Applied Soft Computing最新文献

{"title":"Multi-time Scale Augmented Neural ODEs graph neural for traffic flow prediction with elastic channel variation","authors":"Zihao Chu,&nbsp;Wenming Ma,&nbsp;Mingqi Li,&nbsp;Hao Chen","doi":"10.1016/j.asoc.2025.113513","DOIUrl":"10.1016/j.asoc.2025.113513","url":null,"abstract":"<div><div>Traffic flow prediction is a critical task in traffic management due to the complex and dynamic spatio-temporal correlations inherent in traffic data. While existing methods often employ graph convolutional networks (GNNs) and temporal extraction modules to model spatial and temporal dependencies, respectively, deep GNNs suffer from oversmoothing, which impairs their ability to capture long-term spatial relationships. Augmented Neural Ordinary Differential Equations (ANODEs) offer a solution to this issue by enabling deeper models without oversmoothing, but they struggle with the complexity and variability of traffic data, leading to poor prediction performance. In this study, we propose the Multi-time Scale Augmented Neural ODEs Graph Neural Network (MTEC-AODE) for Traffic Flow Prediction. To address the challenges of complex information processing, we introduce the Elastic Channel Variation strategy, which adjusts the number of channels dynamically. Furthermore, we construct a traffic semantic neighborhood matrix using a Gaussian kernel similarity matrix, which captures semantic relationships across regions, aiding in the construction of a global dynamic traffic model. To handle the variability of traffic data, we develop the Multi-time Scale Augmented Neural ODEs Solver, allowing the model to adapt to different time scales and respond to dynamic changes in traffic patterns. We evaluate our model on several real-world traffic datasets, achieving Mean Absolute Errors (MAE) of 2.01, 3.52, 2.96, 3.20, 15.59, 19.75, 22.14, and 16.22, and Mean Absolute Percentage Errors (MAPE) of 4.48%, 10.17%, 7.22%, 7.66%, 15.21%, 13.98%, 9.72%, and 10.2%. Experimental results show that our method outperformed state-of-the-art benchmarks.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113513"},"PeriodicalIF":7.2,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144633813","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":"GAN-based approach for data imputation and handling class imbalance using one class ensemble","authors":"Pranita Baro,&nbsp;Malaya Dutta Borah","doi":"10.1016/j.asoc.2025.113540","DOIUrl":"10.1016/j.asoc.2025.113540","url":null,"abstract":"<div><div>Class imbalance in real-world datasets is a significant issue that results in bias in the machine learning model and may result in incorrect predictions. In this paper, a GAN-based Multiple Imputation One-Class Ensemble (GMI-OCE) is presented for imbalanced classification in scenarios with missing values in the dataset. The approach uses a hybrid OCC ensemble, incorporating a GAN architecture for imputing missing values and boosting the number of minority class instances without modifying the observed values directly. A two-step bootstrap aggregation is applied using a novel weighting algorithm that considers the accuracy of individual classifiers and their performance on synthetic data. The approach is evaluated on various imbalanced datasets and compared against seven baseline methods. The results indicate that GMI-OCE outperforms in most of the datasets compared to other methods based on various evaluation metrics.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113540"},"PeriodicalIF":7.2,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144632614","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":"Texture-aware 3D Gaussian Splatting for sparse view reconstructions","authors":"Xinyuan Hu ,&nbsp;Changyue Shi ,&nbsp;Chuxiao Yang ,&nbsp;Minghao Chen ,&nbsp;Xiaoling Gu ,&nbsp;Jiajun Ding ,&nbsp;Jifa He ,&nbsp;Jianping Fan","doi":"10.1016/j.asoc.2025.113530","DOIUrl":"10.1016/j.asoc.2025.113530","url":null,"abstract":"<div><div>Recently, 3D Gaussian Splatting (3DGS) has achieved high rendering quality in Novel View Synthesis (NVS). However, as the number of input views decreases, 3DGS fails to recover the details of the captured 3D scene due to insufficient constraints. We find that the difficulty of Gaussian primitives to concentrate on texture-rich areas leads to this reconstruction degradation. To this end, we propose TA-GS, a texture-aware framework for sparse-view NVS tasks. Specifically, TA-GS introduces a Texture-Based Gaussian Migration strategy, which detects low-opacity Gaussian primitives and migrates them to texture-rich regions, improving the fidelity of texture representation. Additionally, we utilize the texture of depth maps and introduce a Depth Texture Alignment method to constrain the geometric structures. To prevent overfitting to sparse input views, TA-GS employs Phantom View Regularization to enrich texture information from additional phantom views. Extensive experiments demonstrate that our approach outperforms previous methods across a variety of datasets, including LLFF, Mip-NeRF360, DTU, and Blender.