Applied Soft Computing最新文献_第10页

Automated conceptual design of mechanisms based on Thompson Sampling and Monte Carlo Tree Search

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Applied Soft Computing

Pub Date : 2025-02-01 DOI: 10.1016/j.asoc.2024.112659

Jiangmin Mao , Yingdan Zhu , Gang Chen , Chun Yan , Wuxiang Zhang

Conceptual design of mechanisms is a crucial part of achieving product innovation as mechanisms perform the transmission and transformation of specific motions in the machine. However, existing approaches for automated synthesis of mechanisms are either inefficient or prone to a loss of optimal solutions. To fill this gap, a systematic online decision-making method using Thompson Sampling (TS) based Monte Carlo Tree Search (MCTS) for automated conceptual design of mechanisms is proposed. The functional transformation relationships between inputs and outputs of the intended mechanism system are used to determine combinatorial patterns. Then, a functional representation model is constructed based on the combination rules of motion features and the inference relationships of function elements to represent a range of primitive mechanisms as fundamental building blocks. Finally, the optimal action selection strategy based on TS is applied into MCTS to develop Dirichlet based Monte Carlo Tree Search (D-MCTS) algorithm for searching mechanism building blocks. In addition, the conceptual design of the beat-up mechanism as well as the stitching and feeding mechanism are conducted to validate the feasibility of the proposed approach. Compared with specialized heuristics, D-MCTS achieves higher efficiency in finding the best combination of mechanism building blocks. Compared with other common algorithms, D-MCTS can always avoid the local optima trap to find the global optimal solution without any necessary hyper-parameter tuning. The proposed method exhibits a more balanced performance in exploration and exploitation, which provides better solutions for mechanism synthesis of given requirements.

{"title":"Automated conceptual design of mechanisms based on Thompson Sampling and Monte Carlo Tree Search","authors":"Jiangmin Mao , Yingdan Zhu , Gang Chen , Chun Yan , Wuxiang Zhang","doi":"10.1016/j.asoc.2024.112659","DOIUrl":"10.1016/j.asoc.2024.112659","url":null,"abstract":"<div><div>Conceptual design of mechanisms is a crucial part of achieving product innovation as mechanisms perform the transmission and transformation of specific motions in the machine. However, existing approaches for automated synthesis of mechanisms are either inefficient or prone to a loss of optimal solutions. To fill this gap, a systematic online decision-making method using Thompson Sampling (TS) based Monte Carlo Tree Search (MCTS) for automated conceptual design of mechanisms is proposed. The functional transformation relationships between inputs and outputs of the intended mechanism system are used to determine combinatorial patterns. Then, a functional representation model is constructed based on the combination rules of motion features and the inference relationships of function elements to represent a range of primitive mechanisms as fundamental building blocks. Finally, the optimal action selection strategy based on TS is applied into MCTS to develop Dirichlet based Monte Carlo Tree Search (D-MCTS) algorithm for searching mechanism building blocks. In addition, the conceptual design of the beat-up mechanism as well as the stitching and feeding mechanism are conducted to validate the feasibility of the proposed approach. Compared with specialized heuristics, D-MCTS achieves higher efficiency in finding the best combination of mechanism building blocks. Compared with other common algorithms, D-MCTS can always avoid the local optima trap to find the global optimal solution without any necessary hyper-parameter tuning. The proposed method exhibits a more balanced performance in exploration and exploitation, which provides better solutions for mechanism synthesis of given requirements.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"170 ","pages":"Article 112659"},"PeriodicalIF":7.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143213402","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}

引用次数: 0

Conditional generative adversarial networks for the generation of strong ground motion parameters using KiK-net ground motion records

