Jiaxin Zhang , Yulong Wang , Tongcun Liu , Lei Zhang , Wei Li , Jianxin Liao
{"title":"Hyperbolic spatial-temporal network for session-based recommendation","authors":"Jiaxin Zhang , Yulong Wang , Tongcun Liu , Lei Zhang , Wei Li , Jianxin Liao","doi":"10.1016/j.asoc.2025.113996","DOIUrl":"10.1016/j.asoc.2025.113996","url":null,"abstract":"<div><div>The goal of Session-based recommendation is to recommend the next item a user may be interested in based on historical click data, without relying on user profiles. Compared to the data from other recommendation scenarios, session data is typically more sparse, thus Self-supervised learning (SSL), which derives ground-truth from raw data, has gradually gained attention. Existing SSL methods usually augment data by dropping or transforming original sequences in Euclidean space, leading to two problems. Firstly, in Euclidean space, it is difficult to handle the distorted distribution and hierarchical nature of session data, and secondly, the essential spatial and temporal features of session data are usually overlooked. To address these issues, we propose the <em>Hyperbolic Spatial-Temporal Network</em> (<strong><em>HSTN</em></strong>), which enhances the performance by generating positive samples with the spatial-temporal features of session data in hyperbolic space. Specifically, we first use a hyperbolic hypergraph neural network as the base encoder to mitigate the influence of distorted data distribution. Then, we employ a spatial-temporal feature learning module to generate positive samples with spatial-temporal features for contrastive learning. Extensive experiments on three real-world datasets demonstrate that our proposed method achieves improvements of 11.6 %, 17.6 %, and 38.4 %, respectively, compared to current state-of-the-art methods, under P@10 metric.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113996"},"PeriodicalIF":6.6,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145268122","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":"FSAMLM: A few-shot adaptation multimodal large model for cross-domain fault diagnosis","authors":"Xin Zhang , Shixi Liu , Li Jiang , Yibing Li","doi":"10.1016/j.asoc.2025.113985","DOIUrl":"10.1016/j.asoc.2025.113985","url":null,"abstract":"<div><div>Recent advances in deep learning have demonstrated remarkable performance in mechanical fault diagnosis. However, most existing approaches are designed for a single data modality or a single task, limiting their flexibility and generalization ability. Moreover, common knowledge across different tasks remains largely unexploited. Inspired by the success of large-scale models in natural language processing and computer vision, integrating multimodal data with multi-task learning strategies may significantly improve the performance of fault diagnosis models. Nonetheless, a significant gap exists between general pre-trained knowledge and domain-specific expertise, posing considerable challenges for effective integration. To address these issues, we propose a novel few-shot adaptation multimodal large model (FSAMLM) that efficiently captures shared representations through a parameter-efficient fine-tuning strategy, structured in two stages. Specifically, we first design a low-rank adaptation meta-learning (LoRAML) framework, which employs low-rank decomposition on pre-trained parameters to reduce computational complexity and improve robustness. This approach not only accelerates adaptation to new few-shot tasks but also preserves pre-training knowledge effectively. During the second fine-tuning stage, we implement target-domain training with limited samples and introduce a training-free inference option for real-world deployment. Experimental validation using six public datasets demonstrates FSAMLM’s superior domain generalization capability in few-shot fault diagnosis tasks compared to existing methods. The code repository is publicly available at: <span><span>https://github.com/ohhyeeaah/FSAMLM-for-fault-diagnosis</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113985"},"PeriodicalIF":6.6,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145221557","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":"A proximal policy optimization driven hyper-heuristic for workers constrained hybrid flow shop problem","authors":"Shengnan Ding , Weishi Shao , Zhongshi Shao , ShengTao Peng , Dechang Pi , Jiaquan Gao","doi":"10.1016/j.asoc.2025.113990","DOIUrl":"10.1016/j.asoc.2025.113990","url":null,"abstract":"<div><div>As the production environment becomes increasingly complex, the integration of soft computing techniques becomes essential for addressing resource-constrained scheduling problems. This paper delves into a worker constrained hybrid flow shop scheduling problem (WHFSP) that integrates worker resources at each processing stage. A mixed-integer linear programming (MILP) model is constructed which enables the use of mathematical solvers to obtain optimal solutions for small-scale instances. Additionally, a novel soft computing-based scheduling framework, namely a proximal policy optimization-based hyper-heuristic algorithm (PPO-HH), is proposed. It automatically selects the most suitable low-level