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113530"},"PeriodicalIF":7.2,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144597272","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 hybrid genetic tabu search algorithm for metro crew scheduling based on a space-time-state network","authors":"Feng Xue ,&nbsp;Peng Liang ,&nbsp;Ying Yang ,&nbsp;Jincheng Wang","doi":"10.1016/j.asoc.2025.113574","DOIUrl":"10.1016/j.asoc.2025.113574","url":null,"abstract":"<div><div>The crew scheduling problem is highly important for the operation and management of urban rail transit. It is essential to reasonably design an approach for optimizing the crew schedule within the constraints of a provided train diagram so that the schedule is highly versatile and can meet the actual operational demand. Additionally, better results can be achieved by using an optimization method, which can reduce operating costs and satisfy crew members’ working preferences to the greatest extent possible to achieve a more rational distribution of tasks. Unlike traditional space-time networks that merely describe spatiotemporal movement trajectories, this study innovatively introduces state attributes to ensure solution feasibility during search. Using these attributes, we establish a space-time-state network for crew scheduling modeling. This model has the objective of reducing task connection time and personnel costs. To solve the provided model, a hybrid genetic tabu search (HGTS) algorithm is created by considering the distinctive characteristics of two methods: tabu search (TS) and genetic algorithm (GA), where TS handles local search and GA performs global optimization. The HGTS algorithm can efficiently address the complex metro crew scheduling problem and obtain an improved crew scheduling plan. The proposed method is validated against data from Chengdu Metro Line 5. Results demonstrate that our constructed methodology can effectively reduce the personnel costs and connection time of crew scheduling over the manual scheduling plan: a total of 148 crew duties were obtained, with an optimization rate of 10.30 % and a total connection time of 198 h 44 min 49 s, with an optimization rate of 7.71 %. Furthermore, the proposed method has a higher computational speed and enhanced stability than the shortest-path faster algorithm based on the greedy approach (G-SPFA) method, especially for large-scale data. Additionally, as a hybrid algorithm, HGTS delivers superior solutions compared to standalone GA and TS. This advantage is evidenced by key metrics: HGTS achieved a total duty duration of 725 h 31 min 51 s versus GA's 778 h 38 min 10 s and TS's 749 h 11 min 31 s, while also demonstrating tighter crew efficiency with standard deviations of 0.067, 0.077, and 0.085 for HGTS, GA, and TS respectively.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113574"},"PeriodicalIF":7.2,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144632613","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":"Multi-objective multi-optima ensemble binary optimization algorithm for identifying optimal set of features for ECG-based identification","authors":"Mamata Pandey,&nbsp;Anup Kumar Keshri","doi":"10.1016/j.asoc.2025.113556","DOIUrl":"10.1016/j.asoc.2025.113556","url":null,"abstract":"<div><div>Reducing the number of input features for a machine learning model decreases its complexity and computation time. However, it is crucial to choose the best set of features without compromising the model's performance. There could be several subsets of features with optimal behavior. Evolutionary algorithms are great for feature optimization. However, different evolutionary algorithms may produce different solutions, and their performance is influenced by the size of the data and the types of features. To address these issues, three popular algorithms, Genetic Algorithm (GA), Particle Swarm Optimization (PSO), and Binary Differential Evolution (BDE) have been adapted to accommodate multiple populations for achieving multiple optima. The BDE algorithm applied here is a novel variant with modified mutation and crossover operators. Then they are combined to create a novel 'Multi-Objective Multi-Optima Ensemble Binary Optimization Algorithm. The algorithm has been tested on 71 fiducial ECG features including temporal, amplitude, distance, slope, angular, and HRV features for ECG-based identification. These 71 features can identify individuals using the SVM classifier with 98 % accuracy. With 71 features, there could be a maximum of <span><math><msup><mrow><mn>2</mn></mrow><mrow><mn>71</mn></mrow></msup></math></span> subsets. The optimization objective is to find all feature subsets that maximize classifier accuracy while minimizing the number of features. The ensemble optimizer has found 190 unique optimized subsets. These subsets have been analyzed to identify critical features for identification. The most optimal subset with the minimum number of features and maximum accuracy has been identified. The practical implementation of an ECG-based identification system requires an efficient system that can process incoming signals, extract features from the signal, and identify individuals in the shortest time possible. To speed up the processing of the input signal, a novel DFA-based algorithm has been proposed to identify fiducial points P, Q, R, S, and T from an ECG signal. The proposed algorithm applies to both recorded and live ECG signals.