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Applied Soft Computing

Pub Date : 2025-02-01 DOI: 10.1016/j.asoc.2025.112730

Zhenning Ba , Jingxuan Zhao , Fangbo Wang , Linghui Lyu

Prediction of ground motion parameters is important for seismic risk analysis and the seismic design of structures. The limited availability of strong ground motion records (peak ground acceleration (PGA) > 100 gal) limits the application of machine learning techniques in this domain. In this study, we utilized the KiK-net database to identify three site parameters—V_S30, V_S-Zhole, and borehole depth—corresponding to the peak underground acceleration (PUA) on the downhole and spectral acceleration (Sa) across 15 different periods as input features for the machine learning model. The output parameters were PGA and Sa across 15 different periods. Pairs of strong ground motion records (PGA > 100 gal) are extracted from the original KiK-net database and processed using a conditional generative adversarial network (cGAN) model to synthesize ground motion parameters, then the original KiK-net database is supplemented. Regression prediction tests were performed on both the original and augmented databases and evaluated using three regression metrics: root mean square error (RMSE), mean absolute error (MAE), and R². The findings indicate that the model trained with synthetic data generated by the cGAN enhances the prediction accuracy of ground motion parameters. A comparative analysis of the ground motion amplification coefficients between station records and cGAN-generated data across various spectral periods was performed. The amplification coefficients across different spectral periods aligned with the ground-motion records, thereby validating the reliability of the cGAN-generated synthetic ground-motion data from a seismological perspective. Furthermore, the ground motion parameters produced by the cGAN represent a valuable resource for future machine learning investigations.

{"title":"Conditional generative adversarial networks for the generation of strong ground motion parameters using KiK-net ground motion records","authors":"Zhenning Ba , Jingxuan Zhao , Fangbo Wang , Linghui Lyu","doi":"10.1016/j.asoc.2025.112730","DOIUrl":"10.1016/j.asoc.2025.112730","url":null,"abstract":"<div><div>Prediction of ground motion parameters is important for seismic risk analysis and the seismic design of structures. The limited availability of strong ground motion records (peak ground acceleration (PGA) > 100 gal) limits the application of machine learning techniques in this domain. In this study, we utilized the KiK-net database to identify three site parameters—<em>V</em><sub>S30</sub>, <em>V</em><sub>S-Zhole</sub>, and borehole depth—corresponding to the peak underground acceleration (PUA) on the downhole and spectral acceleration (Sa) across 15 different periods as input features for the machine learning model. The output parameters were PGA and Sa across 15 different periods. Pairs of strong ground motion records (PGA > 100 gal) are extracted from the original KiK-net database and processed using a conditional generative adversarial network (cGAN) model to synthesize ground motion parameters, then the original KiK-net database is supplemented. Regression prediction tests were performed on both the original and augmented databases and evaluated using three regression metrics: root mean square error (RMSE), mean absolute error (MAE), and R<sup>2</sup>. The findings indicate that the model trained with synthetic data generated by the cGAN enhances the prediction accuracy of ground motion parameters. A comparative analysis of the ground motion amplification coefficients between station records and cGAN-generated data across various spectral periods was performed. The amplification coefficients across different spectral periods aligned with the ground-motion records, thereby validating the reliability of the cGAN-generated synthetic ground-motion data from a seismological perspective. Furthermore, the ground motion parameters produced by the cGAN represent a valuable resource for future machine learning investigations.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"170 ","pages":"Article 112730"},"PeriodicalIF":7.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143213473","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}

引用次数: 0

A priority rule heuristic for the multi-skilled resource-constrained project scheduling problem

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Applied Soft Computing

Pub Date : 2025-02-01 DOI: 10.1016/j.asoc.2025.112776

Guillaume Vermeire , Mario Vanhoucke

This research presents a priority rule heuristic approach for the multi-skilled resource-constrained project scheduling problem. The approach is based on a parallel schedule generation scheme which includes a new resource assignment procedure. The scheme combines three types of priority rules in order to schedule activities and assign resources to the skill requirements of these activities. In computational experiments, skill- and resource rule combinations are evaluated and selected based on two metrics using a Pareto Front approach. These rule combinations are then integrated with various activity priority rules after which their solution quality is evaluated. The heuristic approach and the selected rules are then employed to solve all project instances of the MSLIB dataset. It is shown that, on average, the presented approach is able to obtain solutions with a comparable quality to the solution quality of a meta-heuristic procedure from literature. Additionally, new best known solutions are obtained for the MSLIB dataset. The practical applicability of the heuristic is validated by solving empirical project instances.

