{"title":"MCSMOTE: A transition matrix-driven oversampling technique for class imbalance","authors":"Fatih Sağlam , Mehmet Ali Cengiz","doi":"10.1016/j.asoc.2025.113906","DOIUrl":"10.1016/j.asoc.2025.113906","url":null,"abstract":"<div><div>Class imbalance presents a challenge in machine learning, often skewing predictive performance toward the majority class and undermining the accuracy of minority class predictions. To address this, we introduce MCSMOTE, a novel resampling method that employs a transition matrix-based Monte Carlo mechanism for generating synthetic samples. MCSMOTE differentiates itself by modeling the relationships among features and leveraging probabilistic transitions to generate synthetic data points that effectively capture the underlying data structure. This approach ensures enhanced representativeness of the minority class while approximating the local structure of the minority class and thereby generating samples that reflect the underlying data patterns. Comprehensive experiments across 63 diverse imbalanced datasets demonstrate that MCSMOTE consistently outperforms nine widely used resampling techniques—NORES, ROS, SMOTE, ADASYN, BLSMOTE, RWO, SMOTEWB, DeepSMOTE, RWO, and GQEO—when evaluated using multiple classifiers and six key performance metrics: balanced accuracy, F1-score, G-mean, MCC, ROCAUC, and ROCAUC. Results show that MCSMOTE achieves the highest average performance across all metrics. Friedman and Nemenyi tests confirm that these improvements are statistically significant. An ablation study further highlights the stability and effectiveness of MCSMOTE’s hyperparameter choices across different data characteristics. These findings establish MCSMOTE as a powerful and reliable solution for addressing class imbalance in machine learning applications.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113906"},"PeriodicalIF":6.6,"publicationDate":"2025-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222105","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}
Tianyi Zhou , Zijie Huang , Hui Lin , Zhaobin Zhou , Jia Hu
{"title":"MACityChat: Integrating remote sensing professional large model with general-purpose large model for multi-domain urban land use analysis","authors":"Tianyi Zhou , Zijie Huang , Hui Lin , Zhaobin Zhou , Jia Hu","doi":"10.1016/j.asoc.2025.113929","DOIUrl":"10.1016/j.asoc.2025.113929","url":null,"abstract":"<div><div>Urbanization remains a global trend, with urban land use being a key component of the process. The effective integration and management of land use are critical for the sustainable development of cities. Traditional urban land use analysis methods can fit dynamic models of land use changes nonlinearly, but they face two challenges: First, the analysis process of existing technologies is often a black-box, with unknown principles, reducing the reliability and authenticity of results. Second, traditional machine learning can only analyze urban land use changes from a single domain, such as remote sensing, overlooking the influence of economic and sociological factors. We propose an interpretable urban land use change analysis task and design MACityChat, a framework that combines remote sensing-specific large models with general-purpose large language models for multidisciplinary generalized analysis, while also visualizing the model’s analytical results. In this framework remote sensing images are input into a remote sensing large model, which transforms the semantic objects in the images into textual descriptions. These descriptions are then fed into a general-purpose large language model. A regional tag-guiding module directs the general-purpose language model to incorporate local economic, policy, and cultural knowledge to perform generalized analysis. Finally, the analysis results are visualized on the remote sensing images, providing a detailed examination of urban land use. Extensive experiments show that MACityChat can provide detailed and effective analyses of urban land use changes and visualize these analyses, offering an interpretable and superior solution to urban land use problems.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113929"},"PeriodicalIF":6.6,"publicationDate":"2025-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145183698","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}
Xinjie Wan , Hao Pu , Paul Schonfeld , Yang Ran , Taoran Song , Lihui Peng
