Cadmus Yuan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Cadmus Yuan has a total of 40 co-authors. Here are the top co-authors by number of shared publications:
- Xuejun Fan
Department of Mechanical Engineering, Lamar University, Beaumont, USA
5 shared publications - Hongyu Tang
Department of Microelectronics
3 shared publications Yen-Chang Huang
Institute of Oral Sciences, Chung Shan Medical University, Taichung, Taiwan
3 shared publicationsGuang-Qian Shen
Department of Mechanical and Computer-Aided Engineering, Feng Chia University, Taichung, Taiwan
2 shared publications- Guoqi Zhang
Delft University of Technology
2 shared publications Jian-Cheng Hong
2 shared publications
Latest Publications
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 155421-155432
Artificial Intelligence (AI) for Semiconductor Packaging
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 155421-155432
Open AccessArtificial Intelligence (AI) for Semiconductor Packaging
- 29 November 2025
- Published by Elsevier in Microelectronics Reliability
- Vol. 176
AI surrogate modeling for PBGA solder joint fatigue risk assessment
- 29 November 2025
- Published by Elsevier in Microelectronics Reliability
- Vol. 176
AI surrogate modeling for PBGA solder joint fatigue risk assessment
- 6 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Design Optimization Study by AI-Surrogate PBGA Reliability Model and Transfer Learning
Solder joint reliability remains an important design issue for ball grid array (BGA) type packaging. Although the finite element model is used to optimize the solder joint reliability, the finite element method's deterministic nature limits the generation of statistical information. In this paper, the mechanical characteristics of the solder joint of PBGA packaging are represented by a validated and simplified finite model. We then explored an AI-surrogate model to capture the solder joint fatigue characteristics of PBGA packaging and then applied the same surrogate for design optimization and process uncertainty. Various techniques are integrated, including finite model simplification and validation methods, AI training methods, AI model optimization, and statistical uncertainty assessment methods. The result shows the seamless integration of all the above-mentioned methods. Moreover, when different AI-surrogate models are applied, stable optimization results can be expected when the proper AI-surrogate model training criterion can be met. Furthermore, the transfer learning technique is applied to obtain a revised NN model against the new database, and its prediction capability is demonstrated.
- 6 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Design Optimization Study by AI-Surrogate PBGA Reliability Model and Transfer Learning
Solder joint reliability remains an important design issue for ball grid array (BGA) type packaging. Although the finite element model is used to optimize the solder joint reliability, the finite element method's deterministic nature limits the generation of statistical information. In this paper, the mechanical characteristics of the solder joint of PBGA packaging are represented by a validated and simplified finite model. We then explored an AI-surrogate model to capture the solder joint fatigue characteristics of PBGA packaging and then applied the same surrogate for design optimization and process uncertainty. Various techniques are integrated, including finite model simplification and validation methods, AI training methods, AI model optimization, and statistical uncertainty assessment methods. The result shows the seamless integration of all the above-mentioned methods. Moreover, when different AI-surrogate models are applied, stable optimization results can be expected when the proper AI-surrogate model training criterion can be met. Furthermore, the transfer learning technique is applied to obtain a revised NN model against the new database, and its prediction capability is demonstrated.
- Open Access
A Small Database with an Adaptive Data Selection Method for Solder Joint Fatigue Life Prediction in Advanced Packaging
There has always been high interest in predicting the solder joint fatigue life in advanced packaging with high accuracy and efficiency. Artificial Intelligence Plus (AI+) is becoming increasingly popular as computational facilities continue to develop. This study will introduce machine learning (a core component of AI). With machine learning, metamodels that approximate the attributes of systems or functions are created to predict the fatigue life of advanced packaging. However, the prediction ability is highly dependent on the size and distribution of the training data. Increasing the amount of training data is the most intuitive approach to improve prediction performance, but this implies a higher computational cost. In this research, the adaptive sampling methods are applied to build the machine learning model with a small dataset sampled from an existing database. The performance of the model will be visualized using predefined criteria. Moreover, ensemble learning can be used to improve the performance of AI models after they have been fully trained.
