Applied Soft Computing最新文献

{"title":"Analytical-heuristic modeling and optimization for low-light image enhancement","authors":"Axel Martinez ,&nbsp;Emilio Hernandez ,&nbsp;Matthieu Olague ,&nbsp;Gustavo Olague","doi":"10.1016/j.asoc.2025.113546","DOIUrl":"10.1016/j.asoc.2025.113546","url":null,"abstract":"<div><div>Low-light image enhancement remains an open problem, and the new wave of artificial intelligence is at the center of this problem. This work describes the use of genetic algorithms for optimizing analytical models that can improve the visualization of images with poor light. The main goal of low-light image enhancement is to produce an image with features similar to those of a well-taken photograph under optimal lighting conditions. We propose a balanced analytical-heuristic method combined with optimization reasoning to approach a solution to the physical and computational aspects of transforming dark images into visible ones. The experiments demonstrate that the dichotomy-tuned approach ranks at the top among 26 state-of-the-art algorithms in the LOL (LOw-Light) benchmark, reaching a staggering 27.1717 in the synthetic version LOLv2. Moreover, the proposed dichotomy-tuned algorithm provides a pleasant visual appearance with room for improvement. The results show evidence that a simple genetic algorithm combined with analytical reasoning can defeat the current mainstream in a challenging computer vision task through controlled experiments and objective comparisons. This work opens interesting new research avenues for the soft computing community and others interested in analytical and heuristic reasoning.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113546"},"PeriodicalIF":7.2,"publicationDate":"2025-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144632569","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":"Convolutional and ℓ21-norm neural network for bone age estimation","authors":"M.A. Ganaie ,&nbsp;Jha Rohan ,&nbsp;Krish Agrawal ,&nbsp;Rupal Shah ,&nbsp;Anouck Girard ,&nbsp;Joséphine Kasa-Vubu ,&nbsp;M. Tanveer","doi":"10.1016/j.asoc.2025.113456","DOIUrl":"10.1016/j.asoc.2025.113456","url":null,"abstract":"<div><div>Bone age (BA) assessment is critical for evaluating children for potential endocrine, genetic and growth disorders. The evaluation of BA reading may vary among the readers. We use an Inception-v3 convolutional neural network to extract features and propose the novel <span><math><msub><mrow><mi>ℓ</mi></mrow><mrow><mn>21</mn></mrow></msub></math></span>-norm random vector functional link neural network (LR21-RVFL) for the automatic assessment of bone age. Random vector functional link neural network (RVFL) suffers in the presence of noise and outliers due to the squared loss function. To overcome these challenges, we incorporate an <span><math><msub><mrow><mi>ℓ</mi></mrow><mrow><mn>21</mn></mrow></msub></math></span>-norm-based loss function in the RVFL model to improve the robustness of the model. Moreover, we used <span><math><msub><mrow><mi>ℓ</mi></mrow><mrow><mn>21</mn></mrow></msub></math></span>-based regularization to suppress the redundant/irrelevant features and hence, generate a less complex model. The proposed LR21-RVFL model achieves better performance compared to baseline models (except R21-RVFL) in bone age prediction. Moreover, we evaluate the models on the classification of UCI and KEEL datasets.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113456"},"PeriodicalIF":7.2,"publicationDate":"2025-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144665517","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":"Integrating feature selection and fuzzy decision-making: A spherical triangular fuzzy number based framework for large-scale decision-making","authors":"Priya Sharma ,&nbsp;Mukesh Kumar Mehlawat ,&nbsp;Pankaj Gupta ,&nbsp;Weiping Ding","doi":"10.1016/j.asoc.2025.113535","DOIUrl":"10.1016/j.asoc.2025.113535","url":null,"abstract":"<div><div>This study introduces a novel Fuzzy Large-Scale Decision-Making (FLSDM) framework designed to address the complexities of managing a large number of criteria in fuzzy decision-making contexts. While Multi-criteria decision-making (MCDM) methods are widely used across disciplines, traditional approaches often struggle when confronted with high-dimensional decision parameters. To overcome this, we propose an integrated feature selection algorithm that integrates machine learning (ML) algorithms, namely Extreme Gradient Boosting (XGBoost), Support Vector Machine-Recursive Feature Elimination (SVM-RFE), and ReliefF, within a triangular spherical fuzzy (STFN) environment to select the core criteria from a large dataset. Additionally, we extend the Integrated Determination of Objective Criteria Weights (IDOCRIW) and Additive Ratio Assessment (ARAS) methods for the STFN environment to calculate criteria weights and rank alternatives under fuzziness, respectively. The application of the proposed framework is demonstrated through a case study of ranking 10 sustainable energy sources based on a comprehensive set of sustainability indicators, including economic, technical, social, environmental, and political dimensions. Extensive robustness and sensitivity analyses validate the model’s effectiveness in managing complex, large-scale decision scenarios.