{"title":"Transformer-guided optimization of railway alignments using an adaptive rapidly-exploring random tree algorithm","authors":"Xinjie Wan , Hao Pu , Paul Schonfeld , Yang Ran , Taoran Song , Lihui Peng","doi":"10.1016/j.asoc.2025.113977","DOIUrl":"10.1016/j.asoc.2025.113977","url":null,"abstract":"<div><div>Railway alignment design is a crucial part of a railway project. Despite the widespread success of computer-aided alignment optimization methods in determining alignments, effectively exploring the objective function’s descent direction (OFDD) remains challenging, particularly when navigating complex alignment search spaces. To address this issue, it is essential to comprehensively consider key factors, including the global and local environment, explored and unexplored search spaces, as well as established alignment search strategies and potential new ones that may emerge during the OFDD optimizing process. Therefore, an alignment-oriented Transformer framework is formulated in this work. In this framework, various real-world railway cases are input into a stacked Transformer framework to learn an optimized OFDD strategy. Specifically, the model handles regular inputs (i.e., global and local contexts, long-term goals) and irregular inputs (i.e., historical paths) using two separate stacked Transformer encoders. Afterward, an adaptive rapidly-exploring random tree star (Ada-RRT-star) method is developed by integrating the Transformer framework’s output to guide RRT’s search direction as well as to enhance the solution quality. Ultimately, the proposed method is applied to a realistic railway case, where the results demonstrate its superiority over the conventional 3D-RRT-star algorithm in terms of solution quality. Besides, the best alignment generated by the Ada-RRT-star also outperforms the manually-designed alignment.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113977"},"PeriodicalIF":6.6,"publicationDate":"2025-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145183648","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}
Qiong Gu , Yanchi Li , Wenyin Gong , Zhiyuan Yuan , Bin Ning , Chunyang Hu , Jicheng Wu
{"title":"Progressive auto-encoding for domain adaptation in evolutionary multi-task optimization","authors":"Qiong Gu , Yanchi Li , Wenyin Gong , Zhiyuan Yuan , Bin Ning , Chunyang Hu , Jicheng Wu","doi":"10.1016/j.asoc.2025.113916","DOIUrl":"10.1016/j.asoc.2025.113916","url":null,"abstract":"<div><div>In recent years, evolutionary multi-task optimization (EMTO) has emerged as an effective paradigm for solving multiple optimization tasks simultaneously by leveraging knowledge transfer across tasks. The domain adaptation technique plays an important role in EMTO, as it helps align search spaces to support knowledge transfer among tasks. However, most existing domain adaptation methods rely on static pre-training or periodic re-matched mechanism, which do not adapt well to the dynamic change in evolving populations. In this paper, we propose a progressive auto-encoding (PAE) technique that enables continuous domain adaptation throughout the EMTO process. The PAE incorporates two complementary adaptation strategies: i) segmented PAE, which employs staged training of auto-encoders to achieve effective domain alignment across different optimization phases, and ii) smooth PAE, which utilizes eliminated solutions from the evolutionary process to facilitate more gradual and refined domain adaptation. We integrate the PAE into both single-objective and multi-objective multi-task evolutionary algorithms, yielding <em>MTEA-PAE</em> and <em>MO-MTEA-PAE</em>, respectively. Comprehensive experiments conducted on six benchmark suites and five real-world applications validate the effectiveness of our proposed PAE technique in enhancing domain adaptation capabilities within EMTO.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113916"},"PeriodicalIF":6.6,"publicationDate":"2025-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145221561","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":"Decision analytics for Indian culinary tourism: A holistic group approach considering correlation","authors":"Kaushik Debnath, Sankar Kumar Roy","doi":"10.1016/j.asoc.2025.113812","DOIUrl":"10.1016/j.asoc.2025.113812","url":null,"abstract":"<div><div>In today’s data-driven world, making informed decisions in dynamic fields like culinary tourism is crucial. An enhanced multi-attribute decision-making (MADM) model is presented in this study to tackle the uncertainty and interdependencies of India’s culinary tourism landscape. The main goals are to (i) address uncertainty and correlations in MADM scenarios, (ii) calculate objective attribute weights, and (iii) resolve conflicts among alternatives based on preference, indifference, and incomparability. To manage uncertainty, the proposed model incorporates <span><math><mrow><mi>r</mi><mo>,</mo><mi>s</mi></mrow></math></span>-quasirung orthopair fuzzy set (<span><math><mrow><mi>r</mi><mo>,</mo><mi>s</mi></mrow></math></span>-QOFS), while to capture relational dynamics among factors Aczel–Alsina operations based geometric Heronian mean operator is developed. Attribute weighting is performed with MEREC (method based on the removal effects of criteria) method, while a modified ORESTE (organísation, rangement et Synthèse dedonnées relarionnelles (in French)) method within the <span><math><mrow><mi>r</mi><mo>,</mo><mi>s</mi></mrow></math></span>-QOFS is initiated to rank alternatives, introducing a new ranking measure in place of Besson’s traditional rank. Finally, to test the effectiveness and practical value, a case study of culinary tourism destinations across 36 Indian states and union territories is conducted and then ranked using the proposed model. The results highlight southern Indian states as preferred destinations. Thus, this work contributes in two ways: first, by providing a general decision-making model for imprecise and data, and second, by offering valuable insights into the future of Indian culinary tourism.