Open AccessA Small Database with an Adaptive Data Selection Method for Solder Joint Fatigue Life Prediction in Advanced Packaging
There has always been high interest in predicting the solder joint fatigue life in advanced packaging with high accuracy and efficiency. Artificial Intelligence Plus (AI+) is becoming increasingly popular as computational facilities continue to develop. This study will introduce machine learning (a core component of AI). With machine learning, metamodels that approximate the attributes of systems or functions are created to predict the fatigue life of advanced packaging. However, the prediction ability is highly dependent on the size and distribution of the training data. Increasing the amount of training data is the most intuitive approach to improve prediction performance, but this implies a higher computational cost. In this research, the adaptive sampling methods are applied to build the machine learning model with a small dataset sampled from an existing database. The performance of the model will be visualized using predefined criteria. Moreover, ensemble learning can be used to improve the performance of AI models after they have been fully trained.
- Open Access
Simulation education utilizing phantom and angle reference guide in pulmonary nodule CT localization
- Open Access
Coefficient Extraction of SAC305 Solder Constitutive Equations Using Equation-Informed Neural Networks
Equation-Informed Neural Networks (EINNs) are developed as an efficient method for extracting the coefficients of constitutive equations. Subsequently, numerical Bayesian Inference (BI) iterations were applied to estimate the distribution of these coefficients, thereby further refining them. We could generate coefficients optimally aligned with the targeted application scenario by carefully adjusting pre-processing mapping parameters and identifying dataset preferences. Leveraging graphical representation techniques, the EINNs formulation is implemented in temperature- and strain-rate-dependent hyperbolic Garofalo, Anand, and Chaboche constitutive models to extract the corresponding coefficients for lead-free SAC305 solder material. The performance of the EINNs-based extracted coefficients, obtained from experimental results of SAC305 solder material, is comparable to existing studies. The methodology offers the dual advantage of providing the coefficients’ value and distribution against the training dataset.
Open AccessCoefficient Extraction of SAC305 Solder Constitutive Equations Using Equation-Informed Neural Networks
Equation-Informed Neural Networks (EINNs) are developed as an efficient method for extracting the coefficients of constitutive equations. Subsequently, numerical Bayesian Inference (BI) iterations were applied to estimate the distribution of these coefficients, thereby further refining them. We could generate coefficients optimally aligned with the targeted application scenario by carefully adjusting pre-processing mapping parameters and identifying dataset preferences. Leveraging graphical representation techniques, the EINNs formulation is implemented in temperature- and strain-rate-dependent hyperbolic Garofalo, Anand, and Chaboche constitutive models to extract the corresponding coefficients for lead-free SAC305 solder material. The performance of the EINNs-based extracted coefficients, obtained from experimental results of SAC305 solder material, is comparable to existing studies. The methodology offers the dual advantage of providing the coefficients’ value and distribution against the training dataset.
- Open Access
- 9 September 2022
- Published by MDPI AG in Electronics
- Vol. 11 (18) , 2861
An AI-Based Adaptive Surrogate Modeling Method for the In-Service Response of UVLED Modules
The response forecasting of in-service complex electronic systems remains a challenge due to its uncertainty. An AI-based adaptive surrogate modeling method, including offline and online learning procedures, is proposed in this research for different systems with significant variety. The offline learning aims to abstract the knowledge from the known information and represent it as root models. The in-service response is modeled by a linear combination of the online learning of these root models against the continuous new measurement. This research applies a performance measurement dataset of the UVLED modules with considerable deviation to verify the proposed method. Part of the datasets is selected to generate the root models by offline learning, and these root models are applied to the online learning procedures for the adaptive surrogate model (ASM) of the different systems. The results show that after approximately 10 online learning iterations, the ASM achieves the capability of predicting 1000 h of response.