{"title":"A priority rule heuristic for the multi-skilled resource-constrained project scheduling problem","authors":"Guillaume Vermeire , Mario Vanhoucke","doi":"10.1016/j.asoc.2025.112776","DOIUrl":"10.1016/j.asoc.2025.112776","url":null,"abstract":"<div><div>This research presents a priority rule heuristic approach for the multi-skilled resource-constrained project scheduling problem. The approach is based on a parallel schedule generation scheme which includes a new resource assignment procedure. The scheme combines three types of priority rules in order to schedule activities and assign resources to the skill requirements of these activities. In computational experiments, skill- and resource rule combinations are evaluated and selected based on two metrics using a Pareto Front approach. These rule combinations are then integrated with various activity priority rules after which their solution quality is evaluated. The heuristic approach and the selected rules are then employed to solve all project instances of the MSLIB dataset. It is shown that, on average, the presented approach is able to obtain solutions with a comparable quality to the solution quality of a meta-heuristic procedure from literature. Additionally, new best known solutions are obtained for the MSLIB dataset. The practical applicability of the heuristic is validated by solving empirical project instances.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"171 ","pages":"Article 112776"},"PeriodicalIF":7.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143355954","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}

引用次数: 0

Constraint-handling techniques for reusable launch vehicle reentry trajectory optimization using marine predator whale optimizer

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Applied Soft Computing

Pub Date : 2025-02-01 DOI: 10.1016/j.asoc.2024.112637

Ya Su , Yi Liu

Reentry trajectory optimization for reusable launch vehicles (RLVs) is a class of optimal control problems with multiple highly nonlinear constraints. Nature-inspired algorithms (NIAs), which can somehow reduce the reliance on initial points, function differentiability, and convexity, with gradient-free nature and ease of implementation, have been actively applied in RLV reentry trajectory optimization problems. As NIAs are primarily designed for unconstrained optimization, constraint-handling techniques (CHTs), which play a crucial role in addressing RLV trajectory optimization issues and significantly impact the overall quality of the solutions, are necessary to guide the search towards feasible regions. However, the existing literature has not yet, or at least not systematically, investigated how well the current CHTs perform. Additionally, an in-depth analysis of parametric approaches and a performance evaluation framework is not yet available. To bridge this gap, this research constructs a benchmark model based on Space Shuttle reentry scenarios, investigates the effects of collocation type and interpolation method, and compares the performance of eight CHTs. An improved marine predator whale optimization algorithm is developed as a direct search engine, with results analyzed using the Wilcoxon signed rank and the Friedman tests. The results show that the ε-constrained technique and multi-objective-based CHTs with the algorithm, are somewhat superior in overall performance, and can produce relatively high-quality solutions over other competitors, while the optimization framework facilitates algorithm integration and CHTs for RLV reentry trajectory optimization problems.

