong LamThis paper investigates an uplink cooperative Non-Orthogonal Multiple Access (NOMA) system integrated with energy harvesting in ultra-dense wireless networks, where both base stations (BSs) and users are spatially modeled as independent Poisson Point Processes (PPPs). A distance-dependent cooperative transmission scheme is proposed, allowing a near user to assist a far user only when their separation is below a predefined threshold, thus improving cooperation efficiency and reducing unnecessary relay operations. The near user allocates a fraction ϵ of its harvested energy to relay the far user's signal, while the remaining (1−ϵ) portion is used for its own uplink transmission. Using the stretched path loss model and Campbell's theorem, a closed-form expression for the average harvested energy is derived, and system capacity is evaluated through Monte Carlo simulations. Results show that the proposed energy-harvesting cooperative NOMA (EH-C-NOMA) scheme significantly improves network performance, achieving up to 85% higher capacity than the non-cooperative case and 11.63% improvement over full cooperation.
Năm:2026
|
Chủ đề: Trí tuệ nhân tạo
|
Nhà xuất bản:
Comparison of Multi-Expert Ensemble Feature Selection with XAI Explainability for IoT Intrusion Detection on Edge Computing
Le Thi Trang LinhThis paper presents a comprehensive evaluation of four ensemble methods (Majority Voting Method, AdaBoost Algorithm, Stacking Ensemble Method, and Bayesian Decision Method) in multi-expert feature selection systems, constructed from five base experts including SelectKBest, Recursive Feature Elimination, Random Forest, L1-based Selection (Lasso), and Mutual Information, in the context of intrusion detection for Internet of Things (IoT) systems. We utilize the CIC IoT 2023 dataset, a large-scale real-world dataset that reflects diverse attack scenarios in IoT environments. The highlight of this research is the integration of Explainable Artificial Intelligence (XAI) methods to analyze the influence of each expert and feature on the ensemble method results. Additionally, the study conducts multi-criteria comparisons including performance metrics (accuracy, per-class precision, recall, F1-score), computational efficiency (training time, response time, memory usage), and comprehensive classification performance indicators (ROC-AUC, PR-AUC, false positive rate). The experimental results provide in-depth analyses of the advantages and limitations of each method, thereby offering recommendations for selecting the most appropriate ensemble method according to specific IoT system requirements. This research contributes to enhancing the effectiveness and transparency of intrusion detection systems in complex IoT environments.
Năm:2026
|
Chủ đề: Trí tuệ nhân tạo
|
Nhà xuất bản:
Beyond Connectivity: Semantic Communication Unlocks the Hyper-Connected City with 6G
Woongsoo NaThe dawn of 6G technology presents a transformative opportunity to realize the hyper-connected city, characterized by seamless information exchange and intelligent infrastructure. However, achieving this vision requires more than simply increasing bandwidth and connectivity; it demands a shift towards semantic communication, a paradigm that prioritizes the meaning of information. This paper argues that semantic communication, powered by Artificial Intelligence (AI), is an indispensable cornerstone for building truly intelligent and responsive hyper-connected cities. We explore the core framework of semantic communication, examine the specific AI technologies that enable its implementation, and analyze experimental results demonstrating significant improvements in bandwidth efficiency, data recovery, and reduced latency compared to legacy methods. These results, derived from a testbed implementation of a semantic adaptive context lighting system, highlight the potential of AI-powered semantic communication to optimize resource allocation, enhance system resilience, and pave the way for a future where technology seamlessly enhances urban living. Although limited, our testbed results offer a compelling case for semantic communication's future relevance and potential as an enabler for hyper-connected city of the 6G era. Key areas for future research, including scalable AI models, security and privacy protocols, standardized interoperability, and robust real-world deployment and evaluation, are also discussed
Năm:2026
|
Chủ đề: Kỹ thuật điện tử và viễn thông
|
Nhà xuất bản: IEEE Communications Magazine
Implementation of Compressed Sensing Method for Foot Pressure Reconstruction Based on AI
Viet Nguyen VanThe quality of life is significantly affected by abnormalities in gait brought on by diseases like flat feet, Parkinson's disease, or stroke. This work presents the development of a plantar pressure monitoring system that utilizes gait analysis to aid in clinical evaluation and rehabilitation. The technology records pressure data in real-time at key plantar areas using eight Force Sensitive Resistor (FSR) sensors inserted in a mat. Reconstructing high-resolution plantar pressure maps from a small number of sensor inputs is made possible by an innovative use of compressed sensing (CS). For accurate signal reconstruction, the system employs a K-SVD-based dictionary learning framework combined with Orthogonal Matching Pursuit (OMP), where the pressure map is modeled as a sparse signal in a learned representation domain. Experimental results using a public dataset and real measurements demonstrate that the reconstructed images closely match ground truth data, with a Pearson correlation of 95.65%. This proves the feasibility of reconstructing detailed pressure distributions from sparse input data.