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113812"},"PeriodicalIF":6.6,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145158926","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":"Fall detection system using tabular GAN for data augmentation with integration of isolation forest model","authors":"Ali Nawaz , Najah Abu Ali , Amir Ahmad","doi":"10.1016/j.asoc.2025.113931","DOIUrl":"10.1016/j.asoc.2025.113931","url":null,"abstract":"<div><div>Effective Fall Detection Systems (FDS) are essential to minimize the risk of severe injuries among the elderly. However, the limited availability and imbalanced nature of fall detection data pose significant challenges to developing accurate models. This paper proposes a novel approach to enhance fall detection accuracy by using synthetic data generated through Tabular Generative Adversarial Network (GAN), combined with Isolation Forest and Autoencoder models. The dataset was augmented by factors of 5 and 10, and the models were evaluated using the area under the curve-receiver operating characteristic (AUC-ROC) and area under the curve–precision recall (AUC-PR) metrics. Notably, the Isolation Forest model improved from an AUC-ROC of 0.49 and AUC-PR of 0.43 (without augmentation) to 0.59 and 0.63, respectively, with 5x augmentation. Similarly, the Autoencoder showed an increase from 0.4 (AUC-ROC) and 0.2 (AUC-PR) to 0.5 and 0.54 with the same augmentation. These results demonstrate the effectiveness of synthetic data in improving anomaly detection performance. The findings suggest that advanced data augmentation techniques significantly improve FDS, thereby enhancing safety and quality of life for the vulnerable population.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113931"},"PeriodicalIF":6.6,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145221599","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 granularity time series forecasting model combining three-way decision and trend information granule","authors":"Jianuan Qiu, Shuhua Su, Jingjing Qian","doi":"10.1016/j.asoc.2025.113957","DOIUrl":"10.1016/j.asoc.2025.113957","url":null,"abstract":"<div><div>Long-term forecasting of time series plays a vital role across diverse applications but is challenged by error accumulation arising from recursive predictions and the insufficient retention of trend information in conventional methods. To tackle these issues, we propose a novel forecasting model based on granular time series (GTS). The model utilizes an improved <span><math><msub><mrow><mi>L</mi></mrow><mrow><mn>1</mn></mrow></msub></math></span>-trend filtering technique to achieve optimal segmentation of information granules, preserving essential trend features. Subsequently, we introduce dual evaluation functions based on distance similarity to jointly drive the three-way decision (TWD) process for aggregating information granules, thereby effectively reducing error propagation. Finally, the aggregated granules serve as inputs to a long short-term memory (LSTM) neural network to generate accurate forecasts. In addition, the proposed model is evaluated on several real-world datasets through sensitivity and comparative analyses. The results demonstrate that the model exhibits strong performance in long-term forecasting tasks.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113957"},"PeriodicalIF":6.6,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145119755","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}
Xianlun Tang , Xiaodong Qian , Jie Li , Binyu Lu , Wuquan Deng , Weisheng Li
{"title":"Joint stage features modulation for progressive blind face restoration","authors":"Xianlun Tang , Xiaodong Qian , Jie Li , Binyu Lu , Wuquan Deng , Weisheng Li","doi":"10.1016/j.asoc.2025.113974","DOIUrl":"10.1016/j.asoc.2025.113974","url":null,"abstract":"<div><div>A significant challenge for Blind Face Restoration (BFR) is to cope with the degraded information of unknown parameters in face images. The BFR method has evolved from non-prior to prior-based methods, but there are still some shortcomings. The quality of priors seriously affects the restoration results, especially in scenarios with severe degradation. Simultaneously encoding or modulating degraded images directly into the restoration process can introduce degraded information, leading to poor visual perception. Therefore, we propose a progressive restoration model with the joint stage features modulation, named JSFM-GAN. JSFM-GAN can be seen as having two stages. In the first stage, the LQ image is modulated with the facial resolution map to provide a rough structure for recovery. In the second stage, Joint Stage Feature Modulation (JSFM) utilizes the LQ images and stage features for joint modulation on multiple scales to balance fidelity and realism by combining clean spatial information of stage features and the tonal structure of LQ images. At the same time, the Up-Sampling Feature Supplement Block (UFSB) is used to reduce information loss due to channel fading and improve the network’s focus on face components and textures. In addition, we use the stage reconstruction loss and adjusted facial parsing maps to enhance the realism and symmetry of the generated results. Experiments with JSFM-GAN on synthetic and real-world datasets achieve good results, demonstrating the superior performance of our method.