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113556"},"PeriodicalIF":7.2,"publicationDate":"2025-07-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144605094","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":"Hexagonal-hogel holographic stereograms based on neural graphics primitives","authors":"Xingpeng Yan ,&nbsp;Hebin Chang ,&nbsp;Yanan Zhang ,&nbsp;Hairong Hu ,&nbsp;Tao Jing ,&nbsp;Xiaoyu Jiang ,&nbsp;Weifeng Wang","doi":"10.1016/j.asoc.2025.113500","DOIUrl":"10.1016/j.asoc.2025.113500","url":null,"abstract":"<div><div>In this work, a method for generating full-parallax synthetic holographic stereograms using enhanced hexagonal hogels and neural graphics primitives is proposed and implemented. The diffraction effect and defocus aberration are introduced to analyze and compare the reconstruction performance between the proposed hexagonal pupil and the traditional square pupil, indicating that the hexagonal pupil offers better performance for floating-out holographic stereograms. Then, the extraction method for effective perspective image segments, along with the formation strategy for the synthetic effective perspective image (SEPI), is analyzed in detail. Torch-ngp is adopted to generate the SEPI from a limited number of perspective images of the 3D scene, where the direct extraction and rendering mechanism significantly improves the generation efficiency of the SEPI. Experimental results demonstrate that the proposed method can efficiently fabricate full-parallax synthetic holographic stereograms with enhanced reconstruction quality.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113500"},"PeriodicalIF":7.2,"publicationDate":"2025-07-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144597273","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":"Quantitative risk assessment of cruise ship turbochargers using type-2 fuzzy-FMECA and dynamic Bayesian network approach","authors":"Shoaib Ahmed ,&nbsp;Tie Li ,&nbsp;Xinyi Zhou ,&nbsp;Shuai Huang ,&nbsp;Run Chen","doi":"10.1016/j.asoc.2025.113568","DOIUrl":"10.1016/j.asoc.2025.113568","url":null,"abstract":"<div><div>Marine propulsion systems, both traditional and modern electric, face significant risks associated with turbocharger and lubrication system failures. The failure outcomes can be severe, with accidents leading to deaths onboard, damage to machinery causing operational disruption, environmental pollution, and financial losses. While traditional Failure mode, effect, and criticality analysis (FMECA) methods excel in identifying system failures, their reliance on single-point estimates for severity, occurrence, and non-detection may prove limiting. Moreover, employing multiple experts in assessments can introduce biases. Integrating type-2 Fuzzy-FMECA with the linear opinion pool method is a robust approach to address these limitations. Leveraging the collective expertise of multiple experts, this framework enhances risk assessment comprehensiveness and accuracy. Focusing on the Carnival Freedom cruise ship incident near the Cayman Islands in October 2019, this study aims to develop a comprehensive risk assessment framework for assessing marine engine turbocharger and lubrication system risks. This study showed a strong positive correlation of 0.99 between the traditional risk prioritization number and the proposed type-2 fuzzy logic method, demonstrating its validity as a reliable alternative. This method effectively identified critical machinery failures, such as low-pressure switch and pressure control valve malfunctions, consistently aligning with the results of Traditional methods. It combines a dynamic Bayesian network for handling uncertainty with an interval type-2 fuzzy expert system and a bow-tie model. This framework enables both qualitative hazard identification and quantitative risk assessment. This risk analysis approach holds practical applicability in real-world scenarios, and its outcomes significantly provide actionable insights to mitigate and eliminate potential failures. Ultimately, it reduces the risk and improves the safety and reliability of cruise ship operations, providing a tangible solution to a pressing problem in the field.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113568"},"PeriodicalIF":7.2,"publicationDate":"2025-07-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144633814","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":"Multi-branch global Transformer‐assisted network for fault diagnosis","authors":"Xiaorui Shao ,&nbsp;Chang-Soo Kim","doi":"10.1016/j.asoc.2025.113572","DOIUrl":"10.1016/j.asoc.2025.113572","url":null,"abstract":"<div><div>Fault Diagnosis (FD) is critical in smart manufacturing, enabling predictive maintenance, reducing operational costs, and enhancing system reliability. To deal with this task more accurately, this paper proposes a generative, effective, and novel framework, a multi-branch global Transformer-assisted network (MBGTNet), for accurate FD. First, a multi-branch global-wide one-dimension convolution operation (MBG-WideConv1D) is proposed to obtain global features in different views. Meanwhile, a Transformer assist scheme (TAS) is designed to leverage the Transformer's global feature extraction capacity. The features extracted by the Transformer are fused with those extracted with MBG-WideConv1D by minimizing their pairwise correlation alignment (CORAL) distances. Benefiting from the well-designed MBG-WideConv1D and TAS, the global features hidden in the raw signals are fully extracted from multiple viewpoints. Each branch of global features is then fed into a one-dimension convolutional neural network (1DCNN) to extract local features in a multi-supervised scheme (MSS) that helps each branch learn thoroughly. Furthermore, the proposed method employs a local feature correlation enhancement scheme (LFCS) to reduce distribution differences and increase robustness among the local features of each branch. As a result, the final features used for FD are a fusion of multi-view global and local features with strong robustness, enabling accurate FD in noisy environments. Comparative experiments on four datasets, including CWRU, MFPT, SU Bearing, and SU Gear, validate the proposed method's effectiveness, achieving over 99.6 % accuracy across four datasets. Moreover, the TAS and LFCS's generalities have been demonstrated on two 1DCNNs and hybrid CNN-LSTMs with four subsets. Also, the effectiveness of each component in the proposed framework has been thoroughly analyzed.