Engineering Applications of Artificial Intelligence最新文献

{"title":"Elements discriminative non-negative matrix factorization for data clustering","authors":"Jie Li ,&nbsp;Xuzhu Shen ,&nbsp;Chaoqian Li ,&nbsp;Yaotang Li","doi":"10.1016/j.engappai.2025.111210","DOIUrl":"10.1016/j.engappai.2025.111210","url":null,"abstract":"<div><div>Semi-supervised non-negative matrix factorization (NMF) is widely used in data clustering because it can generate more discriminative representations for high-dimensional data by leveraging partial label information. To advance research in this field, we propose a novel method, Element Discriminative NMF (EDNMF), which incorporates discrimination constraints based on the element ratio and element difference of the new representations of labeled data points. EDNMF is implemented in two variants, each with an iterative algorithm for solving the optimization problem. We further analyze the computational complexity and convergence of these algorithms. A key advantage of EDNMF is that its learned representations can serve directly as a clustering assignment matrix, thereby simplifying the clustering process. Extensive experiments on eight real-world datasets demonstrate that EDNMF consistently outperforms baseline methods, confirming its effectiveness in improving clustering performance. The code is available at <span><span>https://github.com/ljisxz/EDNMF</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111210"},"PeriodicalIF":7.5,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144107981","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deep gate information bottleneck-based prediction model for complex disease-related micro-ribonucleic acids via heterogeneous biological networks","authors":"Yanbu Guo ,&nbsp;Yiyang Xin ,&nbsp;Jinde Cao ,&nbsp;Yaoli Xu ,&nbsp;Dongming Zhou","doi":"10.1016/j.engappai.2025.111222","DOIUrl":"10.1016/j.engappai.2025.111222","url":null,"abstract":"<div><div>Accurate prediction of potential association patterns is becoming a routine and essential method for detecting, analyzing, and controlling diseases. However, the complexity of biological networks poses significant challenges to computational methods between micro-ribonucleic acids and diseases. In this work, we propose a flexible multi-view gate information bottleneck-driven prediction model for complex disease-related micro-ribonucleic acid prediction. Our proposed model can reduce noise and redundant information from complex biological networks between different scale representations via the information bottleneck mechanism, and then enhances the robustness and generalization performance. Unlike other computational methods, we design the gate variational information bottleneck via a shrinking and an enlarging gate mechanism for multi-view different-order feature learning. The gate variational information bottleneck fuses the shared similarity and the view-specific embedding to obtain discriminative representation, and then eliminates redundant information and enhances task-relevant patterns. Next, the information bottleneck-based model is parameterized by a gate variational autoencoder and the reparameterization trick. Extensive experiments on different genomic datasets show the superior performance of our model compared to baselines, and the proposed model could effectively support the validation of complex disease-related micro-ribonucleic acids. It also shows that the model performance can be effectively improved by multi-view embedding learning and gate variational information bottleneck.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111222"},"PeriodicalIF":7.5,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144108143","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Predicting mechanical properties of magnetorheological elastomers during the manufacturing process using a new machine learning method","authors":"Qiyu Wang ,&nbsp;Lai Peng ,&nbsp;Yurui Shen ,&nbsp;Hua Dezheng ,&nbsp;Xinhua Liu ,&nbsp;Zhixiong Li ,&nbsp;Sumika Chauhan ,&nbsp;Govind Vashishtha","doi":"10.1016/j.engappai.2025.111160","DOIUrl":"10.1016/j.engappai.2025.111160","url":null,"abstract":"<div><div>Precisely predicting the shear storage modulus of magnetorheological elastomer (MRE) is crucial for effective vibration control. Traditional methods, however, are time-consuming and resource-intensive. This study introduces a novel hybrid deep learning model (RICBM) to efficiently predict this modulus. RICBM combines Random Forest (RF), an improved football team training algorithm, convolutional neural networks (CNN), bidirectional long short-term memory networks (BiLSTM), and multi-head attention (MAT) mechanisms. Initially, the significance of chosen input features is assessed through the RF. Subsequently, spatial pyramid matching (SPM) chaotic mapping and nonlinear weighting factors are employed to enhance the performance of the Football team training algorithm (FTTA). Next, a CNN-BiLSTM model is developed and the improved FTTA (IFTTA) is utilized to refine its parameters. Ultimately, the multi-head attention mechanism is utilized to highlight crucial input features, thereby further enhancing the predictive capabilities of the model. MRE samples with varied preparation parameters are prepared and tested by a rheometer in this study, resulting in a database of MRE shear storage modulus with 6 input and 1 output feature. The proposed model is applied to this database, alongside various comparative models. The experimental results demonstrate that RF effectively processes data and enhances prediction accuracy. The IFTTA enhances the CNN-BiLSTM model's predictive accuracy for the shear storage modulus of MRE. The resulting model shows significant effectiveness in making these predictions.