{"title":"Constraint-handling techniques for reusable launch vehicle reentry trajectory optimization using marine predator whale optimizer","authors":"Ya Su , Yi Liu","doi":"10.1016/j.asoc.2024.112637","DOIUrl":"10.1016/j.asoc.2024.112637","url":null,"abstract":"<div><div>Reentry trajectory optimization for reusable launch vehicles (RLVs) is a class of optimal control problems with multiple highly nonlinear constraints. Nature-inspired algorithms (NIAs), which can somehow reduce the reliance on initial points, function differentiability, and convexity, with gradient-free nature and ease of implementation, have been actively applied in RLV reentry trajectory optimization problems. As NIAs are primarily designed for unconstrained optimization, constraint-handling techniques (CHTs), which play a crucial role in addressing RLV trajectory optimization issues and significantly impact the overall quality of the solutions, are necessary to guide the search towards feasible regions. However, the existing literature has not yet, or at least not systematically, investigated how well the current CHTs perform. Additionally, an in-depth analysis of parametric approaches and a performance evaluation framework is not yet available. To bridge this gap, this research constructs a benchmark model based on Space Shuttle reentry scenarios, investigates the effects of collocation type and interpolation method, and compares the performance of eight CHTs. An improved marine predator whale optimization algorithm is developed as a direct search engine, with results analyzed using the Wilcoxon signed rank and the Friedman tests. The results show that the <em>ε</em>-constrained technique and multi-objective-based CHTs with the algorithm, are somewhat superior in overall performance, and can produce relatively high-quality solutions over other competitors, while the optimization framework facilitates algorithm integration and CHTs for RLV reentry trajectory optimization problems.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"170 ","pages":"Article 112637"},"PeriodicalIF":7.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143212935","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}

引用次数: 0

Dynamic Q&A multi-label classification based on adaptive multi-scale feature extraction

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Applied Soft Computing

Pub Date : 2025-02-01 DOI: 10.1016/j.asoc.2025.112740

Ying Li, Ming Li, Xiaoyi Zhang, Jin Ding

In community question answering (CQA), questioners use labels for question and answer (Q&A) classification when asking questions. Since the answerers do not have the same understanding and perspective of the question, the original labels cannot accurately reflect the Q&A categories with constantly given answers. Therefore, this paper proposes a dynamic Q&A multi-label classification approach based on adaptive multi-scale feature extraction. First, global and local semantic features of Q&As are extracted based on bidirectional long short-term memory network and convolutional neural network models, respectively. Second, the label features extraction and fusion method is proposed. The semantic features of the labels are extracted, the label structure graph based on horizontal and vertical dependencies is constructed, and the label structure and semantic features are fused using the graph attention network integrating multi-head self-attention mechanism. Afterward, the label-aware local features of Q&As are constructed using the attention mechanism and fused with global features of Q&A using the multi-head self-attention, thereby multi-scale fusion classification features of Q&A are established. Then, to adaptively extract the core multi-scale fusion features, a multi-objective feature selection model is established and an improved binary multi-objective Sinh Cosh optimizer algorithm is proposed to solve the model. Finally, a classification prediction layer based on a multilayer perceptron is constructed to obtain the multi-label classification results of Q&A documents. The experimental results based on real Q&A data show the superior performance of the proposed method and validate the effectiveness of the proposed four modules.

{"title":"Dynamic Q&A multi-label classification based on adaptive multi-scale feature extraction","authors":"Ying Li, Ming Li, Xiaoyi Zhang, Jin Ding","doi":"10.1016/j.asoc.2025.112740","DOIUrl":"10.1016/j.asoc.2025.112740","url":null,"abstract":"<div><div>In community question answering (CQA), questioners use labels for question and answer (Q&A) classification when asking questions. Since the answerers do not have the same understanding and perspective of the question, the original labels cannot accurately reflect the Q&A categories with constantly given answers. Therefore, this paper proposes a dynamic Q&A multi-label classification approach based on adaptive multi-scale feature extraction. First, global and local semantic features of Q&As are extracted based on bidirectional long short-term memory network and convolutional neural network models, respectively. Second, the label features extraction and fusion method is proposed. The semantic features of the labels are extracted, the label structure graph based on horizontal and vertical dependencies is constructed, and the label structure and semantic features are fused using the graph attention network integrating multi-head self-attention mechanism. Afterward, the label-aware local features of Q&As are constructed using the attention mechanism and fused with global features of Q&A using the multi-head self-attention, thereby multi-scale fusion classification features of Q&A are established. Then, to adaptively extract the core multi-scale fusion features, a multi-objective feature selection model is established and an improved binary multi-objective Sinh Cosh optimizer algorithm is proposed to solve the model. Finally, a classification prediction layer based on a multilayer perceptron is constructed to obtain the multi-label classification results of Q&A documents. The experimental results based on real Q&A data show the superior performance of the proposed method and validate the effectiveness of the proposed four modules.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"170 ","pages":"Article 112740"},"PeriodicalIF":7.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143212942","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}

