Applied Soft Computing最新文献

{"title":"Spectrum sparsity-driven learnable wavelet packet transform for denoising and its application in fault diagnosis","authors":"Xin Kang,&nbsp;Jianxin Cui,&nbsp;Junsheng Cheng,&nbsp;Yu Yang,&nbsp;Qiyao Liu","doi":"10.1016/j.asoc.2026.114713","DOIUrl":"10.1016/j.asoc.2026.114713","url":null,"abstract":"<div><div>Domain-specific noise is a key cause of cross-domain data distribution shifts and a major obstacle in cross-domain fault diagnosis. Most existing denoising methods are built on the assumptions that noise is weak and Gaussian, which often do not hold in the industrial context, thus limiting their effectiveness. To address this, we propose a spectrum sparsity-driven learnable wavelet packet transform (SSLWPT) model for unsupervised, small-sample mechanical signal denoising and feature enhancement. First, based on the vibration properties of rotating machinery, where stationary and cyclostationary components carry critical health-related information, we propose a spectrum sparsity-driven mechanism to actively capture these components for adaptive feature enhancement. Next, an autoencoder mimicking wavelet packet decomposition and reconstruction is proposed, with learnable wavelet bases and denoising parameters. Under the joint constraints of reconstruction loss, parameter regularization, and spectrum sparsity loss, the model achieves effective signal denoising. We further integrate the proposed SSLWPT as a preprocessing tool into a two-stage framework for cross-domain fault diagnosis. Both theoretical analysis and experimental results validate the superiority of the proposed model in denoising. Moreover, cross-domain diagnosis experiments conducted on bearing, gear and motor datasets confirm that our method achieves significantly better generalization than related denoising methods and SOTA domain generalization methods.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"191 ","pages":"Article 114713"},"PeriodicalIF":6.6,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080107","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"Shopping recommendation algorithm based on knowledge graph and dynamic association graph model","authors":"Li Gao,&nbsp;Shenglong Yang,&nbsp;Dunlu Peng,&nbsp;Hongjun Li","doi":"10.1016/j.asoc.2026.114636","DOIUrl":"10.1016/j.asoc.2026.114636","url":null,"abstract":"<div><div>With the maturation of deep learning and big data technologies, the application of recommendation algorithms can help e-commerce platforms acquire new users and enhance conversion rates. This paper proposes a novel e-commerce recommendation algorithm, which is based on a combination model of knowledge graphs and dynamic association graphs and is designed in two stages, In the offline stage: Firstly, a knowledge graph attention network is built based on the features of product attributes and consumers’ historical shopping behavior, thereby obtaining key feature attributes of the products. Secondly, a dynamic association graph network is built by connecting the consumer behavior sequence with a dynamic graph structure, and key features of the selection behavior are obtained successively through a Graph Convolutional Neural Network (GCN), a Gated Recurrent Unit (GRU), and an attention mechanism. Thirdly, the output features are fused to generate a recommendation list, and the algorithm records the consumer’s shopping path, updates the database and dynamic association graph, and can automatically adjust according to the shopping path. Experiments on three Amazon datasets demonstrate that the algorithm exhibits superior recommendation performance and robustness in actual shopping scenarios with complex association features.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"191 ","pages":"Article 114636"},"PeriodicalIF":6.6,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080108","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"Green AI driven quantile boosting ensembles for sustainable real estate valuation","authors":"Muhammet Ali Kadıoğlu ,&nbsp;Mehmet Yavuz Yağcı","doi":"10.1016/j.asoc.2026.114616","DOIUrl":"10.1016/j.asoc.2026.114616","url":null,"abstract":"<div><div>In today’s volatile economic environment, the accurate estimation of apartment prices is crucial for maintaining financial predictability. Nevertheless, the development of high-performing artificial intelligence (AI) models often entails substantial computational costs, which, in turn, raise critical concerns regarding environmental sustainability, economic viability, and equitable accessibility. This study addresses these challenges by proposing a comprehensive, energy-efficient framework for real estate price estimation grounded in the principles of Green AI. Utilizing a dataset of over 13 million geolocated listings in Turkey (January 2018–September 2024), we implement an end-to-end machine learning pipeline, including data preparation, outlier detection, quantile-based ensemble modeling, hyperparameter tuning using Optuna, and model evaluation. The performance of models—LightGBM, Gradient Boosting Regressor, XGBoost, and Random Forest—is assessed through a dual lens of estimation accuracy and environmental efficiency, incorporating metrics such as training time, energy consumption, and prediction latency. Results show that LightGBM offers the best balance between accuracy and resource efficiency, making it the recommended baseline for large-scale or real-time valuation systems, while GBR and XGBoost may be used in R&amp;D settings under energy constraints. The paper also highlights the importance of integrating dynamic market indicators and climate risk variables to enhance model robustness and sustainability. Overall, this research advances a holistic framework for responsible and scalable real estate valuation, contributing to the growing field of sustainable AI deployment in high-impact sectors.