Chih-Hung Wu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Chih-Hung Wu has a total of 37 co-authors. Here are the top co-authors by number of shared publications:
- Chih-Feng Liu
YajanTech Co. Ltd., Tainan, Taiwan
2 shared publications Chiung-Hui Tsai
2 shared publications- Chiung-Hui Tsai
Department of Electrical Engineering, National University of Kaohsiung, Kaohsiung, Taiwan
2 shared publications - Shie-Jue Lee
Department of Electrical Engineering, National Sun Yat-Sen University, Kaohsiung, Taiwan
2 shared publications Chang-Yeh Hsieh
Department of Electrical Engineering, National University of Kaohsiung, Kaohsiung, Taiwan
1 shared publication- Chen-Sen Ouyang
Department of Information Engineering, I-Shou University, Kaohsiung, Taiwan
1 shared publication
Latest Publications
- Open Access
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Journal of Industrial and Management Optimization
- Vol. 22 (1) , 182-213
A genetic algorithm for tool-constrained scheduling with dynamic expedited orders in smart manufacturing
This paper addresses a practical scheduling challenge in smart manufacturing systems, where dynamic expedited orders and tool change constraints must be handled in real time. While deep learning and multi-objective optimization methods offer strong theoretical capabilities, they are often unsuitable for industrial deployment due to high computational demands, training costs, and lack of transparency. We propose a lightweight, interpretable genetic algorithm framework that incorporates a real-time insertion mechanism and tooling-aware encoding to minimize disruptions when handling expedited orders. The method jointly considers machine-tool compatibility, job priorities, and system stability. It is validated using real production data from a precision screw manufacturing plant, demonstrating superior performance in handling expedited orders and tool constraints. The proposed method requires no training, supports online adjustments, and is well-suited for deployment in real-world production settings.
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Journal of Industrial and Management Optimization
- Vol. 22 (1) , 182-213
Open AccessA genetic algorithm for tool-constrained scheduling with dynamic expedited orders in smart manufacturing
This paper addresses a practical scheduling challenge in smart manufacturing systems, where dynamic expedited orders and tool change constraints must be handled in real time. While deep learning and multi-objective optimization methods offer strong theoretical capabilities, they are often unsuitable for industrial deployment due to high computational demands, training costs, and lack of transparency. We propose a lightweight, interpretable genetic algorithm framework that incorporates a real-time insertion mechanism and tooling-aware encoding to minimize disruptions when handling expedited orders. The method jointly considers machine-tool compatibility, job priorities, and system stability. It is validated using real production data from a precision screw manufacturing plant, demonstrating superior performance in handling expedited orders and tool constraints. The proposed method requires no training, supports online adjustments, and is well-suited for deployment in real-world production settings.
- 10 April 2025
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 84 (34) , 42009-42035
Integrating local binary patterns and convolutional neural networks with a balanced gray encoded pooling mechanism
- 10 April 2025
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 84 (34) , 42009-42035
Integrating local binary patterns and convolutional neural networks with a balanced gray encoded pooling mechanism
- Open Access
- 11 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 190598-190610
Deep-Learning Methods for Defect Inspection of Plated Through Holes With Clustering-Based Auto-Labeling and GAN-Based Model Training
This paper presents the integration of several deep learning techniques for defect inspection of plated through-hole (PTH) on printed circuit boards (PCBs). In our proposed system, the object detection technology of You Only Look Once (YOLO) allocates the position of PTHs; a semi-automatic clustering mechanism distinguishes normal and defective PTHs for collecting training data. A convolution neural network based on the ResNet framework is established for detecting PTH defects. A retrain mechanism is designed for retracting misclassified PTHs and updating the recognition model. The generative adversarial network (GAN) is employed to deal with the insufficiency of negative training samples. The detection model can achieve 98.96% accuracy with continuous retraining and data augmentation. Further, a filtering mechanism based on template analysis eliminates ambiguous outputs of GAN and improves the accuracy of the detection model. This study integrates these deep learning techniques to establish a PTH defect detection system tested in a PCB manufactory against real PTHs and gained an overall prediction accuracy of 99.48%.
- 11 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 190598-190610
Open AccessDeep-Learning Methods for Defect Inspection of Plated Through Holes With Clustering-Based Auto-Labeling and GAN-Based Model Training
This paper presents the integration of several deep learning techniques for defect inspection of plated through-hole (PTH) on printed circuit boards (PCBs). In our proposed system, the object detection technology of You Only Look Once (YOLO) allocates the position of PTHs; a semi-automatic clustering mechanism distinguishes normal and defective PTHs for collecting training data. A convolution neural network based on the ResNet framework is established for detecting PTH defects. A retrain mechanism is designed for retracting misclassified PTHs and updating the recognition model. The generative adversarial network (GAN) is employed to deal with the insufficiency of negative training samples. The detection model can achieve 98.96% accuracy with continuous retraining and data augmentation. Further, a filtering mechanism based on template analysis eliminates ambiguous outputs of GAN and improves the accuracy of the detection model. This study integrates these deep learning techniques to establish a PTH defect detection system tested in a PCB manufactory against real PTHs and gained an overall prediction accuracy of 99.48%.