引用次数: 0

Pigmented skin disease classification via deep learning with an attention mechanism

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Applied Soft Computing

Pub Date : 2025-02-01 DOI: 10.1016/j.asoc.2024.112571

Jinbo Chen , Qian Jiang , Zhuang Ai , Qihao Wei , Sha Xu , Baohai Hao , Yaping Lu , Xuan Huang , Liuqing Chen

Pigmented skin disease is common; doctors need to observe and analyze pigmented skin disease images for diagnostic purposes. However, due to regional differences, diagnoses are subjective, resulting in high misdiagnosis rates. Therefore, this paper proposes a deep learning-based method for classifying pigmented skin disease images named the skin-global attention block (Skin-GAB). This method automatically classifies pigmented skin disease images through a system architecture that includes image augmentation, image segmentation, cluster analysis, segmented and original image classification, and network fusion. Additionally, this paper utilizes the GAB attention mechanism to encode the height, width, and channel of the feature maps, allowing the model to automatically learn crucial features from pigmented skin disease images and focus its attention on task-relevant information, thereby capturing disparities in input feature maps and further enhancing the model’s performance. The experimental results show that the proposed method performs well in terms of accuracy and practicality. Compared to using Xception as the classification network and the convolutional block attention module (CBAM) as the attention mechanism on the HAM10000 dataset, the system architecture proposed in this paper provides an improvement in accuracy of 2.89%. Therefore, this method will provide more accurate and efficient technical support for medical fields such as pigmented skin disease diagnosis and treatment.

{"title":"Pigmented skin disease classification via deep learning with an attention mechanism","authors":"Jinbo Chen , Qian Jiang , Zhuang Ai , Qihao Wei , Sha Xu , Baohai Hao , Yaping Lu , Xuan Huang , Liuqing Chen","doi":"10.1016/j.asoc.2024.112571","DOIUrl":"10.1016/j.asoc.2024.112571","url":null,"abstract":"<div><div>Pigmented skin disease is common; doctors need to observe and analyze pigmented skin disease images for diagnostic purposes. However, due to regional differences, diagnoses are subjective, resulting in high misdiagnosis rates. Therefore, this paper proposes a deep learning-based method for classifying pigmented skin disease images named the skin-global attention block (Skin-GAB). This method automatically classifies pigmented skin disease images through a system architecture that includes image augmentation, image segmentation, cluster analysis, segmented and original image classification, and network fusion. Additionally, this paper utilizes the GAB attention mechanism to encode the height, width, and channel of the feature maps, allowing the model to automatically learn crucial features from pigmented skin disease images and focus its attention on task-relevant information, thereby capturing disparities in input feature maps and further enhancing the model’s performance. The experimental results show that the proposed method performs well in terms of accuracy and practicality. Compared to using Xception as the classification network and the convolutional block attention module (CBAM) as the attention mechanism on the HAM10000 dataset, the system architecture proposed in this paper provides an improvement in accuracy of 2.89%. Therefore, this method will provide more accurate and efficient technical support for medical fields such as pigmented skin disease diagnosis and treatment.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"170 ","pages":"Article 112571"},"PeriodicalIF":7.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143213075","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}

引用次数: 0

A multi-dimensional co-evolutionary algorithm for multi-objective resource-constrained flexible flowshop with robotic transportation