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"191 ","pages":"Article 114616"},"PeriodicalIF":6.6,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080297","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"Dual-path prompt-enhanced semi-supervised segmentation for retinal vessel quantification in OCTA imaging","authors":"Yuemei Luo ,&nbsp;Yuan Li ,&nbsp;Lei Tao ,&nbsp;Jun Xu ,&nbsp;Linbo Liu","doi":"10.1016/j.asoc.2026.114778","DOIUrl":"10.1016/j.asoc.2026.114778","url":null,"abstract":"<div><div>Retinal vascular analysis plays a vital role in the early detection and management of ocular diseases, which remain a significant global health challenge. Optical Coherence Tomography Angiography (OCTA) provides non-invasive, dye-free, high-resolution visualization of retinal microvasculature, yet accurate segmentation of vascular structures—particularly large vessels—remains difficult due to noise, depth variability, and the scarcity of annotated data. To address the need for efficient vessel segmentation under limited annotation, this paper proposes a semi-supervised learning framework based on a dual-path Segment Anything Model (SAM), tailored for large vessel segmentation in OCTA images. The framework incorporates the following innovations: (1) Dual-decoder collaborative training, which integrates prompt-based and prompt-free segmentation to improve generalization from minimal labeled data; (2) Cross-prompt strategy, which establishes bidirectional guidance between dual decoder branches using pseudo-prompt points derived from unlabeled data; and (3) Prompt consistency regularization, which enforces output stability under varying spatial prompts, mitigating the sensitivity of segmentation results to prompt locations. A hybrid point sampling scheme—combining center and random points—further improves prompt reliability. Evaluated on the OCTA-500 dataset using only 5% labeled samples, the proposed framework achieves Dice scores of 0.9134 and 0.8924 at 3 mm and 6 mm scan depths, respectively, surpassing existing semi-supervised methods. These results demonstrate the method’s effectiveness, robustness, and potential for scalable, annotation-efficient retinal image analysis in clinical practice.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"192 ","pages":"Article 114778"},"PeriodicalIF":6.6,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146191886","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 direction-aware ant colony optimization for smooth mobile robot path planning in complex environments","authors":"Zimin Xu ,&nbsp;Yanying Chen ,&nbsp;Lin Lin ,&nbsp;Jinyan Huang ,&nbsp;Guoli Wang","doi":"10.1016/j.asoc.2026.114782","DOIUrl":"10.1016/j.asoc.2026.114782","url":null,"abstract":"<div><div>Global path planning is a core problem in mobile robotics, aiming to generate collision-free trajectories that are not only short but also smooth and suitable for real-world deployment. Although Ant Colony Optimization (ACO) has been widely applied due to its population-based global search capability, its standard formulation still suffers from inherent limitations in complex environments. Heuristic information solely based on distance to the goal, together with uniformly initialized pheromone trails, provides weak global directional guidance in the early search stage, leading to largely unguided exploration. Moreover, the conventional state transition rule does not explicitly penalize sharp direction changes, often producing zigzag paths with excessive turning, while homogeneous pheromone updating slows the emergence of high-quality solutions and delays convergence. To address these issues, this paper proposes a direction-aware ant colony optimization (DAACO) framework, which integrates four complementary mechanisms in a unified manner. Specifically, an enhanced heuristic function combines distance information with global and local directional consistency to strengthen goal-oriented guidance. A directional pheromone initialization mechanism embeds estimated path cost and directional alignment into the initial pheromone distribution to reduce early-stage randomness. A dynamic smooth state transition model incorporates an angle-based smoothing factor together with an adaptive pseudo-random selection scheme to promote path continuity. Finally, a heterogeneous pheromone update strategy differentiates reinforcement according to path quality and emphasizes elite solutions to improve convergence stability. Extensive simulations on simple, complex, and large-scale grid environments demonstrate that DAACO consistently outperforms the standard ACO and several state-of-the-art ACO variants. Compared with standard ACO, DAACO achieves average reductions of 5.02%–44.12% in path length, 43.75%–81.08% in the number of turns, and 86.07%–98.14% in convergence iterations across different scenarios, confirming its effectiveness and robustness for smooth mobile robot path planning in complex environments.