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111160"},"PeriodicalIF":7.5,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144108148","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Fabrication and characterization of biological biosensors in sports injury treatment: High sensitivity of silver oxide using artificial neural network modeling","authors":"Youliang Wu ,&nbsp;B. Kamyab Moghadas ,&nbsp;Dongqiang Gu ,&nbsp;S. Salahshour ,&nbsp;M. Hashemi","doi":"10.1016/j.engappai.2025.111120","DOIUrl":"10.1016/j.engappai.2025.111120","url":null,"abstract":"<div><div>This study focuses on the potential of biosensor technology to revolutionize sports injury management. Conventional methods for sports injury management, such as physical examinations and imaging techniques, often lack the sensitivity and real-time monitoring capabilities required to track the healing process effectively. These methods are also invasive, relying on blood sampling, which can be uncomfortable for athletes. In contrast, the proposed wearable biosensor offers a non-invasive, painless alternative by measuring biomarkers like myoglobin and creatine kinase in sweat. This study introduces a novel graphene field-effect transistor biosensor integrated into a wristband, combined with an artificial neural network (ANN) model to predict material properties and optimize biomarker detection. The results show the potential of this technology to revolutionize sports injury management by providing real-time, accurate, and non-invasive monitoring of injury progression and recovery. The results indicate that changes in compressive strength and porosity have an impact on dissolution rate, pore size growth, and chemical stability. Lower compressive strength leads to an increase in dissolution rate, while higher compressive strength promotes pore size growth and chemical stability. The accuracy of the ANN model's predictions was evaluated using linear regression and demonstrated acceptable error levels compared to experimental testing. Among the nanocomposite hydrogel scaffolds containing silver oxide nanoparticles, a specific sample showed noteworthy characteristics, including a compressive strength of 2.4 Mega Pascal, 55 % porosity, 22 % dissolution rate, 27 % pore size growth, and 65 % chemical stability.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111120"},"PeriodicalIF":7.5,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144116203","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Hybrid transformer-convolution neural network architecture for emoji-aware hostile post classification with custom attention mechanisms","authors":"Santosh Rajak,&nbsp;Ujwala Baruah","doi":"10.1016/j.engappai.2025.111037","DOIUrl":"10.1016/j.engappai.2025.111037","url":null,"abstract":"<div><div>In the current digital atmosphere, social media platforms have achieved a pervasive attendance, facilitating unprecedented levels of connectivity while simultaneously nurturing environments conducive to hostility. Hindi is the most spoken language in the Indian subcontinent and the third most spoken worldwide. Twitter has 30.3 million active users in India, with many communicating in Hindi, highlighting the need for specialized content moderation tools beyond English. To address the need to identify hostile content in Hindi, we developed a dataset of 8300 Devanagari tweets, annotated into five classes, providing a nuanced analytical framework. We introduce a novel Scaled Multi-Head Attention-Convolution Neural Network Fusion Model (SMA-CFM) to confront the complexities of multiclass classification in this domain. The proposed model extracts contextual features from text and emoji embeddings, refining them via Scaled Multi-Head Attention. Convolution Neural Network (CNN) layers further capture local dependencies, ensuring classification across different hostility categories. Finally, the fused features are processed through a dense layer for final classification. Compared to Long Short-Term Memory (LSTM) and CNN, our fusion SMA-CFM approach consistently outperforms existing techniques by effectively capturing both textual and emoji semantics. This improvement highlights the importance of integrating attention mechanisms with convolutional networks for hostile post-classification. Our Cross-Lingual Language Model with RoBERTa (XLM-RoBERTa) embedding with emoji vector achieves superior performance, with F1-scores of 79.86% (Abusive), 78.58% (Defamation), 81.08% (Hate), and 76.40% (Non-Hostile). Global vectors for word representation(GloVe) with emoji vector performs best for Offensive (78.58%). The findings confirm our approach effectively improves Hindi hostile post classification.