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Applied Soft Computing

Pub Date : 2025-02-01 DOI: 10.1016/j.asoc.2024.112689

Jia-ke Li , Rong-hao Li , Jun-qing Li , Xin Yu , Ying Xu

In this study, a realistic flexible or hybrid flowshop scheduling problem (HFS) is investigated, in which the following constraints are embedded, i.e., resource-dependent processing, robotic arm loading, and transportation. To solve the considered problem, a multi-dimensional co-evolutionary algorithm (MDCEA) is proposed to minimize makespan and total energy consumption (TEC) simultaneously. First, in the MDCEA, solutions are encoded by a three-dimensional vector with a two-phase decoding heuristic. Then, the initialized population is divided into three subsets to focus on different search tasks. To improve the efficiency of the global search task, a dual-population-based variable dimension cooperative search method is developed. In addition, to explore the promising non-dominated solutions in different dimensions, a Q-learning-based dimension detection search method is designed for the local search task. Finally, to keep the diversity in the evolutionary process, a knowledge-based individual transfer strategy is conducted for populations. The proposed algorithm was tested on 25 randomly generated instances, and detailed comparisons verified the efficiency and robustness compared to six state-of-the-art algorithms was achieved.

{"title":"A multi-dimensional co-evolutionary algorithm for multi-objective resource-constrained flexible flowshop with robotic transportation","authors":"Jia-ke Li , Rong-hao Li , Jun-qing Li , Xin Yu , Ying Xu","doi":"10.1016/j.asoc.2024.112689","DOIUrl":"10.1016/j.asoc.2024.112689","url":null,"abstract":"<div><div>In this study, a realistic flexible or hybrid flowshop scheduling problem (HFS) is investigated, in which the following constraints are embedded, i.e., resource-dependent processing, robotic arm loading, and transportation. To solve the considered problem, a multi-dimensional co-evolutionary algorithm (MDCEA) is proposed to minimize makespan and total energy consumption (TEC) simultaneously. First, in the MDCEA, solutions are encoded by a three-dimensional vector with a two-phase decoding heuristic. Then, the initialized population is divided into three subsets to focus on different search tasks. To improve the efficiency of the global search task, a dual-population-based variable dimension cooperative search method is developed. In addition, to explore the promising non-dominated solutions in different dimensions, a Q-learning-based dimension detection search method is designed for the local search task. Finally, to keep the diversity in the evolutionary process, a knowledge-based individual transfer strategy is conducted for populations. The proposed algorithm was tested on 25 randomly generated instances, and detailed comparisons verified the efficiency and robustness compared to six state-of-the-art algorithms was achieved.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"170 ","pages":"Article 112689"},"PeriodicalIF":7.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143213081","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}

引用次数: 0

Artificial intelligence for accurate classification of respiratory abnormality levels using image-based features and interpretable insights

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Applied Soft Computing

Pub Date : 2025-02-01 DOI: 10.1016/j.asoc.2024.112678

Wei Zeng , Liangmin Shan , Qinghui Wang , Fenglin Liu , Ying Wang , Chengzhi Yuan , Shaoyi Du

Accurate classification of respiratory abnormality levels is crucial for early detection and diagnosis of respiratory diseases, making it a pivotal area in the field of medical diagnostics. This study proposes a novel artificial intelligence approach for accurate classification of respiratory abnormality levels. By transforming respiratory sound time-series data into image representations using recurrent plot, Markov transition field, and Gramian angular field, we capture intricate temporal patterns and spatial relationships. A deep neural network autonomously extracts discriminative features from these representations, subsequently integrated into machine learning classifiers. Leveraging the International Conference on Biomedical and Health Informatics (ICBHI) database, our methodology achieves remarkable classification accuracy of 100% for both binary and four-class scenarios, accurately distinguishing normal from abnormal sounds, and discriminating between crackles, wheezes, and their combinations. The SHapley Additive exPlanations (SHAP) method enhances interpretability, providing insights into feature importance and decision-making processes. This interpretable and high-performing approach offers significant promise for enhancing the accuracy and reliability of respiratory disorder diagnosis and treatment planning in clinical settings, potentially improving patient outcomes and healthcare efficiency.