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113535"},"PeriodicalIF":7.2,"publicationDate":"2025-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144632570","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":"Protein-peptide interaction region residues prediction using a generative sampling technique and ensemble deep learning-based models","authors":"Shima Shafiee ,&nbsp;Abdolhossein Fathi ,&nbsp;Ghazaleh Taherzadeh","doi":"10.1016/j.asoc.2025.113603","DOIUrl":"10.1016/j.asoc.2025.113603","url":null,"abstract":"<div><h3>Motivation</h3><div>Predicting protein-peptide interactions advances the understanding of drug design, protein biological functions, and cellular processes. Researchers have proposed various experimental and computational methods to identify interactions between proteins and peptides. However, traditional experimental approaches are laborious, time-consuming, and inefficient. Motivated by these challenges, a novel computational method is developed to detect protein-peptide interaction region residues from protein data, providing a complementary approach to experimental techniques.</div></div><div><h3>Method</h3><div>We designed a computational method for identifying protein-peptide interaction region residues, by incorporating a generative sampling technique with ensemble deep learning (DL) model using various features derived from protein sequences and structures. The proposed method relied on three pipelines: pre-processing, processing, and post-processing. The pre-processing pipeline converted the amino acid sequence into an image-like input representation to capture vital residue interactions. Also to overcome class imbalance challenge and non-binding over-predicting drawback, it employs a generative sampling technique for balancing the training data. Afterwards, to achieve more reliable prediction of protein-peptide interaction, a processing pipeline is designed that incorporates three independent DL sub-models. Subsequently, in the post-processing pipeline to obtain final prediction results, the outputs of ensemble DL modules are applied to three layers convolutional neural network.</div></div><div><h3>Results</h3><div>Compared to state-of-the-art sequence- and structure-based methods, the proposed method achieved the highest performance in F-measures (improved by 22.1 %), precision (improved by 3.9 %), and better balance between sensitivity and specificity. Eventually, our various experiments validated the effectiveness of the proposed method as a reliable computational assistant for predicting protein-peptide interaction region residues.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113603"},"PeriodicalIF":7.2,"publicationDate":"2025-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144614883","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":"An interpretable evolutionary broad learning system for damage identification in aircraft structures using Lamb waves","authors":"Gang Chen,&nbsp;Weihan Shao,&nbsp;Fudong Tang,&nbsp;Hu Sun","doi":"10.1016/j.asoc.2025.113577","DOIUrl":"10.1016/j.asoc.2025.113577","url":null,"abstract":"<div><div>Deep learning (DL) has gained significant attention in Lamb wave-based structural health monitoring (SHM). However, existing DL approaches for damage identification in aircraft structures require manually designed network architectures tailored to specific tasks, resulting in substantial computational overhead and hindering real-time monitoring applications. To overcome these limitations, this study proposes a novel damage identification method for aircraft structures based on Lamb waves and an interpretable evolutionary broad learning system (EBLS), which can automatically learn the complex nonlinear relationship between damage features in Lamb wave signals and structural health conditions. The proposed method incorporates a novel particle swarm optimization with square wave switching mechanism (SWSPSO) to systematically explore and optimize the complex hyperparameter space of the broad learning system (BLS). This intelligent optimization enables automated generation of task-specific BLS architectures for damage identification without manual intervention. The interpretability of EBLS is rigorously investigated through locally interpretable model-agnostic explanations (LIME), revealing physically meaningful correlations between critical feature contributions and fundamental Lamb wave propagation characteristics. Experimental validation employs a comprehensive Lamb wave dataset acquired through lead zirconate titanate (PZT) sensors mounted on aircraft structural components, encompassing diverse damage scenarios with varying locations and severity levels. Experimental results demonstrate that EBLS significantly outperforms traditional deep learning models, achieving over 0.95 accuracy in damage identification tasks while reducing computational efficiency by an order of magnitude and enhancing interpretability.