Năm:2026
|
Chủ đề: Trí tuệ nhân tạo
|
Nhà xuất bản:
Boosting Business Success: The Impact of MetaversePayments on Firm Performance
Nguyễn Thành LuânPayment services in the virtual world have become an essential part of economic transactions at a time when the metaverse is significantly changing digital interactions. This study presents the Metaverse Technology Acceptance Model (MTAM), an expanded version of the conventional Technology Acceptance Model (TAM). Its purpose is to investigate the factors that impact the performance of companies in the metaverse payment industry. Data from 366 working professionals with experience in the metaverse was collected using a purposive convenience sampling technique. The results of a thorough analysis of the route and the use of structural equation modeling (SEM) indicate that business performance is significantly affected by metaverse perceived usefulness (MPU), metaverse trust (MET), attitude towards usage (ATU), technology readiness (TR), and environmental sustainability practices (ESP). The research provides a good understanding of the intricate impacts of the metaverse payment service area. This scholarly research enhances the theoretical framework of metaverse payment and provides strategic insights for firms aiming to thrive in this competitive and expanding field. The statement emphasizes the significance of employing user-centric models to enhance corporate performance and negotiate the complexities of financial operations in the metaverse.
Năm:2026
|
Chủ đề: Trí tuệ nhân tạo
|
Nhà xuất bản: Journal of Entrepreneurship and Innovation in Emerging Economies
RESEARCH ON THE IMPACT OF FINANCIAL AND HUMAN CAPITAL USE EFFICIENCY ON FINANCIAL PERFORMANCE AT RETAIL ENTERPRISES LISTED ON THE VIETNAMESE STOCK EXCHANGE
Khuong Huyen DucThis study examines the effects of capital employed efficiency (CEE) and human capital efficiency (HCE), two key components of intellectual capital operationalized through Pulic’s Value Added Intellectual Coefficient (VAIC) framework, on the financial perfor-
mance of retail firms listed on the Vietnamese stock exchange, while controlling for firm size and the type of financial statements (separate versus consolidated). Drawing on an unbalanced panel of 34 listed retailers from 2016 to 2023, we estimate pooled OLS,
fixed- and random-effects models, and generalized least squares to improve robustness under heteroskedasticity and autocorrelation. Financial performance is measured by return on assets (ROA), return on equity (ROE), and Tobin's Q. Results show that both
CEE and HCE are positively and statistically significantly related to ROA, ROE, and Tobin's Q, implying that efficient use of invested capital and productive utilization of human resources jointly raise profitability and market valuation. Firm size is negatively
associated with ROA but positively associated with Tobin's Q, suggesting that scale can bring coordination costs that dampen operating efficiency while strengthening growth expectations and valuation. Moreover, performance patterns vary by reporting structure: firms issuing consolidated statements display systematically different ROA and ROE outcomes relative to firms reporting only separate statements, highlighting the importance of financial statement type when interpreting VAIC-based indicators. Over-
all, the study provides retail-sector evidence on intellectual capital efficiency in Vietnam's emerging market and offers implications for managers, investors, and policy makers. Managers should balance capital allocation and working-capital discipline with
investments in employee capability, incentives, and retention; investors can use CEE and HCE as complementary signals of value creation; and policymakers can support transparency and human-capital development to strengthen sector competitiveness. By
focusing on listed retailers, the study enriches VAIC evidence for transitional economies and underscores that both tangible and human resource efficiency are strategic levers in retail competition
Năm:2026
|
Chủ đề: Trí tuệ nhân tạo
|
Nhà xuất bản: FINANCIAL AND CREDIT ACTIVITY: PROBLEMS OF THEORY AND PRACTICE
DistilBERT for Efficient and Accurate Email Phishing Detection: A Benchmark Against Machine and Deep Learning Models
Đàm Minh LịnhEmail phishing remains a persistent cybersecurity threat that exploits human vulnerabilities,
often evading technical safeguards. While machine learning (ML) and deep learning (DL)
have been widely applied for phishing detection, systematic benchmarks comparing
lightweight transformer models with traditional approaches remain limited. This study
addresses this gap by evaluating six models—Naïve Bayes, Random Forest, XGBoost,
LSTM, BiLSTM, and a fine-tuned DistilBERT—on a real-world dataset of 17,538 emails
using three train-test splits (60:40, 70:30, 80:20). DistilBERT consistently outperforms all
baselines across all splits. Under the 80:20 split, it achieves the highest accuracy (98.77%),
precision (99.10%), recall (98.97%), F1-score (99.02%), and AUC (99.91%). Remarkably,
it maintains low computational overhead with a training time of 342 seconds, demonstrating
an optimal trade-off between detection accuracy and efficiency. In contrast, BiLSTM, the
best-performing recurrent model, reaches 97.43% accuracy but produces more false
negatives—a more critical security risk than false positives in phishing detection.