- 3 October 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 53 (10) , 12792-12810
A hybrid deep learning network for forecasting air pollutant concentrations
- 3 October 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 53 (10) , 12792-12810
A hybrid deep learning network for forecasting air pollutant concentrations
- 14 July 2021
- Published by Springer Nature in Information Retrieval Journal
- Vol. 24 (4-5) , 298-321
Using word semantic concepts for plagiarism detection in text documents
- 14 July 2021
- Published by Springer Nature in Information Retrieval Journal
- Vol. 24 (4-5) , 298-321
Using word semantic concepts for plagiarism detection in text documents
- Open Access
- 10 March 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 49760-49773
Migration-Based Load Balance of Virtual Machine Servers in Cloud Computing by Load Prediction Using Genetic-Based Methods
This paper presents a two-stage genetic mechanism for the migration-based load balance of virtual machine hosts (VMHs) in cloud computing. Previous methods usually assume this issue as a job-assignment optimization problem and only consider the current VMHs' loads; however, without considering loads of VMHs after balancing, these methods can only gain limited effectiveness in real environments. In this study, two genetic-based methods are integrated and presented. First, performance models of virtual machines (VMs) are extracted from their creating parameters and corresponding performance measured in a cloud computing environment. The gene expression programming (GEP) is applied for generating symbolic regression models that describe the performance of VMs and are used for predicting loads of VMHs after load-balance. Secondly, with the VMH loads estimated by GEP, the genetic algorithm considers the current and the future loads of VMHs and decides an optimal VM-VMH assignment for migrating VMs and performing load-balance. The performance of the proposed methods is evaluated in a real cloud-computing environment, Jnet, wherein these methods are implemented as a centralized load balancing mechanism. The experimental results show that our method outperforms previous methods, such as heuristics and statistics regression.
- 10 March 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 49760-49773
Open AccessMigration-Based Load Balance of Virtual Machine Servers in Cloud Computing by Load Prediction Using Genetic-Based Methods
This paper presents a two-stage genetic mechanism for the migration-based load balance of virtual machine hosts (VMHs) in cloud computing. Previous methods usually assume this issue as a job-assignment optimization problem and only consider the current VMHs' loads; however, without considering loads of VMHs after balancing, these methods can only gain limited effectiveness in real environments. In this study, two genetic-based methods are integrated and presented. First, performance models of virtual machines (VMs) are extracted from their creating parameters and corresponding performance measured in a cloud computing environment. The gene expression programming (GEP) is applied for generating symbolic regression models that describe the performance of VMs and are used for predicting loads of VMHs after load-balance. Secondly, with the VMH loads estimated by GEP, the genetic algorithm considers the current and the future loads of VMHs and decides an optimal VM-VMH assignment for migrating VMs and performing load-balance. The performance of the proposed methods is evaluated in a real cloud-computing environment, Jnet, wherein these methods are implemented as a centralized load balancing mechanism. The experimental results show that our method outperforms previous methods, such as heuristics and statistics regression.
- 28 December 2020
- Published by Wiley in International Journal of Imaging Systems and Technology
- Vol. 31 (3) , 1300-1316
Local binary pattern encoding schemes for computed tomography image segmentation: An experimental and comparative study
Local binary patterns (LBPs) are used for effective texture representation in various applications. This study explores the clustering consistency and stability of image segmentation when distance‐based clustering methods are used with an LBP. Because data are described by features and attributes, distances among data are also dominated by the definition of features. Moreover, four popular LBP encoding schemes for segmenting computed tomography (CT) images by using fuzzy C‐means are discussed and compared. The experimental results demonstrate several notable phenomena: When the pixels are encoded in different LBP encoding schemes, the distance between these pixels varies considerably. The experimental results indicate that each LBP encoding scheme emphasizes a specific image texture that dominates how pixels are described in the feature space and thus affects the clustering results. Based on the evaluation of CT image segmentation by using the clustering inconsistency index, linear and corner‐like LBPs are particularly suitable for clustering longitudinal sections and cross sections of CT images, respectively.
- 28 December 2020
- Published by Wiley in International Journal of Imaging Systems and Technology
- Vol. 31 (3) , 1300-1316
Local binary pattern encoding schemes for computed tomography image segmentation: An experimental and comparative study
Local binary patterns (LBPs) are used for effective texture representation in various applications. This study explores the clustering consistency and stability of image segmentation when distance‐based clustering methods are used with an LBP. Because data are described by features and attributes, distances among data are also dominated by the definition of features. Moreover, four popular LBP encoding schemes for segmenting computed tomography (CT) images by using fuzzy C‐means are discussed and compared. The experimental results demonstrate several notable phenomena: When the pixels are encoded in different LBP encoding schemes, the distance between these pixels varies considerably. The experimental results indicate that each LBP encoding scheme emphasizes a specific image texture that dominates how pixels are described in the feature space and thus affects the clustering results. Based on the evaluation of CT image segmentation by using the clustering inconsistency index, linear and corner‐like LBPs are particularly suitable for clustering longitudinal sections and cross sections of CT images, respectively.