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111037"},"PeriodicalIF":7.5,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144116528","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Training-free multi-scale neural architecture search for high-incidence cancer prediction","authors":"Jie Zheng,&nbsp;Chunlin He,&nbsp;Wenxing Man,&nbsp;Jing Wang","doi":"10.1016/j.engappai.2025.111089","DOIUrl":"10.1016/j.engappai.2025.111089","url":null,"abstract":"<div><div>Deep neural networks excel in high-incidence cancer prediction; however, designing networks that predict specific cancers is time-consuming and requires expert. The neural architecture search method offers a way to automate network design and has shown success in natural image. However, the small and varying lesion sizes in cancer image pose challenges, and most neural architecture search methods are computationally expensive and exhibit low agent correlation. Therefore, we propose a training-free multi-scale neural architecture search method for high-incidence cancer prediction. We introduce a multi-scale search space to address varying lesion sizes; and identify optimal scale combinations for feature extraction. To reduce computational costs and improve agent correlation, we design a training-free agent that evaluates network performance based on convergence, expressiveness, trainability, and complexity, enabling efficient neural architecture search implementation. Our extensive experiments on the NAS-Bench-201, MedmnistV2, LC25000, BreakHis, and CRC-5000 datasets show that our method outperforms both manually designed networks and state-of-the-art neural architecture search methods. The results demonstrate average improvements of 4.2%, 1.88%, 79.45%, 34.31%, and 31.71% in accuracy, area under the curve, search time, and Kendall and Spearman correlation coefficients, respectively.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111089"},"PeriodicalIF":7.5,"publicationDate":"2025-05-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144108093","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Promoting sustainable urban mobility: An integrated fuzzy decision-making model for assessing autonomous bus alternatives","authors":"Ömer Faruk Görçün ,&nbsp;Mehmet Özçalıcı ,&nbsp;Hasan Emin Gurler ,&nbsp;Dragan Pamucar ,&nbsp;Vladimir Simic","doi":"10.1016/j.engappai.2025.111034","DOIUrl":"10.1016/j.engappai.2025.111034","url":null,"abstract":"<div><div>Today, in addition to the increasing pressures on urban transportation authorities to achieve sustainability goals, it has become necessary to comprehensively evaluate innovative transportation technologies such as autonomous (driverless) buses due to the increasing demand for public transportation applications that will contribute to making urban transportation more sustainable with its environmental, social, economic and social dimensions. In addition, the reservations and hesitations of decision-makers about integrating autonomous buses into urban transportation systems have not been eliminated. These hesitations and reservations are mainly due to critical research, theoretical gaps, and limitations in practice. Considering these gaps, this study presents an innovative model that integrates the fuzzy logarithm methodology of additive weights (LMAW) method and the fuzzy Dombi Bonferroni (DOBI) method to evaluate and rank 20 different autonomous bus alternatives with 33 sustainability criteria. The proposed integrated decision-making procedure can effectively manage complex uncertainties while examining whether autonomous bus alternatives can be integrated into urban transportation systems based on sustainability, considering four-dimensional sustainability criteria. This finding indicates that urban transportation's user-oriented and reliable nature is critical to achieving sustainability goals. In addition, the Proterra Catalyst (A9) is the autonomous bus with the highest sustainability performance for use in urban transport, followed by the Mercedes-Benz Future Bus (A18) and Mercedes-Benz eCitaro (A8). These results regarding alternatives underline the importance of advances in autonomous vehicle technology and making these vehicles more sustainable in evaluation processes.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111034"},"PeriodicalIF":7.5,"publicationDate":"2025-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144107910","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A quality index for construction big data in shield tunneling","authors":"Chao Zhang ,&nbsp;Yuhao Ren ,&nbsp;Qihang Huang ,&nbsp;Renpeng Chen","doi":"10.1016/j.engappai.2025.111023","DOIUrl":"10.1016/j.engappai.2025.111023","url":null,"abstract":"<div><div>The quality of the dataset underpinning the data-driven models predefines the upper limit for their performance yet lacks a quantitative way to be captured for the construction big data generated in earth pressure balance, i.e., EPB, shield tunneling. Herein, a quality index is proposed to fill this gap and formulated as an <span><math><msub><mrow><mi>L</mi></mrow><mrow><mn>2</mn></mrow></msub></math></span> norm of a vector composed of three components, i.e., accuracy, inclusiveness, and informativeness. The accuracy component is the ratio of non-outlier samples, i.e., a dataset containing fewer outliers shows a higher accuracy, reflecting the extent to which the dataset represents the real construction conditions during the tunneling. The inclusiveness component is the normalized envelope area of the dataset being mapped into a two-dimensional space, reflecting the range of diverse construction scenarios that have been included in the dataset. The informativeness component is the dimensionless uncertainty reduction of given data-driven models by the dataset, reflecting the contribution of datasets to the given model’s prediction. The proposed quality index is comprehensively assessed using a big database collected from multiple tunneling projects. A series of sub-datasets deliberately divided from the big database are utilized to train data-driven