HighTech and Innovation Journal
Not a member yet
317 research outputs found
Sort by
Improved Skyline-BP Network for Multi-Track MIDI Music Melody Extraction and Style Classification
Full text linkWith the rapid development of the digital music industry, core challenges have emerged concerning the insufficient accuracy of main melody extraction and the poor style classification effect of multi-track MIDI files. To address these issues, this study proposes a novel model based on an improved Skyline algorithm and an optimized BP neural network. The method first standardizes MIDI data into a Time-Pitch-Intensity feature matrix. An improved Skyline algorithm is then used to integrate pitch saliency calculation with temporal continuity screening, enhancing the anti-interference ability for multi-track melodies. For music style classification, an optimized BP network with Adaptive Moment Estimation (Adam) gradient optimization and Residual Connection (ResConnect) is designed to improve learning efficiency and accuracy. Experimental results demonstrated that the proposed model surpassed comparative models in overall performance, with a classical-style main melody extraction accuracy of 94.6% and a 2-track separation accuracy of 95.2%. The experiments were benchmarked on the Lakh MIDI Dataset and MuseScore MIDI Library. The model also exhibits superior robustness against noise interference and faster convergence speed. This study provides reliable technical support for applications like music creation assistance and copyright retrieval
Fine-Tuned Attribute Weighted Naïve Bayes with Modified Partial Instances Reduction for Gaming Disorder Classification
Full text linkFine Tuning Attribute Weighted Naïve Bayes (FTAWNB) is a reliable modified Naïve Bayes model. Even though it is able to provide high accuracy on ordinal data, this model is sensitive to outliers. To improve the performance of FTAWNB, this research modified the Partial Instances Reduction (PIR) technique to make the FTAWNB more adaptive to outliers. Nevertheless, in contrast to the original PIR technique, which substitutes missing values for data values deemed outliers, the PIR technique suggested in this study replaces data values deemed outliers using a Naïve Bayes weighting approach. The attribute values from the outlier data are replaced with the highest probability values for the attributes in the actual class. This PIR technique is referred to as modified PIR. The FTAWNB model with modified PIR has been evaluated using the Gaming Disorder dataset. Replacing the four attributes with the least amount of information resulted in accuracy gains of 99.74%, an increase of 1.53% over the FTAWNB model. The experimental result shows that adding the modified PIR technique to the FTAWNB model can handle the outlier in the data, proving it by increasing the performance in terms of accuracy, precision, and recall without pruning the dataset used.
Doi: 10.28991/HIJ-2025-06-01-05
Full Text: PD
Assessing AI-Driven Personalization in Smart Cities Using Hybrid Machine Learning and MCDM Approach
Full text linkThis study aims to assess AI-driven personalization strategies in smart cities, focusing on promoting digital inclusion across diverse urban populations. As artificial intelligence becomes increasingly central to urban service delivery, ensuring equitable and effective personalization is critical to preventing the amplification of digital inequality. To address this challenge, a hybrid evaluation framework is proposed, integrating Multi-Criteria Decision Making (MCDM) techniques, specifically Step-wise Weight Assessment Ratio Analysis (SWARA), Linguistic q-Rung Orthopair Fuzzy Numbers (Lq-ROFNs), and the Multi-Attributive Border Approximation Area Comparison (MABAC) with a Machine Learning (ML) classification model based on Random Forest. The framework is applied to stakeholder input from ten Indonesian smart cities, evaluating personalization readiness across five dimensions: accessibility, affordability, user engagement, privacy, and personalization effectiveness. The results indicate that accessibility and user engagement are the most influential criteria, while affordability and privacy are areas requiring strategic policy focus. The integrated model classifies cities by readiness level and identifies sensitivity patterns relevant to inclusive digital policy-making. The novelty of this research lies in its synthesis of MCDM and ML approaches to produce a transparent, scalable, and data-driven tool for evaluating AI personalization. This contributes to inclusive smart city development by aligning AI implementation with broader social equity objectives
Efficient Object Detection with an Optimized YOLOv8x Model