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113572"},"PeriodicalIF":7.2,"publicationDate":"2025-07-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144632571","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":"EEG-based emotion detection using Long Short-Term Memory Network and reinforcement learning for enhanced feature selection","authors":"Junxiu Liu ,&nbsp;Guopei Wu ,&nbsp;Qiang Fu ,&nbsp;Yuling Luo ,&nbsp;Su Yang ,&nbsp;Senhui Qiu ,&nbsp;Yi Cao ,&nbsp;Wei Li","doi":"10.1016/j.asoc.2025.113512","DOIUrl":"10.1016/j.asoc.2025.113512","url":null,"abstract":"<div><div>Brain–computer interface systems can recognize users’ emotions through electroencephalography (EEG). EEG-based human emotion recognition is an emerging field that is gaining significant traction within the realm of brain–computer interfaces. However, due to the complexity and diversity inherent in EEG signals, emotion recognition remains a challenge in pattern recognition. The critical task of selecting salient features from EEG and achieving high recognition accuracy warrants further exploration. In this paper, a hybrid emotion detection system is proposed by incorporating the reinforcement learning mechanism into a deep learning framework. Reinforcement learning is used to recursively select informative features, while a Long Short-Term Memory Network (LSTM) and a deep neural network are employed for enhanced feature selection and emotion recognition. Specifically, the LSTM, based on input features, determines and generates the current state, thereby aiding the policy model in making action decisions. This process successively retains or removes features to improve emotion recognition in the next state. The neural net-based policy model generates the policy actions based on the current state and the corresponding reward signal from the classification result, to control the feature selections for the subsequent states. A public EEG emotion dataset of SEED is used in the experiments. Results show that the proposed network model is effective in feature selections and emotion classifications, which reduces feature dimensions by 11.3% on average, and achieves a higher recognition accuracy of 92.65% compared to other approaches. The proposed system can use the current state info for prediction and adaptive feature selection, which can accommodate the data pattern differences of individual participants and leverage the model for a good performance.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113512"},"PeriodicalIF":7.2,"publicationDate":"2025-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144614433","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":"Imbalanced Broad Learning System with label relaxation and sample weight adaptation","authors":"Yanting Li ,&nbsp;Yusha Wang ,&nbsp;Junwei Jin ,&nbsp;Weiwei Zhang ,&nbsp;Hongwei Tao ,&nbsp;Huaiguang Wu ,&nbsp;C.L. Philip Chen","doi":"10.1016/j.asoc.2025.113543","DOIUrl":"10.1016/j.asoc.2025.113543","url":null,"abstract":"<div><div>The Broad Learning System (BLS), as a lightweight network architecture, has been extensively applied to various classification and regression tasks. However, BLS and its variants remain suboptimal for addressing imbalanced classification problems. These models often pay little attention to the quality of original features. Their supervision mechanisms typically rely on strict binary label matrices, which impose limitations on approximation and fail to align with the underlying data distribution. Additionally, they generally do not differentiate the contributions of majority and minority classes, leading to a bias towards majority classes in predictions. In this paper, we propose a novel imbalanced BLS framework that integrates label relaxation and sample weight adaptation to address challenges in imbalanced classification tasks. First, genetic programming is employed to optimize the original features, improving data representation capability. Then, a latent label space is constructed based on pairwise label relationships, which serves to achieve flexible label relaxation. Furthermore, a dynamic weighting mechanism is proposed based on intra-class and inter-class distributions to balance the influence of majority and minority classes. Extensive experiments conducted on 30 benchmark datasets demonstrate that the proposed method significantly outperforms various state-of-the-art approaches, with average G-mean and AUC scores of 89.6% and 90.0%, respectively. These results validate the effectiveness and superiority of the proposed model in addressing imbalanced classification tasks.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"182 ","pages":"Article 113543"},"PeriodicalIF":7.2,"publicationDate":"2025-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144589195","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}