{"title":"Artificial intelligence for accurate classification of respiratory abnormality levels using image-based features and interpretable insights","authors":"Wei Zeng , Liangmin Shan , Qinghui Wang , Fenglin Liu , Ying Wang , Chengzhi Yuan , Shaoyi Du","doi":"10.1016/j.asoc.2024.112678","DOIUrl":"10.1016/j.asoc.2024.112678","url":null,"abstract":"<div><div>Accurate classification of respiratory abnormality levels is crucial for early detection and diagnosis of respiratory diseases, making it a pivotal area in the field of medical diagnostics. This study proposes a novel artificial intelligence approach for accurate classification of respiratory abnormality levels. By transforming respiratory sound time-series data into image representations using recurrent plot, Markov transition field, and Gramian angular field, we capture intricate temporal patterns and spatial relationships. A deep neural network autonomously extracts discriminative features from these representations, subsequently integrated into machine learning classifiers. Leveraging the International Conference on Biomedical and Health Informatics (ICBHI) database, our methodology achieves remarkable classification accuracy of 100% for both binary and four-class scenarios, accurately distinguishing normal from abnormal sounds, and discriminating between crackles, wheezes, and their combinations. The SHapley Additive exPlanations (SHAP) method enhances interpretability, providing insights into feature importance and decision-making processes. This interpretable and high-performing approach offers significant promise for enhancing the accuracy and reliability of respiratory disorder diagnosis and treatment planning in clinical settings, potentially improving patient outcomes and healthcare efficiency.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"170 ","pages":"Article 112678"},"PeriodicalIF":7.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143213212","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}

引用次数: 0

Air quality index prediction through TimeGAN data recovery and PSO-optimized VMD-deep learning framework

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Applied Soft Computing

Pub Date : 2025-02-01 DOI: 10.1016/j.asoc.2024.112626

Kenan Wang , Tianning Yang , Shanshan Kong , Mingduo Li

With the rapid development of the economy, air pollution has become increasingly severe. Accurate prediction of the Air Quality Index (AQI) is crucial for safeguarding public health and the environment. However, AQI time series exhibit strong randomness and volatility, posing challenges for traditional forecasting methods to achieve precise AQI predictions. Therefore, we propose a new AQI hybrid prediction model, TG-Hybrid model, which integrates generative artificial intelligence, signal decomposition techniques, artificial intelligence methods, and optimization algorithms. In the proposed model, missing values in the data are handled using generative adversarial networks, effectively addressing the issue of a large number of missing values in time series data. Autoregressive integrated moving average is employed to forecast the linear components of the data, while variational mode decomposition decomposes AQI into multiple modes. Particle swarm optimization is used to combine the prediction results of convolutional neural network combined with bidirectional long short-term memory and extreme gradient boosting. Additionally, AQI prediction experiments were conducted using air pollution data from Tangshan and Beijing, and compared with fifteen other models. The results indicate that the root mean square error for Tangshan and Beijing are 6.407 and 7.485, respectively, significantly outperforming other baseline models.

{"title":"Air quality index prediction through TimeGAN data recovery and PSO-optimized VMD-deep learning framework","authors":"Kenan Wang , Tianning Yang , Shanshan Kong , Mingduo Li","doi":"10.1016/j.asoc.2024.112626","DOIUrl":"10.1016/j.asoc.2024.112626","url":null,"abstract":"<div><div>With the rapid development of the economy, air pollution has become increasingly severe. Accurate prediction of the Air Quality Index (AQI) is crucial for safeguarding public health and the environment. However, AQI time series exhibit strong randomness and volatility, posing challenges for traditional forecasting methods to achieve precise AQI predictions. Therefore, we propose a new AQI hybrid prediction model, TG-Hybrid model, which integrates generative artificial intelligence, signal decomposition techniques, artificial intelligence methods, and optimization algorithms. In the proposed model, missing values in the data are handled using generative adversarial networks, effectively addressing the issue of a large number of missing values in time series data. Autoregressive integrated moving average is employed to forecast the linear components of the data, while variational mode decomposition decomposes AQI into multiple modes. Particle swarm optimization is used to combine the prediction results of convolutional neural network combined with bidirectional long short-term memory and extreme gradient boosting. Additionally, AQI prediction experiments were conducted using air pollution data from Tangshan and Beijing, and compared with fifteen other models. The results indicate that the root mean square error for Tangshan and Beijing are 6.407 and 7.485, respectively, significantly outperforming other baseline models.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"170 ","pages":"Article 112626"},"PeriodicalIF":7.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143213269","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}