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"192 ","pages":"Article 114782"},"PeriodicalIF":6.6,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146191883","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"Lightweight-oriented classification using distillation learning and spatial information from tabular data","authors":"Minh-Trieu Tran ,&nbsp;Patrizio Pelliccione ,&nbsp;Phuong T. Nguyen","doi":"10.1016/j.asoc.2026.114798","DOIUrl":"10.1016/j.asoc.2026.114798","url":null,"abstract":"<div><div>Spatial information, along with deep learning approaches, plays a crucial role in classification tasks. However, tabular datasets do not inherently contain spatial information, so researchers often focus primarily on traditional machine learning algorithms or DNNs for simplicity. Thus, these methods cannot fully utilize all the information in tabular data because they are unable to mine spatial (2D) information. Additionally, while deep learning, particularly convolutional neural networks (CNNs), can handle spatial information effectively, it requires significant computational resources to achieve high performance. In this paper, we present a new method, named ESATTO (Efficient Synthetic Spatial Awareness for Tabular daTa classificatiOn), which uses distillation learning and is able to extract spatial features in an efficient way. The proposed model is 27 times smaller in size and has 50 times fewer parameters than standard models, making it highly efficient for resource-limited environments. This lightweight model is particularly beneficial to real-world applications, such as precision agriculture, where computational resources are constrained. The validation shows that our model outperforms in wine quality prediction and performs strongly on medical data, highlighting its versatility for various tasks. The code is available at <span><span>https://github.com/tmtgssi/ESATTO</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"192 ","pages":"Article 114798"},"PeriodicalIF":6.6,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146191880","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"Frequency hopping synchronization for satellite communication system using reinforcement learning","authors":"Sangkeum Lee","doi":"10.1016/j.asoc.2026.114648","DOIUrl":"10.1016/j.asoc.2026.114648","url":null,"abstract":"<div><div>Satellite communication systems (SCSs) deployed in tactical environments must maintain reliable links under severe interference, jamming attempts, and large path loss. From a soft computing perspective, frequency-hopping (FH) synchronization in such uncertain and highly dynamic channels is a sequential decision problem that benefits from adaptive, data-driven control. In this paper, we propose a reinforcement learning (RL)–driven FH synchronization framework for dehop–rehop SCSs, where a conventional serial search performs coarse acquisition and a proximal policy optimization (PPO) agent with a GCN–Bi-LSTM network refines the uplink hop timing. The RL agent interacts with the stochastic channel, observing signal-energy patterns and learning to minimize both mean acquisition time (MAT) and mean-squared error (MSE) of the timing estimate without requiring an explicit channel model. Mathematical analysis and Monte Carlo simulations show that the proposed hybrid method reduces the average number of hops required for synchronization by 58.17 % and the MSE of uplink hop-timing estimation by 76.95 % compared with a conventional serial-search scheme. Relative to an early–late-gate synchronization method that combines serial search with an LSTM network, the average number of hops is further reduced by 12.24 % and the MSE by 18.5 %. These results demonstrate that the PPO-based GCN–Bi-LSTM agent provides a flexible soft-computing solution that can adapt to rapidly varying SCS operating conditions while significantly improving FH synchronization performance.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"191 ","pages":"Article 114648"},"PeriodicalIF":6.6,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146039727","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"Accurate solution of high-dimensional partial differential equations: Research on data-physics collaborative modeling and adaptive sampling","authors":"Haoran Du ,&nbsp;Zhao Liu ,&nbsp;Ping Zhu","doi":"10.1016/j.asoc.2025.114472","DOIUrl":"10.1016/j.asoc.2025.114472","url":null,"abstract":"<div><div>Solving high-dimensional partial differential equations (PDEs) remains a fundamental challenge in modeling complex systems in science and engineering. Traditional methods often suffer from significant accuracy limitations due to the \"curse of dimensionality.\" In this work, we propose a bidirectionally coupled data-physics collaborative modeling framework that integrates a Variational Autoencoder (VAE) with a Physics-Informed Neural Network (PINN), significantly improves the solution accuracy of high-dimensional PDEs. On the data-driven side, the VAE compresses high-dimensional data into a latent space to extract essential features; on the physics-constrained side, the PINN embeds physical priors via residual minimization, achieving deep integration of physical laws and data representations. Building upon this, we introduce a dynamically enhanced version, VAE-PINN-DS, which employs an adaptive sampling strategy to allocate computational resources to regions with high residual gradients, low latent-space density, and high uncertainty. Numerical experiments demonstrate the effectiveness of VAE-PINN-DS in solving the Poisson equation (5D–15D), non-uniform elliptic equation (10D), and the 20-dimensional Korteweg-de Vries (KdV) equation. Compared with the baseline PINN and other mainstream methods such as Galerkin and Deep Ritz, VAE-PINN-DS reduces the relative L2 error by 1–3 orders of magnitude and exhibits strong robustness in ultra-high-dimensional scenarios. These results highlight that the synergy between data-driven dimensionality reduction and physics-constrained learning provides a scalable new paradigm for solving high-dimensional PDEs.