Full text linkThis study developed an efficient object detection model for indoor environments, addressing common challenges such as occlusions, varying lighting, and cluttered scenes. We evaluated several YOLOv8 variants—ranging from nano to extra-large—and introduced an optimized YOLOv8x model. Our approach combines structured pruning, quantization-aware training, and advanced data preprocessing techniques, including augmentation and noise reduction, to improve model performance while reducing computational demands. The models were developed and evaluated using a carefully selected indoor object detection dataset featuring ten common categories. Performance was measured through key metrics like precision, recall, and mean average precision (mAP). Among them, the fine-tuned YOLOv8x clearly outshined the baseline models, reaching a training precision of 0.577, a recall of 0.572, and an [email protected] of 0.537. When tested on new data, it demonstrated even better generalization, delivering a precision of 0.502, a recall of 0.528, and an [email protected] of 0.480—proving robustness and reliability in real-world scenarios. These results demonstrate that pruning and quantization can significantly reduce model complexity without sacrificing accuracy, which helps to detect indoor objects. In essence, it is optimized for indoor object detection, offering promising applications in smart environments, surveillance, and robotics
The Influence of Social Media Influencer Attributes on Brand Equity and Purchase Intention
Full text linkThis study examines how the attributes of social media influencers (SMIs) collectively shape brand equity and purchase intention in the digital marketplace. Building on classical endorsement theories—namely the Source Attractiveness Model, Source Credibility Model, Product Match-Up Hypothesis, and Meaning Transfer Theory—this research develops an integrated framework to explain how influencer traits translate into brand-related outcomes. Data were collected from 200 active social media users in Thailand and analyzed using structural equation modeling (SEM) to test the proposed causal relationships. The results reveal that source attractiveness, product match-up, and meaning transfer significantly enhance brand equity, while source credibility demonstrates a marginal yet positive effect. Furthermore, brand equity strongly predicts purchase intention and mediates the effects of influencer attributes. Theoretically, this study extends endorsement research by integrating four fragmented models into a unified influencer-based framework, advancing understanding of how digital influencers shape consumer-based brand equity. Managerially, the findings guide marketers in selecting influencers whose image, credibility, and symbolic meanings align strategically with brand identity to maximize consumer engagement and behavioral intent
Multi-Time Scale Coordinated Optimization of Energy Systems Under Flexible Load Response
Full text linkSince the development of society, people's demand for and use of energy have become increasingly diverse rather than remaining monotonous. Flexible load response serves as the core medium for integrating various energy sources. However, the operational performance of units within the energy system has not been ideal, and operating costs remain difficult to control. To address these challenges, this study investigates multi-time scale collaborative optimization of energy systems based on flexible load response, utilizing a combination of qualitative and quantitative methods. The research encompasses optimization architecture, optimization models, computational case studies, and validation. The results indicate that, during load response experiments, implementing an intra-day coordinated plan—specifically by further reducing thermal and electrical loads during peak hours—can significantly decrease the peak-valley difference. Additionally, in the cost comparison analysis, the operating cost was reduced by 1.47%, thereby addressing the shortcomings of traditional energy system coordination and optimization. Overall, the approach offers notable improvements both in economic performance and in system coordination and optimization, demonstrating considerable foresight
Integrating Social Cognitive Neuroscience into Digital Learning: Strengthening Pre-Service Teachers' Learning Design Competencies
Full text linkThis study investigates the development of learning design competencies among pre-service teachers (PSTs) through an integrated framework combining instructional design, social cognitive neuroscience (SCN), and digital learning innovations (DLIs). While SCN is often associated with neuroscience, this study applies SCN principles to educational contexts, focusing on cognitive and social processes that influence teaching and learning. Using a mixed-methods quasi-experimental design, the framework was validated by experts and implemented with 60 PSTs in experimental and control groups over 12 weeks. The experimental group engaged with SCN-informed DLIs, including virtual classroom simulations, adaptive feedback systems, and reflective learning tools, while the control group followed demonstration-based instruction. Findings revealed significant improvements in the experimental group's competencies, particularly in reflection (Cohen's d = 2.48) and implementation (Cohen's d = 2.27). The completion rate of virtual modules reached 92.5%, with 85% of sessions incorporating interactive digital tools. These results highlight the effectiveness of integrating SCN-informed DLIs for fostering adaptive, reflective, and innovative teaching skills. The framework bridges theoretical insights with practical applications, providing a scalable model for enhancing digital learning design competencies in teacher education. Doi: 10.28991/HIJ-2025-06-01-021 Full Text: PD