models by three commonly used algorithms, i.e., random forest, neural network, and K-nearest neighbors, for mapping three target functions widely concerned in tunneling, i.e., torque, thrust, and penetration. It is shown that the proposed quality index of the training data unfailingly excellently correlates with the performance of the data-driven models (R-values <span><math><mo>&gt;</mo></math></span> 0.91) regardless of algorithms, target functions, and sample sizes.The proposed quality index serves as a theoretical basis for a series of practical application scenarios, e.g., training data selection, and core dataset development. A practical application based on the Changsha project illustrated that the training dataset selected using the quality index can significantly boost the performance of the developed data-driven models by more than 38% and reduce training time by more than 26%.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111023"},"PeriodicalIF":7.5,"publicationDate":"2025-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144107980","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Two-stage vision system: Application of multi-perspective object detection network and character recognition network in industrial product classification","authors":"Shengjie Jin,&nbsp;Zhen Cao,&nbsp;Chaogang Yu","doi":"10.1016/j.engappai.2025.111190","DOIUrl":"10.1016/j.engappai.2025.111190","url":null,"abstract":"<div><div>In industrial automation, precise identification of incoming goods using vision systems significantly enhances production efficiency and reduces shipment errors. To address issues of size discrepancies in target detection and the need for further category code identification, a two-stage visual classification system has been introduced. The first stage presents a novel You Only Look Once-DualSight Fusion Network (YOLO-DSF) model that adopts multi-perspective backbones: one enhances fine-detail extraction via a lightweight space-to-depth downsampling module (LSPDown), and the other captures global features through a lightweight GhostNet structure. A newly developed FocusFusion module (FFM) merges the outputs from these perspectives, reinforcing small-target detection and maintaining the detection capabilities for multi-scale objects. In the detection head, a low-level feature layer is introduced for improved performance on tiny objects, and a lightweight design is applied to control computational overhead, thereby address staircase convergence issues while preserving efficiency. Compared to the baseline model, the YOLO-DSF achieves a Mean Average Precision at Intersection over Union of 0.50 ([email protected]) of 98.6 %, an improvement of 4.4 %. Moreover, tests on the Northeastern University Surface Defect (NEU-DET) public dataset have also demonstrated the superiority of this model. The second stage involves a U-shaped denoising network (UDNet) that acts as a preprocessor in character recognition, effectively reducing background noise and boosting character visibility, resulting in an accuracy improvement of 1.5 %. Field tests have demonstrated that the system achieves an object detection accuracy of 96.3 % and a goods classification accuracy of 98.9 %, thereby verifying its practicality and value in industrial applications.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111190"},"PeriodicalIF":7.5,"publicationDate":"2025-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144108066","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Adaptive lightweight temporal convolutional network with context-aware downsampling strategy for traffic flow prediction","authors":"Shuai Zhang ,&nbsp;Xiang Yin ,&nbsp;Wenyu Zhang ,&nbsp;Jiyuan Xu ,&nbsp;Xin Jing","doi":"10.1016/j.engappai.2025.111208","DOIUrl":"10.1016/j.engappai.2025.111208","url":null,"abstract":"<div><div>Accurate traffic flow prediction is crucial for intelligent transportation systems, which remains challenging due to intricate spatiotemporal features. Despite the promising performance of recent advances in spatiotemporal models, expensive computational costs make them difficult to be used with limited hardware resources. Moreover, existing models tend to cope with multiple temporal patterns of traffic flow in a coarse-grained manner, making it difficult to deeply extract temporal features. In this study, a novel deep learning model is proposed to address the aforementioned issues and achieve accurate traffic flow prediction. First, a new context-aware downsampling strategy is proposed for reducing the computational cost of the model, which provides contextual information for the traffic flow at each time step and then performs downsampling to reduce the sequence length, thus making the model more lightweight. Second, a new adaptive lightweight temporal convolutional module is proposed to extract temporal features deeply, which can adaptively update model parameters in a lightweight and fine-grained manner to deal with multiple temporal patterns of traffic flow. Third, the proposed model employs spatiotemporal embedding to efficiently learn the underlying spatiotemporal patterns of traffic flow. Extensive experiments on multiple real-world datasets validate the effectiveness and robustness of the proposed model.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"156 ","pages":"Article 111208"},"PeriodicalIF":7.5,"publicationDate":"2025-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144108146","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}