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113974"},"PeriodicalIF":6.6,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145221562","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 Zhu , Fengmei Liu , Jingzhe Sun , Haiping Huang
{"title":"Deadline-constrained and bi-objective workflow scheduling with fuzziness in cloud computing systems","authors":"Jie Zhu , Fengmei Liu , Jingzhe Sun , Haiping Huang","doi":"10.1016/j.asoc.2025.113970","DOIUrl":"10.1016/j.asoc.2025.113970","url":null,"abstract":"<div><div>Deploying intelligent expert systems on the cloud is a cost-effective solution for handling compute-intensive service requests, where each request is treated as a workflow processing procedure. Cloud workflows break down complex service requests into smaller tasks and take advantage of automatic maintenance services provided by the cloud workflow management system (CWMS). CWMS can orchestrate task execution, handle dependencies, and direct dynamic scaling of resources based on workload demands. For a CWMS, the main challenge lies in the uncertainty of workflow scheduling, i.e., the processing time, the data transmission time, and the due date are not crisp values. This paper investigates the problem of bi-objective workflow scheduling under fuzziness, aiming to minimize both the total rental cost and the degree of user dissatisfaction. Triangular fuzzy numbers are used to represent the uncertainty of temporal parameters. Two general pricing models in cloud systems are considered: on-demand and reserved price structures. A bi-objective fuzzy workflow scheduling framework is proposed, which consists of workflow sequencing, fuzzy solution generation and solution improvement components. The workflow sequencing component determines the priorities of the workflows. The fuzzy solution generation component assigns tasks to appropriate resources. The Simulated Annealing with Variable Neighborhood Search (SAVNS) method is developed for the solution improvement component. The experimental results demonstrate that the proposal can achieve better effectiveness and robust performance than the baseline algorithms compared. The proposed method can offer practical solutions for CWMS to optimize cost-performance trade-offs in uncertain environments.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113970"},"PeriodicalIF":6.6,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145268652","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":"Multilinear hyperspectral unmixing based on autoencoder and recurrent neural network","authors":"Zehui Jin , Xiaorui Yi , Yue Liu , Hongjuan Zhang","doi":"10.1016/j.asoc.2025.113972","DOIUrl":"10.1016/j.asoc.2025.113972","url":null,"abstract":"<div><div>Spectral unmixing techniques estimate the endmember spectra and corresponding abundance fractions that constitute the pixels of hyperspectral remote sensing images, revealing the mixing mechanisms of materials within pixels. In recent years, deep learning has shown significant potential in advancing spectral unmixing, particularly in nonlinear scenarios. However, most existing nonlinear models rely on bilinear mixing frameworks, with limited focus on high-order nonlinear models. This restricts their ability to capture complex interactions such as multiple light scattering events. To address this issue, this work proposes an unsupervised unmixing method leveraging an autoencoder network framework and the multilinear mixing model (MLM). It employs a recurrent neural network (RNN) in the decoder to simulate the multiple scattering of light between materials. Unlike conventional multilinear approaches that rely on explicit mathematical formulations, the proposed method leverages the RNN to automatically learn and approximate the nonlinear interactions of light. Moreover, the RNN weights are adaptively updated during training and interpreted as transition probabilities representing further light interactions among materials, endowing the model structure with explicit physical interpretation. Besides, a new stopping criterion is also designed, which ensures better RNN weights are obtained during backpropagation. Experiments conducted on both synthetic and real datasets demonstrate the better performance of the proposed method.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"185 ","pages":"Article 113972"},"PeriodicalIF":6.6,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222107","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}