heuristic strategies based on the current state and historical data, facilitating efficient exploration and exploitation of the complex decision space. Several low-level heuristics including perturbative and local search operators are developed to explore the solution space. Subsequently, a high-level control strategy based on proximal policy optimization is proposed. A solution quality evaluation function and a reward mechanism based on problem characteristics are formulated. This mechanism provides feedback to PPO-HH based on the degree of alignment between the actions taken by the agent and the objectives, gradually optimizing the selection of low-level heuristic strategies. Eventually, it generates a probability distribution for each low-level heuristic in the given environment. Comprehensive numerical experiments are conducted to evaluate the performance of both the MILP model and the components of the PPO-HH algorithm. The comparison results show that PPO-HH is effective and efficient for solving the WHFSP.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113990"},"PeriodicalIF":6.6,"publicationDate":"2025-09-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145183700","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}
Ye Liu , Jiahao Wang , Nan Zhang , Xiaodong Qian , Shaojun Chai
{"title":"Beyond training horizons: A physics-informed U-Net transformer for time-dependent CO₂ storage simulation in environmental applications","authors":"Ye Liu , Jiahao Wang , Nan Zhang , Xiaodong Qian , Shaojun Chai","doi":"10.1016/j.asoc.2025.113987","DOIUrl":"10.1016/j.asoc.2025.113987","url":null,"abstract":"<div><div>CO<sub>2</sub> geological storage is a key strategy for reducing greenhouse gas emissions, requiring accurate modeling of subsurface CO<sub>2</sub> migration is essential for effective storage planning and risk assessment. Conventional numerical simulations, which solve time-dependent nonlinear partial differential equations, provide detailed physical insights but are computationally demanding, especially for large-scale or long-term scenarios. To improve computational efficiency, surrogate models based on machine learning have been increasingly investigated. Methods such as deep learning and physics-Informed neural networks aim to approximate the behavior of physical systems, offering potential reductions in simulation time. However, these approaches often require extensive case-specific datasets and are typically limited to fixed time horizons defined during training, which can restrict their generalizability and practical application. This study presents a time-aware surrogate modeling framework that combines convolutional neural networks with self-attention mechanisms to address these limitations. Drawing inspiration from autoregressive forecasting used in sequential learning models, the proposed approach captures temporal dependencies through iterative prediction of system states.The framework requires only a short-term numerical simulation to initialize the physical system, after which it can generate predictions for arbitrarily extended time horizons without the need for retraining. By enabling long-term forecasting, the method improves efficiency and supports repeated scenario evaluations, such as site screening and well placement optimization. Such predictive capabilities are particularly valuable in addressing environmental sustainability goals, where rapid and scalable simulations are essential for managing long-term subsurface processes under climate-related constraints.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113987"},"PeriodicalIF":6.6,"publicationDate":"2025-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222104","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}
Jie Ouyang , Yangfan Liang , Hong Sun , Xianchao Zhang , Jingxue Chen , Gao Liu , Zhiquan Liu , Yining Liu
{"title":"A dual-branch self-supervised contrastive learning framework for emotion recognition based on time-frequency fusion","authors":"Jie Ouyang , Yangfan Liang , Hong Sun , Xianchao Zhang , Jingxue Chen , Gao Liu , Zhiquan Liu , Yining Liu","doi":"10.1016/j.asoc.2025.113958","DOIUrl":"10.1016/j.asoc.2025.113958","url":null,"abstract":"<div><div>Emotion recognition based on electroencephalography(EEG) signals is becoming a prominent research hotspot due to its wide-ranging applications in brain–computer interfaces (BCIs), mental health assessment, and human-computer interaction. Traditional emotion recognition methods often rely on supervised learning, which requires large amounts of labeled data to effectively train deep models. However, EEG signals exhibit inherent complexity and substantial variability across individuals and sessions, making it challenging to obtain consistent and reliable labels. In this paper, we propose a novel pretraining framework for EEG-based emotion recognition that enables mutual learning between time-domain and time-frequency-domain representations, while requiring simple network architectures. Experimental results demonstrate that our method achieves 84.39 % accuracy on the SEED dataset, and 89.01 % valence accuracy and 79.75 % arousal accuracy on the DEAP dataset using only 10 % labeled data, indicating strong performance under limited label conditions. Furthermore, we evaluate the transfer learning capability of our framework by pretraining it on the SEED dataset and then fine-tuning it on SEED-V. This cross-dataset transfer leads to a 1.9 % absolute improvement in classification accuracy on SEED-V, demonstrating the effectiveness of the learned representations in generalizing across datasets.