Local Economic Autonomy and Enterprises’ Green Total Factor Productivity: A Policy Substitution Perspective
Full text linkGlobal climate change and environmental degradation necessitate a transition toward sustainable economic development. The factors influencing green total factor productivity (GTFP), a crucial measure for sustainable economic growth, have garnered significant attention. This study investigates the impact of local economic autonomy on enterprises’ GTFP in China, integrating both fiscal and environmental autonomy. Using panel data from 2008 to 2021, a generalized difference-in-differences (DID) model combined with the non-radial SBM-ML index measures GTFP. Findings indicate that while fiscal autonomy promotes GTFP, environmental autonomy hinders it, resulting in an overall negative effect of economic autonomy. A policy substitution effect emerges; wherein local governments prioritize environmental regulation over support for science and technology. Additionally, industrial structure upgrading plays a role in mitigating the negative impact of autonomy, offering empirical evidence relevant to sustainable development policies in transition economies
Application of Deep Learning for Stock Prediction Within the Framework of Portfolio Optimization in Quantitative Trading
Full text linkThis paper proposes a method for stock prediction and portfolio optimization as a part of quantitative trading based on a combination of Bi-RNN and a modified snake optimization algorithm (MSOA) to build optimal portfolios and outperform conventional models and benchmarks. Methods/Analysis: We employ the Bi-RNN model, which processes historical stock data in both forward and backward directions to unveil intricate temporal dependencies. MSOA is used to fine-tune the hyperparameters of the Bi-RNN with enhancements such as Latin Hypercube Sampling for initialization, dynamic temperature adjustment, adaptive learning rates, and hybrid exploration-exploitation mechanisms. The Markowitz mean-variance approach is used to optimize the portfolio from asset allocations that the MSOA then improves. The model is evaluated on the S&P 500 from 1993 to 2020. Results: Such findings in experiments indicate that the proposed model outperforms baseline models, e.g., LSTM, GRU, and HMM, with lower Mean Squared Percentage Error (MSPE) values and higher Sharpe ratios of constructed portfolios. For instance, Portfolio 3 produced a 10.9% expected return with a standard deviation of 12.9%, delivering risk-adjusted returns that exceed those of the S&P 500. Novelty/Improvement: A strong integrated approach of deep learning and advanced optimization techniques is proposed for stock prediction and portfolio optimization, which achieves notable improvements in terms of accuracy and efficiency. The proposed approach overcomes the drawbacks of traditional algorithms, making it a valuable tool for financial decision-making
Digital Transformation in Higher Education: Enhancing Support Services Through Mobile Apps
Full text linkThe fourth industrial revolution has considerably enhanced technology, resulting in disruptive developments across various industries, including higher education. Generation Z, known as digital natives, has specific digital preferences, making mobile applications essential for improving their educational experience. The present study aims to investigate the digital transition in higher education, emphasizing using mobile applications as campus support services for Generation Z students. The study investigates the factors influencing mobile app acceptance and usage, intending to enhance educational support services' efficiency and quality among 100 students from different universities. By performing multiple linear regression, the study revealed that perceived usefulness is the most critical factor driving students' intention to use mobile apps in higher education. In contrast, other elements such as ease of use, competence, accessibility, and data privacy were not deemed significant concerns by the students. The findings are intended to advise higher education institutions on integrating mobile apps to assist Generation Z better, eventually leading to increased student engagement and satisfaction. This study emphasizes the significance of mobile technology in current educational contexts and offers practical insights for universities looking to use digital technologies to improve campus support services. However, the outcome may vary for students from different demographic and socio-economic backgrounds