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"191 ","pages":"Article 114472"},"PeriodicalIF":6.6,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146039726","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"Empirical group sparse decomposition: Algorithm and application","authors":"Sainan Li ,&nbsp;Jing Wu ,&nbsp;Zhengbin Yan ,&nbsp;Jiayi Zheng ,&nbsp;Xianbin Zheng","doi":"10.1016/j.asoc.2026.114752","DOIUrl":"10.1016/j.asoc.2026.114752","url":null,"abstract":"<div><div>Periodic pulse feature components are widely present in complex signals across various fields such as biomedical and aerospace engineering, and their accurate extraction is crucial for monitoring. Aiming to address the core challenges of extracting periodic pulse feature components, this paper proposes a novel signal decomposition method, Empirical Group Sparse Decomposition (EGSD). First, EGSD constructs a unimodal pre-segmentation strategy to determine the frequency band range of the decomposition modes, ensuring the physical significance of the pre-decomposed modes. Next, EGSD designs a group sparse filter bank to decompose the signal into a series of fundamental modes, thereby extracting the hidden periodic pulse feature components within the signal. Finally, EGSD proposes a periodic frequency similarity fusion strategy to fuse the fundamental mode components into a series of eigenmode functions, effectively overcoming the over-decomposition issue. As a novel signal decomposition method, EGSD enables effective signal separation and extracts feature information concealed within noise. This paper validates the effectiveness of EGSD through simulated cases and three real-world applications: mechanical vibration signal feature extraction, EEG signal denoising, and ECG signal separation. The experimental results demonstrate that the EGSD method effectively separates signals, extracts feature information, and mitigates noise interference, underscoring its practical value.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"192 ","pages":"Article 114752"},"PeriodicalIF":6.6,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146191785","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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":"Cost investigations and ANFIS implementation for M/M/1/L queueing model with general vacation, customer discouragement, and F-policy","authors":"Khushbu S. Antala,&nbsp;Sudeep Singh Sanga","doi":"10.1016/j.asoc.2026.114667","DOIUrl":"10.1016/j.asoc.2026.114667","url":null,"abstract":"<div><div>This work investigates the M/M/1/L queueing model within real-life scenarios, highlighting its substantial impact on various domains such as call centers, distributed computing systems, healthcare facilities, traffic management centers, and other service-oriented operations. It focuses on the finite capacity queueing model with general vacations and customer discouragement under the <em>F</em>-policy. Additionally, we consider essential queueing features that account for customers’ behaviors, such as balking and reneging. When customers encounter long queues, they may experience discouragement and consequently choose not to join the queue or leave without receiving service. Implementation of the ‘<em>F</em>-policy’ is an effective strategy for the admission of customers into the system to reduce congestion resulting from excessive customer arrivals. Further, the server chooses to take a vacation when no customer is present in the system. The vacation time is not exponentially distributed; instead, we consider the general case of vacation time for a more comprehensive analysis. The mathematical development of the M/M/1/L queueing model is carried out using the Chapman-Kolmogorov steady-state equations by introducing supplementary variables corresponding to remaining vacation times. Subsequently, the Laplace-Stieltjes transform and recursive method are employed to solve these equations and establish probability distributions. Further, various performance metrics, such as the number of customers in the system, throughput, customer loss, long-run probabilities, etc., are derived. These performance measures will assist system organizers in making informed decision-making strategies. A numerical example is also presented, illustrating the impact of input parameters and customers’ discouraged behavior on various performance metrics. The adaptive neuro-fuzzy inference system, which is built on an artificial neural network and a support vector regression, a machine learning technique, is employed to validate the numerical results. Moreover, the nonlinear cost function is formulated with the service and vacation rates as decision variables. The cost is minimized using quasi-Newton methods and several metaheuristics (particle swarm optimization, bat algorithm, and differential evolution variants). These algorithms are utilized to compare the optimal values of the cost function. The queueing model’s practical application is demonstrated through its implementation in fog computing systems.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"191 ","pages":"Article 114667"},"PeriodicalIF":6.6,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080042","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}