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113577"},"PeriodicalIF":7.2,"publicationDate":"2025-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144614881","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":"AttenHideNet: A novel deep learning-based image steganography method using a lightweight U-net with soft attention","authors":"Younis M. Younis ,&nbsp;Ramadhan J. Mstafa ,&nbsp;Shamal AL-Dohuki","doi":"10.1016/j.asoc.2025.113583","DOIUrl":"10.1016/j.asoc.2025.113583","url":null,"abstract":"<div><div>Image-to-image steganography, embedding secret information within images while preserving visual quality, has become essential due to growing demands for secure and efficient digital communication. Traditional methods often struggle to achieve high embedding capacity without sacrificing imperceptibility. Recent advancements in deep learning have offered promising solutions by enabling more complex data embedding strategies. In this paper, we propose AttenHideNet, a novel deep learning-based steganography method leveraging a lightweight U-Net architecture (&lt;1.2 million parameters) combined with soft attention mechanisms. By utilizing the YUV color space instead of RGB, our method significantly improves embedding efficiency, capacity, and visual imperceptibility. AttenHideNet achieves an embedding capacity of up to 24 bits per pixel (bpp) while maintaining high visual quality. The soft attention mechanism dynamically identifies and prioritizes embedding in less perceptually sensitive image regions. Experimental results on benchmark datasets demonstrate that AttenHideNet achieves superior visual quality (PSNR up to 52.67 dB) compared to state-of-the-art methods, with low latency (18 ms/image) and minimal memory usage (4.11 MB), making it suitable for real-time applications. Despite these advantages, the method shows limited robustness under firm JPEG compression and geometric transformations, highlighting essential directions for future research.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113583"},"PeriodicalIF":7.2,"publicationDate":"2025-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144632615","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 dual-branch encoder context-aware fusion network for ultrasound image segmentation","authors":"Ning Yang ,&nbsp;Xinhui Jia ,&nbsp;Chunyu Hu ,&nbsp;Yuang Zhang ,&nbsp;Lei Lyu","doi":"10.1016/j.asoc.2025.113538","DOIUrl":"10.1016/j.asoc.2025.113538","url":null,"abstract":"<div><div>Accurate segmentation of lesion regions in ultrasound images remains a challenging task. Recent research has focused on integrating Transformers and CNNs to leverage their complementary strengths. However, most existing methods employ coarse fusion strategies that often lead to the loss of critical local details, such as lesion boundaries. Additionally, these methods fail to fully leverage the Transformer’s capability for global context modeling, thereby limiting their effectiveness in enhancing comprehensive feature representation. To this end, we propose a dual-branch encoder context-aware fusion network (DECF-Net) for automatic and robust lesion segmentation. The network introduces a parallel dual-branch encoder architecture to simultaneously capture global information and maintain sensitivity to the low-level context. We present a progressive feature extraction (PFE) module suitable for the Transformer branch, which aims to effectively suppress clutter noise and emphasize local features. In order to facilitate the interaction and fusion of feature information between different branches, we further introduce a supplementary feature fusion (SFF) module. In addition, we present a spatial channel attention bridge (SCAB) module to enhance the features of skip connections, which can extract multi-stage and multi-scale context information. Experimental results show that DECF-Net exhibits competitive segmentation performance in both qualitative and quantitative evaluation.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113538"},"PeriodicalIF":7.2,"publicationDate":"2025-07-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144632573","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":"Trend-Aware Mechanism for Metaheuristic Algorithms","authors":"Junbo Jacob Lian ,&nbsp;Kaichen Ouyang ,&nbsp;Rui Zhong ,&nbsp;Yujun Zhang ,&nbsp;Shipeng Luo ,&nbsp;Ling Ma ,&nbsp;Xincan Wu ,&nbsp;Huiling Chen","doi":"10.1016/j.asoc.2025.113505","DOIUrl":"10.1016/j.asoc.2025.113505","url":null,"abstract":"<div><div>In metaheuristic algorithms, historical search-position data often remain underutilized despite their potential to reveal valuable movement trends and promising search directions. To address this limitation, we propose the Trend-Aware Mechanism (TAM), which leverages historical position information to enhance the position updating process. TAM identifies the primary direction of movement by deriving a trend line from the population’s positions over the two most recent iterations. It evaluates candidate optimal positions by assessing the fitness of the K nearest points along this trend line. To effectively balance exploration and exploitation, TAM