引用次数: 0

TFGNet: Frequency-guided saliency detection for complex scenes

IF 7.2 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Applied Soft Computing

Pub Date : 2025-02-01 DOI: 10.1016/j.asoc.2024.112685

Yi Wang , Ruili Wang , Juncheng Liu , Rui Xu , Tianzhu Wang , Feng Hou , Bin Liu , Na Lei

Salient object detection (SOD) with accurate boundaries in complex and chaotic natural or social scenes remains a significant challenge. Many edge-aware or/and two-branch models rely on exchanging global and local information between multistage features, which can propagate errors and lead to incorrect predictions. To address this issue, this work explores the fundamental problems in current U-Net architecture-based SOD models from the perspective of image spatial frequency decomposition and synthesis. A concise and efficient Frequency-Guided Network (TFGNet) is proposed that simultaneously learns the boundary details (high-spatial frequency) and inner regions (low-spatial frequency) of salient regions in two separate branches. Each branch utilizes a Multiscale Frequency Feature Enhancement (FFE) module to learn pixel-wise frequency features and a Transformer-based decoder to learn mask-wise frequency features, improving a comprehensive understanding of salient regions. TFGNet eliminates the need to exchange global and local features at intermediate layers of the two branches, thereby reducing interference from erroneous information. A hybrid loss function is also proposed to combine BCE, IoU, and Histogram dissimilarity to ensure pixel accuracy, structural integrity, and frequency distribution consistency between ground truth and predicted saliency maps. Comprehensive evaluations have been conducted on five widely used SOD datasets and one underwater SOD dataset, demonstrating the superior performance of TFGNet compared to state-of-the-art methods. The codes and results are available at https://github.com/yiwangtz/TFGNet.

{"title":"TFGNet: Frequency-guided saliency detection for complex scenes","authors":"Yi Wang , Ruili Wang , Juncheng Liu , Rui Xu , Tianzhu Wang , Feng Hou , Bin Liu , Na Lei","doi":"10.1016/j.asoc.2024.112685","DOIUrl":"10.1016/j.asoc.2024.112685","url":null,"abstract":"<div><div>Salient object detection (SOD) with accurate boundaries in complex and chaotic natural or social scenes remains a significant challenge. Many edge-aware or/and two-branch models rely on exchanging global and local information between multistage features, which can propagate errors and lead to incorrect predictions. To address this issue, this work explores the fundamental problems in current U-Net architecture-based SOD models from the perspective of image spatial frequency decomposition and synthesis. A concise and efficient Frequency-Guided Network (TFGNet) is proposed that simultaneously learns the boundary details (high-spatial frequency) and inner regions (low-spatial frequency) of salient regions in two separate branches. Each branch utilizes a Multiscale Frequency Feature Enhancement (FFE) module to learn pixel-wise frequency features and a Transformer-based decoder to learn mask-wise frequency features, improving a comprehensive understanding of salient regions. TFGNet eliminates the need to exchange global and local features at intermediate layers of the two branches, thereby reducing interference from erroneous information. A hybrid loss function is also proposed to combine BCE, IoU, and Histogram dissimilarity to ensure pixel accuracy, structural integrity, and frequency distribution consistency between ground truth and predicted saliency maps. Comprehensive evaluations have been conducted on five widely used SOD datasets and one underwater SOD dataset, demonstrating the superior performance of TFGNet compared to state-of-the-art methods. The codes and results are available at <span><span>https://github.com/yiwangtz/TFGNet</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50737,"journal":{"name":"Applied Soft Computing","volume":"170 ","pages":"Article 112685"},"PeriodicalIF":7.2,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143213335","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0