Additional experiments reveal that DistilBERT maintains stable performance across
different data splits, with AUC values consistently above 0.998. The confusion matrix
analysis shows that DistilBERT misclassifies only 25 legitimate emails as phishing (false
positives) and misses only 23 phishing emails (false negatives), significantly outperforming
all baseline models. These findings demonstrate that lightweight transformer models like
DistilBERT offer a practical, scalable, and cost-effective solution for real-time phishing
email detection, effectively bridging the gap between high accuracy and production-ready
deployability.
Năm:2026
|
Chủ đề: Khoa học máy tính và thông tin
|
Nhà xuất bản: Ingénierie des Systèmes d’Information
Integrating CSR and ESG into Sustainable Development Strategies: Practical Solutions for Businesses in the Era of Climate Change
Vũ Thị Hạnh TâmIn the context of escalating climate change, integrating Corporate Social Responsibility (CSR) and Environmental, Social, and Governance (ESG) criteria into sustainable development strategies has become decisive for business survival and growth. Despite strengthened awareness of their importance, many companies, particularly in emerging economies like Vietnam, still struggle with effectively integrating these elements into business operations. This study aims to evaluate the impact of six key factors on CSR and ESG integration effectiveness while proposing practical solutions for businesses in the era of climate change. Using a mixed-methods approach with a sample of 270 enterprises in Vietnam, results show that leadership commitment (β = 0.398) and climate change impacts (β = 0.254) are the two most influential factors, followed by regulatory framework, stakeholder pressure, customer awareness, and financial resources. The regression model explains 64.5% of the variance in CSR and ESG integration effectiveness. The results of this study will serve as a scientific basis for developing a supportive policy framework and sustainable business strategy, affirming that in the context of climate change, integrating CSR and ESG is not only an ethical responsibility but also a smart business strategy to enhance resilience while creating long-term value.
Năm:2026
|
Chủ đề: Trí tuệ nhân tạo
|
Nhà xuất bản: Proceedings of the 8th International Conference on Corporate Social Responsibility and Sustainable Development, Springer Proceedings in Business and Economics
VIVID: A Culturally Grounded Benchmark Exposing the Figurative Language Gap in Vietnamese NLP
Đỗ Trần TúWe present VIVID (Vietnamese Idioms for Validation and Interpretation Depth), the first systematic benchmark for evaluating culturally grounded figurative language understanding in Vietnamese. VIVID comprises 1,636 idioms and proverbs annotated with five complexity traits (literal expressions, pragmatic nuances, Sino-Vietnamese terms, uncommon vocabulary, folk knowledge) and seven semantic themes. We establish an evaluation framework com- bining generative and discriminative tasks, proposing an LLM-as-a-Judge approach with aspect-based prompting validated against human judgment (Cohen’s κ = 0.792). Evaluating eight state-of-the-art models reveals critical gaps: Vietnamese-specialized models drastically underperform multilingual systems (VinaLLaMA-7B: 0.13 vs. GPT-4o: 2.46), and even top models achieve less than 50% correctness on average. Notably, few-shot prompting does not universally improve performance, with GPT-4o exhibiting degradation due to stylistic overfitting. Our analysis exposes systematic failures including literal over-interpretation, lexical gaps, and pragmatic flattening, demonstrating that current models lack cultural competence for nuanced figurative interpretation. VIVID provides an essential tool for advancing figurative language understanding in culturally rich contexts. We release codes and datasets at https://github.com/ReML-AI/VIVID.
Năm:
|
Chủ đề: Trí tuệ nhân tạo
|
Nhà xuất bản: