Engineering Applications of Artificial Intelligence最新文献

{"title":"Reducing the estimation bias and variance in reinforcement learning via Maxmean and Aitken value iteration","authors":"Fanghui Huang ,&nbsp;Wenqi Han ,&nbsp;Xiang Li ,&nbsp;Xinyang Deng ,&nbsp;Wen Jiang","doi":"10.1016/j.engappai.2025.112502","DOIUrl":"10.1016/j.engappai.2025.112502","url":null,"abstract":"<div><div>The value-based reinforcement leaning methods suffer from overestimation bias, because of the existence of max operator, resulting in suboptimal policies. Meanwhile, variance in value estimation will cause the instability of networks. Many algorithms have been presented to solve the mentioned, but these lack the theoretical analysis about the degree of estimation bias, and the trade-off between the estimation bias and variance. Motivated by the above, in this paper, we propose a novel method based on Maxmean and Aitken value iteration, named MMAVI. The Maxmean operation allows the average of multiple state–action values (Q values) to be used as the estimated target value to mitigate the bias and variance. The Aitken value iteration is used to update Q values and improve the convergence rate. Based on the proposed method, combined with Q-learning and deep Q-network, we design two novel algorithms to adapt to different environments. To understand the effect of MMAVI, we analyze it both theoretically and empirically. In theory, we derive the closed-form expressions of reducing bias and variance, and prove that the convergence rate of our proposed method is faster than the traditional methods with Bellman equation. In addition, the convergence of our algorithms is proved in a tabular setting. Finally, we demonstrate that our proposed algorithms outperform the state-of-the-art algorithms in several environments.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"162 ","pages":"Article 112502"},"PeriodicalIF":8.0,"publicationDate":"2025-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222673","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":"Reinforcement learning-based trajectory tracking optimal control for underactuated unmanned surface vehicles under asymmetric input saturation","authors":"Ziping Wei,&nbsp;Jialu Du","doi":"10.1016/j.engappai.2025.112307","DOIUrl":"10.1016/j.engappai.2025.112307","url":null,"abstract":"<div><div>For underactuated unmanned surface vehicles (USVs) under asymmetric input saturation caused by thrust-limit characteristics, as well as unknown dynamics and ocean environmental disturbances, a trajectory tracking optimal control (TTOC) scheme is proposed using the reinforcement learning (RL) method. Through coordinate transformations and mathematical derivation, an underactuated USV motion model is transformed into the standard affine nonlinear form. To address the asymmetric input saturation of underactuated USVs, a new inverse hyperbolic tangent-type penalty function is designed for control inputs, relaxing the assumption of input saturation limits being symmetric. Based on RL methods and adaptive neural networks (NNs), an actor-critic NN framework is developed, with weight update laws designed for NNs. This framework learns the TTOC law for underactuated USVs through the online interaction of actor and critic NNs while adapting to unknown dynamics and disturbances. In particular, a robustifying term is designed and added to the output of an actor NN to compensate for the adverse effects of a lumped residual term, which enhances the robustness of the TTOC law and thereby achieves asymptotic regulation of trajectory tracking errors. Theoretical analyses and simulation results indicate that the proposed TTOC scheme enables underactuated USVs to asymptotically track the desired trajectory.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"162 ","pages":"Article 112307"},"PeriodicalIF":8.0,"publicationDate":"2025-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222668","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 multiple convolution and bilayer acceleration model for precise and efficient early urban fire detection in complex scenarios","authors":"Pei Shi ,&nbsp;Jun Lu ,&nbsp;Yachen Xu ,&nbsp;Quan Wang ,&nbsp;Yonghong Zhang ,&nbsp;Liang Kuang ,&nbsp;Deji Chen ,&nbsp;Guangyan Huang","doi":"10.1016/j.engappai.2025.112555","DOIUrl":"10.1016/j.engappai.2025.112555","url":null,"abstract":"<div><div>AI advancement enables earlier and more effective urban fire detection, crucial for slowing fire spread. However, hardware limitations make precise and efficient detection under limited resources a major challenge. Moreover, earlier detection of fire requires the identification of smoke, which further exacerbates the difficulty of detecting algorithms since smoke's inherent low-contrast visual properties produce feature blurring from the surrounding background. In this paper, we propose a novel multiple convolutions and bilayer accelerate (MCBA) model for effective early urban fire detection in terms of precision, lightweight and efficiency, which takes advantage of the mainstream You Only Look Once version 8 (YOLOv8) to training and testing the early fire detection model. In our MCBA model, three optimization techniques have been developed to balance lightweight and precision. First, it designs a new multi-convolution (MC) structure to reduce the size of the original backbone network by avoiding complex or skipping connections. Second, the model includes a novel design of a bilayer accelerate mechanism (BAM) at the neck to minimize the interference of redundant background information in multiple scenarios. Third, we provide a precision compensation strategy (PCS) at the neck to enhance the feature extraction and aggregation capabilities, enabling effective detection of small fire areas. The experiments demonstrate that our proposed MCBA model achieves higher performance in terms of precision and efficiency compared with 17 counterpart detection models. It exhibits superior performance with minimal parameter count and the lowest computational complexity among the compared methods. The model shows strong potential for deployment in early urban fire detection across a variety of real-world scenarios.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"162 ","pages":"Article 112555"},"PeriodicalIF":8.0,"publicationDate":"2025-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222674","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":"Analysing sustainable industrial wastewater treatment technologies using circular Fermatean fuzzy multi-attribute group decision making with decision experts’ confidence levels","authors":"Prayosi Chatterjee,&nbsp;Mijanur Rahaman Seikh","doi":"10.1016/j.engappai.2025.112549","DOIUrl":"10.1016/j.engappai.2025.112549","url":null,"abstract":"<div><div>The evaluation of sustainable industrial wastewater treatment techniques is vital for preserving environmental integrity and protecting public health. Industrial processes often generate wastewater containing toxic compounds like metal contaminants, toxic chemicals, and complex organic compounds, posing serious risks to ecosystems and human well-being. This study proposes a robust multi-attribute group decision-making framework to assess five treatment alternatives across twelve sub-criteria. The evaluation model employs circular Fermatean fuzzy numbers to capture uncertainty and imprecision in expert judgements. To enhance the accuracy of data aggregation, four novel Schweizer–Sklar weighted aggregation operators are introduced, integrating varying confidence levels. Criteria weights are determined through a hybrid approach combining the subjective Ranking Comparison (RANCOM) method and the objective Opinion Weight Criteria Method (OWCM), ensuring balanced prioritization. Alternatives are ranked using the Alternative Ranking Order Method Accounting for Two-Step Normalization (AROMAN), a novel technique for improved discrimination and consistency. Results reveal that the membrane bioreactor as the most sustainable treatment with score 0.821, outperforming activated sludge process, by 25.34%. The lowest-ranked option is chemical coagulation and flocculation, scoring 0.622. Sensitivity analysis, performed by varying three parameters, shows reasonable stability with an average correlation value of 0.71. Comparative analysis shows an average Spearman’s rank correlation of 0.86, confirming reliability. The study recommends prioritizing membrane bioreactor adoption in industrial treatment plants to enhance efficiency and water reuse. By promoting effective treatment solutions, the study contributes to reducing industrial pollution, enhancing water reuse, and advancing environmental sustainability.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"162 ","pages":"Article 112549"},"PeriodicalIF":8.0,"publicationDate":"2025-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222684","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":"Efficient active flow control strategy for confined square cylinder wake using deep learning-based surrogate model and reinforcement learning","authors":"Meng Zhang,&nbsp;Mustafa Z. Yousif,&nbsp;Minze Xu,&nbsp;Haifeng Zhou,&nbsp;Linqi Yu,&nbsp;Hee-Chang Lim","doi":"10.1016/j.engappai.2025.112468","DOIUrl":"10.1016/j.engappai.2025.112468","url":null,"abstract":"<div><div>This study introduces a deep learning surrogate model-based reinforcement learning (DL–MBRL) for active control of two-dimensional (2D) wake flow past a square cylinder confined between parallel walls using antiphase jets. In the training of this framework, a proximal policy optimisation (PPO) reinforcement learning agent alternates its interaction between a deep learning-based surrogate model (DL–SM) and a computational fluid dynamics (CFD) simulation to suppress wake vortex shedding, thereby significantly reducing computational costs. The DL–SM, built with a Transformer for temporal dynamics and a multiscale enhanced super-resolution generative adversarial network (MS–ESRGAN) for spatial reconstruction, is trained on 2D direct numerical simulation wake flow data to effectively and accurately emulate complex nonlinear flow behaviours. Compared to standard model-free reinforcement learning, the DL–MBRL approach reduces training time by about 50% while maintaining or improving wake stabilisation. Specifically, it achieves approximately a 98% reduction in shedding energy and a 95% reduction in the standard deviation of the lift coefficient, demonstrating strong suppression of vortex shedding. By leveraging the inherent stochasticity of DL–SM, DL–MBRL also addresses the nonzero mean lift coefficient issue observed in model-free methods, promoting more robust exploration. These results highlight the potential of the framework for extension to practical and industrial flow control problems.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"162 ","pages":"Article 112468"},"PeriodicalIF":8.0,"publicationDate":"2025-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222669","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":"LIP-MC: Multi-Constraint Label Independent Prediction in label distribution learning","authors":"Gui-Lin Li ,&nbsp;Ruili Wu ,&nbsp;Xiaorui Qian ,&nbsp;Qiang Zhu ,&nbsp;Heng-Ru Zhang","doi":"10.1016/j.engappai.2025.112493","DOIUrl":"10.1016/j.engappai.2025.112493","url":null,"abstract":"<div><div>Label distribution learning can resolve label ambiguity precisely by determining how well each label describes an instance. Traditional label distribution learning algorithms frequently attempt to model the complex relationships between all labels. This not only adds complexity to the model but may also reduce prediction accuracy due to label conflicts. In this paper, we propose a novel label distribution learning algorithm based on Multi-Constraint Label Independent Prediction (LIP-MC), intended to promote the rationality and accuracy of prediction results by simplifying the prediction process and combining multiple constraints. Specifically, label independent prediction values are generated for each label through sparsity constraints and weight coefficient matrices. Subsequently, a novel transformation model is designed to combine all separate label predictions and produce the final label distribution. Furthermore, smoothness constraints and logarithmic similarity constraints were introduced to enhance the model’s performance and generalization ability. On fourteen real datasets, the experiment was carried out, and the comparison results against seven advanced algorithms under seven evaluation metrics confirmed that the proposed algorithm is superior.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"162 ","pages":"Article 112493"},"PeriodicalIF":8.0,"publicationDate":"2025-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222670","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":"Enhanced targeted attacks on Graph Neural Networks via Average Gradient and Perturbation Optimization","authors":"Yang Chen ,&nbsp;Bin Zhou ,&nbsp;Haixing Zhao ,&nbsp;Padarti Vijaya Kumar","doi":"10.1016/j.engappai.2025.112530","DOIUrl":"10.1016/j.engappai.2025.112530","url":null,"abstract":"<div><div>Graph Neural Networks (GNNs) are vulnerable to adversarial attacks that cause performance degradation by adding small perturbations to the graph. Gradient-based attacks are among the most widely used methods and have demonstrated strong performance across various attack scenarios. However, most gradient attacks use greedy strategies to generate perturbations, which tend to fall into local optima, leading to underperformance of the attack. To address the above problem, we propose an attack (Average Gradient and Perturbation Optimization Attack, AGPOA) on GNNs, which consists of an average gradient calculation and a perturbation optimization module. In the average gradient calculation module, we compute the average of the gradient information over all moments to guide the attack to generate perturbed edges, which stabilizes the direction of the attack update and gets rid of undesirable local maxima. We use a perturbation optimization module to limit the attack budget and further improve performance. Furthermore, we demonstrate the theoretical superiority of AGPOA over traditional gradient-based attack methods through attack loss variance. The experimental results show that AGPOA improves the misclassification rate by 2%–8% compared to other state-of-the-art models in the node classification task.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"162 ","pages":"Article 112530"},"PeriodicalIF":8.0,"publicationDate":"2025-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222815","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 lightweight self-attention metric network for bird species recognition in intelligent bird repellent equipment","authors":"Jiangjian Xie ,&nbsp;Shanshan Xie ,&nbsp;Baican Li ,&nbsp;Yujie Zhong ,&nbsp;Chunhe Hu ,&nbsp;Junguo Zhang ,&nbsp;Björn W. Schuller","doi":"10.1016/j.engappai.2025.112546","DOIUrl":"10.1016/j.engappai.2025.112546","url":null,"abstract":"<div><div>Bird damages to power transmission lines pose significant operational risks, and intelligent bird repellent equipment (IBRE) requires accurate species recognition for effective long-term repellent. We propose a novel lightweight self-attention metric network (LSAM-Net) for few-shot bird species recognition in the vicinity of power transmission lines, aiming to enhance the performance of IBRE. LSAM-Net integrates a simple attention mechanism (SimAM) to emphasize critical spatial and channel features, thereby enhancing the extraction of key semantic information from bird images. Additionally, a self-correlation representation (SCR) module is employed to capture local structural patterns, effectively mitigating the impact of pseudo-features and improving the network’s capacity to learn discriminative representations. To promote the utilization of local discriminative information in few-shot classification, LSAM-Net leverages earth mover’s distance (EMD) to compute structural similarity between images. For efficient deployment, we apply knowledge distillation to further reduce model complexity. Extensive experiments conducted on Bird-65, CUB200, 2011, miniImageNet, and Fewshot-CIFAR100 demonstrate that LSAM-Net achieves superior performance compared to state-of-the-art methods, while maintaining a compact architecture. On the Bird-65 and CUB200-2011 datasets, LSAM-Net requires only 4.75 and 1.18 giga floating-point operations (GFLOPs), and achieves inference speed improvements of 52.9 % and 48.9 %, respectively, over the self-attention metric network (SAM-Net). Further optimization with TensorRT yields additional reductions in inference time by 43.6 ms and 53.7 ms, respectively. These improvements significantly support species-specific repellent strategies, thereby enhancing the long-term effectiveness of IBRE systems.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"162 ","pages":"Article 112546"},"PeriodicalIF":8.0,"publicationDate":"2025-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222814","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":"Physics-driven prior learning-based deep unrolling for underwater image enhancement","authors":"Thuy Thi Pham ,&nbsp;Hansung Yu ,&nbsp;Truong Thanh Nhat Mai ,&nbsp;Chul Lee","doi":"10.1016/j.engappai.2025.112472","DOIUrl":"10.1016/j.engappai.2025.112472","url":null,"abstract":"<div><div>We propose a physics-driven prior learning-based algorithm unrolling approach for underwater image enhancement that leverages the advantages of both model- and learning-based approaches while overcoming their limitations. Model-based algorithms are theoretically robust because of prior knowledge of the underlying physics but may degrade image quality due to modeling inaccuracies. On the other hand, learning-based algorithms exhibit better adaptivity but inferior interpretability due to their black-box models and neglect of domain knowledge. In this work, we first formulate underwater image enhancement as a joint optimization problem with physics-based underwater-related priors and two learnable regularizers to compensate for modeling inaccuracies. Then, we solve the problem by reformulating it as a set of subproblems, which are then solved iteratively. Finally, we unroll the iterative algorithm into a deep neural network comprising a series of blocks, in which the optimization variables and regularizers are updated using closed-form solutions and learned deep neural networks, respectively. Experimental results on several datasets demonstrate that the proposed algorithm outperforms state-of-the-art underwater image enhancement algorithms on both quantitative and qualitative comparisons. The source code and pretrained models will be available at <span><span>https://github.com/thithuypham/BLUE-Net</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"162 ","pages":"Article 112472"},"PeriodicalIF":8.0,"publicationDate":"2025-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145204200","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":"Three-way dynamic clustering algorithms based on generalized neighborhood relations in incomplete hybrid information systems with applications in medical decision-making","authors":"Haoran Sun ,&nbsp;Bingzhen Sun ,&nbsp;Xixuan Zhao ,&nbsp;Qiang Bao ,&nbsp;Xiaoli Chu","doi":"10.1016/j.engappai.2025.112508","DOIUrl":"10.1016/j.engappai.2025.112508","url":null,"abstract":"<div><div>In addressing clinical challenges related to chronic diseases, the effective use of medical data in decision-making is often hindered by issues such as incompleteness, heterogeneity, and the need for continuous updates. To cope with these challenges, this study introduces a three-way dynamic clustering strategy built upon generalized neighborhood relations, aiming to enhance clustering robustness, strengthen the model’s ability to manage uncertainty, and support adaptability to dynamically evolving data. First, generalized neighborhood relations are constructed in incomplete hybrid information systems. An evaluation function is defined from two perspectives: the number of similar attributes between objects and the distance between objects, thereby optimizing similarity measurement and accurately characterizing the data structure. Second, three-way decision rules are introduced to effectively handle uncertainty in objects while maintaining classification accuracy, thereby improving the interpretability and adaptability of the clustering model. Furthermore, to accommodate the dynamic nature of medical data, a dynamic incremental clustering method based on neighborhood information is proposed to ensure that newly added patient data can be efficiently integrated into existing clusters, enhancing model real-time performance and computational efficiency. Experiments conducted on real clinical data from Chronic kidney disease (CKD) patients validate the proposed method. The results demonstrate that, compared to existing clustering algorithms, the proposed method outperforms in terms of F1-score and Rand Index evaluation metrics. It also exhibits higher applicability in patient classification, core and boundary domain partitioning, and dynamic data processing, providing effective support for precision stratified management of chronic disease patients and intelligent medical decision-making.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"162 ","pages":"Article 112508"},"PeriodicalIF":8.0,"publicationDate":"2025-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145204202","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}