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113958"},"PeriodicalIF":6.6,"publicationDate":"2025-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145268123","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":"Improving mutation strategies in differential evolution with a new pbest selection mechanism","authors":"Jan Popič, Borko Bošković, Janez Brest","doi":"10.1016/j.asoc.2025.113978","DOIUrl":"10.1016/j.asoc.2025.113978","url":null,"abstract":"<div><div>Differential evolution, which belongs to a group of population-based algorithms, has received a lot of research attention since its introduction in 1995. A population-based algorithm is required to guide individuals to visit potentially better basins of attraction in the search space when searching for a globally optimal solution. Additionally, individuals need to interact with each other during an evolutionary process to explore the search space effectively. In this paper, we propose a novel pbest selection mechanism for <em>DE/current-to-pbest</em> mutation strategy and its variants designed to enhance the potential for exploration of different attraction basins. The proposed mechanism enforces a minimal distance between the selected pbest individual and all other better individuals. This means that possible candidates for the pbest individual, used in mutation, are further spaced apart. As a result, the likelihood that the new trial vector will be generated in a different attraction basin of the search space is increased. The mechanism is incorporated into the L-SHADE, jSO, and L-SRTDE algorithms, and its effectiveness is evaluated using CEC’24 benchmark functions. Experimental results demonstrate improvements in the performance of the selected algorithms, particularly in higher-dimensional problem instances.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113978"},"PeriodicalIF":6.6,"publicationDate":"2025-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145268654","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}
Xingyuan Lu , Yanbing Xue , Leida Li , Shiyin Li , Zan Gao
{"title":"Mamba-based multi-branch cost aggregation for stereo matching","authors":"Xingyuan Lu , Yanbing Xue , Leida Li , Shiyin Li , Zan Gao","doi":"10.1016/j.asoc.2025.113973","DOIUrl":"10.1016/j.asoc.2025.113973","url":null,"abstract":"<div><div>This study presents Mamba-Based Multi-Branch Cost Aggregation for Stereo Matching (MMBStereo), an innovative real-time stereo matching framework with high performance. The core innovation lies in the Mamba-based multi-branch cost aggregation network, which uses a unique three-branch aggregation strategy. The Mamba Aggregation Branch integrates the State Space Model from the Mamba structure, replacing conventional convolution and Transformer methods, significantly enhancing network performance and efficiency. The Spatial Aggregation Branch addresses the loss of spatial texture information, improving the scene’s contextual representation. Meanwhile, the Edge Aggregation Branch enhances edge responses, improving object boundary detection accuracy. Through a carefully designed multi-branch fusion strategy, the framework improves disparity prediction accuracy while maintaining real-time inference. Our method achieves competitive accuracy with non-real-time stereo matching frameworks, surpassing existing lightweight solutions in the widely recognized KITTI benchmark tests.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113973"},"PeriodicalIF":6.6,"publicationDate":"2025-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145268120","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}
Shujiao Liao , Nan Zhou , Ling Wei , Weiping Ding , Yidong Lin
{"title":"Rapid attribute and scale selection with adaptive three-way sampling and neighborhood rough set","authors":"Shujiao Liao , Nan Zhou , Ling Wei , Weiping Ding , Yidong Lin","doi":"10.1016/j.asoc.2025.113966","DOIUrl":"10.1016/j.asoc.2025.113966","url":null,"abstract":"<div><div>As a key issue in knowledge reduction for multi-scale data, attribute and scale selection has attracted increasing attention in recent years. However, with the rapid growth of data volumes, most existing methods are inefficient for large-scale multi-scale data and struggle to effectively handle heterogeneous multi-scale data, significantly limiting their practical applications. To address this situation, this paper proposes a rapid attribute and scale selection method to deal with large nominal-and-numerical mixed multi-scale decision systems (NN-MDSs). First, the theory and algorithm of adaptive incremental support vector data description (AISVDD) are presented. The AISVDD algorithm overcomes the limitations of traditional support vector data description methods and efficiently processes both class-balanced and class-imbalanced