employs an adaptive covariance mechanism to generate high-dimensional random vectors, dynamically adjusting the update strategies. We integrate TAM with four prominent metaheuristic algorithms – PSO, SHADE, JaDE, and CMA-ES – and conduct an extensive parameter sensitivity analysis to ensure robustness. Comparative evaluations across five performance metrics demonstrate that TAM significantly improves search efficiency and consistently achieves superior results on standard benchmark functions. Moreover, TAM’s practical applicability is validated through real-world problems in engineering design, feature selection, and photovoltaic model parameter extraction. The open-source implementation of TAM will be publicly available at <span><span>https://github.com/junbolian/Trend-Aware-Mechanism</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113505"},"PeriodicalIF":7.2,"publicationDate":"2025-07-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144614880","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 matheuristic method for the automated guided vehicle scheduling problem with flexible charging and job release","authors":"Shanshan Zhou ,&nbsp;Zheng Wang ,&nbsp;Yantong Li ,&nbsp;Xin Wen","doi":"10.1016/j.asoc.2025.113531","DOIUrl":"10.1016/j.asoc.2025.113531","url":null,"abstract":"<div><div>Automated guided vehicles (AGVs) are vital in modern manufacturing for efficient material transport, requiring optimized scheduling to enhance system performance. This study introduces a new AGV scheduling problem that integrates task assignments, processing sequences, and flexible charging operations while accounting for job release times. The problem is NP-hard since it combines parallel machine scheduling and bin packing. We first formulate it as a mixed-integer linear program (MILP), which is strengthened by a set of valid inequalities. To address practical-sized instances, a matheuristic approach combining MILPs and an adaptive large neighborhood search is proposed. Key innovations include a three-step initialization algorithm for high-quality solutions, tailored destroy-and-repair operators, and a specialized evaluation function for efficient makespan approximation. Extensive experiments on 360 instances show that the matheuristic outperforms the commercial solver CPLEX in both solution quality and efficiency. Sensitivity analyses offer managerial insights into factors like charging strategies and energy management, supporting decision-making in AGV scheduling. A performance profit plot and a time-to-target plot are drawn to further validate the performance of the proposed matheuristic. The method also achieves 230 new best solutions for benchmark problems with slight modifications, demonstrating its versatility and effectiveness.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113531"},"PeriodicalIF":7.2,"publicationDate":"2025-07-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144654292","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 multispectral pansharpening method based on CNN-DI network with mixture of experts","authors":"Zhongyuan Guo ,&nbsp;Jia Lei ,&nbsp;Shihua Zhou ,&nbsp;Bin Wang ,&nbsp;Nikola K. Kasabov","doi":"10.1016/j.asoc.2025.113499","DOIUrl":"10.1016/j.asoc.2025.113499","url":null,"abstract":"<div><div>The process of fusing two complementary data, panchromatic and multispectral images, to create high-resolution multispectral (HRMS) images is known as pansharpening. Combining detail injection (DI) methods with convolutional neural networks (CNN) for improved HRMS image fusion quality is a research hotspot due to their interpretability and large-scale data processing capabilities, respectively. Nevertheless, the current hybrid models typically concatenate CNN and traditional techniques, limiting the ability to utilize the benefits of both approaches. This paper presents a new hybrid network, multispectral pansharpening method based on CNN-DI network with mixture of experts (CDN-MoE), using detail injection theory to design a deep learning framework. Specifically, we first create the mixture of detail inject experts network (MoDIE-Net) that mixes training pairs of full- and reduced-resolution images to enhance model generalization. Next, the adaptive correlation residual network (ACR-Net) is suggested to find the correlation between the spectral and spatial features of the source images. Finally, the global information injection network (GII-Net) is established to strengthen the accuracy of fusion results by integrating the context of input images. Additionally, to reduce the loss of spectral features during the upsampling process, the spectral reconstruction network (SR-Net) is proposed. We perform both qualitative and quantitative experiments on the GaoFen-2, IKONOS, and WorldView-2 datasets at various resolutions. Our approach has advantages over other SOTA pansharpening methods currently available in terms of visual effects and objective metrics.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113499"},"PeriodicalIF":7.2,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144614823","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}