- 1 December 2018
- Published by Springer Nature
- p. 277-283
A Comparative Study on Encoding Methods of Local Binary Patterns for Image Segmentation
- 1 December 2018
- Published by Springer Nature
- p. 277-283
A Comparative Study on Encoding Methods of Local Binary Patterns for Image Segmentation
- 22 December 2014
- Published by Taylor & Francis in Journal of the Chinese Institute of Engineers
- Vol. 38 (4) , 458-468
Associative classification with a new condenseness measure
Associative classification (AC) is a branch of data mining that utilizes association rules (ARs) for classification. ARs are extracted from databases that satisfy some statistical criteria such as minimal support. However, in some practical applications, useful ARs may be found among infrequent, but closely related, itemsets that are filtered out by high minimal support. In this study, a new measure, named “condenseness,” is presented for evaluating whether infrequent ruleitems that are filtered out by minimal support can form strong ARs for classification. For an infrequent ruleitem, the condenseness is the average of lift of all ARs that can be generated from the ruleitem. A ruleitem with a high condenseness means that its elements are closely related and can serve for AC even if it does not have high support. Based on the concept of condenseness, a new associative classifier is developed and presented – condensed association rules for classification (CARC). CARC generates ARs using a modified Apriori algorithm and develops new strategies of rule inference. With the condenseness measure and strategies for rule inference, more useful ARs can be produced and improve the effectiveness of association classification. Empirical evidences show that CARC mitigates the problems caused by setting too high/low minimal support and has a better performance on classification.
- 22 December 2014
- Published by Taylor & Francis in Journal of the Chinese Institute of Engineers
- Vol. 38 (4) , 458-468
Associative classification with a new condenseness measure
Associative classification (AC) is a branch of data mining that utilizes association rules (ARs) for classification. ARs are extracted from databases that satisfy some statistical criteria such as minimal support. However, in some practical applications, useful ARs may be found among infrequent, but closely related, itemsets that are filtered out by high minimal support. In this study, a new measure, named “condenseness,” is presented for evaluating whether infrequent ruleitems that are filtered out by minimal support can form strong ARs for classification. For an infrequent ruleitem, the condenseness is the average of lift of all ARs that can be generated from the ruleitem. A ruleitem with a high condenseness means that its elements are closely related and can serve for AC even if it does not have high support. Based on the concept of condenseness, a new associative classifier is developed and presented – condensed association rules for classification (CARC). CARC generates ARs using a modified Apriori algorithm and develops new strategies of rule inference. With the condenseness measure and strategies for rule inference, more useful ARs can be produced and improve the effectiveness of association classification. Empirical evidences show that CARC mitigates the problems caused by setting too high/low minimal support and has a better performance on classification.
- 17 July 2013
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 62 (12) , 3218-3230
Target Position Estimation by Genetic Expression Programming for Mobile Robots With Vision Sensors
Target location is an important task in robotics applications. For different application purposes, the positions of targets are usually described by various coordinate systems. Closed-form formulas that describe the relationships between two coordinate systems serve as a means for coordinate transformation. However, the existence of unavoidable measurement errors and uncertainty makes closed-form formulas less reliable. Besides, the closed-form formulas usually involve operations of matrix inversion and transpose that usually consume a considerable amount of computing resources. This paper defines the problem of coordinate transformation on mobile robots as a regression problem and employs the techniques of gene expression programming to discover the regression models. With such regression models, coordinate transformation can be done by simpler formulas with lower processing costs. The proposed techniques have been implemented and integrated with a four-wheeled robot equipped with vision sensors and have been verified in real environments. The experiments demonstrate the effectiveness and performance of the proposed method. To the best of our knowledge, this is the first study on the underlying problem using genetic-based techniques.
- 17 July 2013
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 62 (12) , 3218-3230
Target Position Estimation by Genetic Expression Programming for Mobile Robots With Vision Sensors
Target location is an important task in robotics applications. For different application purposes, the positions of targets are usually described by various coordinate systems. Closed-form formulas that describe the relationships between two coordinate systems serve as a means for coordinate transformation. However, the existence of unavoidable measurement errors and uncertainty makes closed-form formulas less reliable. Besides, the closed-form formulas usually involve operations of matrix inversion and transpose that usually consume a considerable amount of computing resources. This paper defines the problem of coordinate transformation on mobile robots as a regression problem and employs the techniques of gene expression programming to discover the regression models. With such regression models, coordinate transformation can be done by simpler formulas with lower processing costs. The proposed techniques have been implemented and integrated with a four-wheeled robot equipped with vision sensors and have been verified in real environments. The experiments demonstrate the effectiveness and performance of the proposed method. To the best of our knowledge, this is the first study on the underlying problem using genetic-based techniques.