data. By using the algorithm, support vectors can be quickly obtained from large data. Next, an adaptive three-way sampling technique is derived by combining AISVDD and three-way decision. With this technique, support vectors are extracted as sampling results and put into the boundary region, and outliers are seen as noise and put into the negative region. This significantly reduces the data size and improves the data quality. Then, a neighborhood rough set model is built to describe NN-MDSs. Multiple concepts and properties are discussed in the model. Finally, a heuristic attribute and scale selection algorithm is designed to simultaneously choose attributes and scales from the sampled NN-MDS. Detailed experiments demonstrate the effectiveness and superiority of the proposed method. The method performs better than state-of-the-art attribute and scale selection methods on both computational efficiency and classification performance under six benchmark classifiers. It is powerful in handling large NN-MDSs with complex characteristics. This work provides new insights into the complex multi-scale data processing.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113966"},"PeriodicalIF":6.6,"publicationDate":"2025-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145268121","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":"Evaluating CEO hubris effects on sustainable performance in the IC design industry: An integrated dynamic network DEA framework with machine learning","authors":"Sheng-Wei Lin, Yu-Rou Lin","doi":"10.1016/j.asoc.2025.113986","DOIUrl":"10.1016/j.asoc.2025.113986","url":null,"abstract":"<div><div>This study introduces an integrated analytical framework combining dynamic network data envelopment analysis (DNDEA) with machine learning to assess the impact of CEO hubris on sustainable performance in the integrated circuit (IC) design industry. Our two-stage DNDEA model evaluates operational and R&D efficiency separately, incorporating intermediate factors including profit and ESG scores. We develop a novel text-based measure of CEO hubris by analyzing the contrast between confidence and conservatism language in annual shareholder reports. This hubris measure is then incorporated into predictive models, where we compare traditional linear regression against advanced machine learning approaches—support vector regression (SVR) and random forest (RF)—using cross-validation and hyperparameter optimization. The analysis reveals a significant negative correlation between CEO hubris and operational and R&D efficiency. Notably, the non-linear models (SVR and RF) demonstrate superior predictive accuracy compared to linear regression across varying levels of CEO hubris. These findings yield two primary contributions: first, they establish the critical need for monitoring hubristic leadership behavior in innovation-intensive industries, given their detrimental effect on organizational efficiency. Second, they validate the effectiveness of combining text analytics, DNDEA efficiency metrics, and machine learning for evaluating leadership impact on firm performance. This methodology provides a comprehensive framework for analyzing leadership dynamics in the IC design sector and offers an adaptable template for similar analyses across technology-driven industries.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113986"},"PeriodicalIF":6.6,"publicationDate":"2025-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145183699","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":"Federated deep embedded clustering under privacy protection","authors":"Xiao Xu , Hong Liao , Xu Yang","doi":"10.1016/j.asoc.2025.113963","DOIUrl":"10.1016/j.asoc.2025.113963","url":null,"abstract":"<div><div>Deep clustering, a prominent research focus in data mining, utilizes deeply embedded features to reveal the intrinsic statistical structure of data. However, existing deep clustering models rely on centralized datasets, which are impractical in scenarios with data silos and privacy constraints, leading to degraded clustering performance. Considering the privacy protection characteristics of federated learning, this paper incorporates the idea of federated learning into deep clustering, and proposes a federated deep embedding clustering (FDEC) model under privacy protection. FDEC follows a universal client-server structure, coordinating training between clients through a central server to obtain a unified central model. The server updates the global deep embedding and cluster centers based on a hybrid federated averaging strategy, while each client conducts two-stage deep clustering on local data without sharing raw data. To enhance robustness under non-independent and identically distributed (non-IID) conditions, the hybrid strategy improves parameter aggregation effectiveness. Experimental results on both IID and non-IID datasets demonstrate that FDEC is more robust and consistently outperforms centralized deep clustering methods.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113963"},"PeriodicalIF":6.6,"publicationDate":"2025-09-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222106","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}