- 9 September 2022
- Published by MDPI AG in Electronics
- Vol. 11 (18) , 2861
Open AccessAn AI-Based Adaptive Surrogate Modeling Method for the In-Service Response of UVLED Modules
The response forecasting of in-service complex electronic systems remains a challenge due to its uncertainty. An AI-based adaptive surrogate modeling method, including offline and online learning procedures, is proposed in this research for different systems with significant variety. The offline learning aims to abstract the knowledge from the known information and represent it as root models. The in-service response is modeled by a linear combination of the online learning of these root models against the continuous new measurement. This research applies a performance measurement dataset of the UVLED modules with considerable deviation to verify the proposed method. Part of the datasets is selected to generate the root models by offline learning, and these root models are applied to the online learning procedures for the adaptive surrogate model (ASM) of the different systems. The results show that after approximately 10 online learning iterations, the ASM achieves the capability of predicting 1000 h of response.
- Open Access
- 11 June 2022
- Published by Elsevier in Journal of Dental Sciences
- Vol. 18 (1) , 248-263
Biomechanical evaluation of bridge span with three implant abutment designs and two connectors for tooth-implant supported prosthesis: A finite element analysis
- 11 June 2022
- Published by Elsevier in Journal of Dental Sciences
- Vol. 18 (1) , 248-263
Open AccessBiomechanical evaluation of bridge span with three implant abutment designs and two connectors for tooth-implant supported prosthesis: A finite element analysis
- Open Access
- 11 December 2021
- Published by Oxford University Press (OUP) in Journal of Mechanics
- Vol. 37, 659-668
The research on the dental bridge model-making process based on the curing shrinkage epoxy and residual stress reduction
Elderly people suffer from more and more teeth problems. The tooth-implant-supported prosthesis provides a reliable solution to missing teeth patients. The proper dental prosthesis design to prevent overstress is essential due to the mechanical characteristics of the dental bridge abutments are different. The finite element method is widely applied, but proper experimental validation is required. The curing shrinkage epoxy is applied for the photoelasticity measurement because its mechanical property is close to the cancellous bone. A series of process developments, including mold design, residual stress releasing and artificial soft film making, is accomplished in this research to simulate the mechanical response of dental bridges in practice. The process is proven and can be accomplished at the dentist's workshop. The transmission photoelasticity technique is applied to measure the residual stress distribution and it nondestructively provides the continuous improvement guideline. The model-making procedure and tools are proven to be available at the dental workshop. Following the model-making procedure, the dental bridge model shows a low residual stress level that the photoelasticity system cannot detect. Excellent reproducibility of the proposed procedure has been validated. These models exhibit stable maximum stress of 2.13 MPa around the natural tooth apex and apical implant region when a 300 N vertical loading is applied upon the dental bridge. Finally, a finite element model of the dental bridge, including the natural tooth and dental implant, is built and validated by the photoelastic measurement.
- 11 December 2021
- Published by Oxford University Press (OUP) in Journal of Mechanics
- Vol. 37, 659-668
Open AccessThe research on the dental bridge model-making process based on the curing shrinkage epoxy and residual stress reduction
Elderly people suffer from more and more teeth problems. The tooth-implant-supported prosthesis provides a reliable solution to missing teeth patients. The proper dental prosthesis design to prevent overstress is essential due to the mechanical characteristics of the dental bridge abutments are different. The finite element method is widely applied, but proper experimental validation is required. The curing shrinkage epoxy is applied for the photoelasticity measurement because its mechanical property is close to the cancellous bone. A series of process developments, including mold design, residual stress releasing and artificial soft film making, is accomplished in this research to simulate the mechanical response of dental bridges in practice. The process is proven and can be accomplished at the dentist's workshop. The transmission photoelasticity technique is applied to measure the residual stress distribution and it nondestructively provides the continuous improvement guideline. The model-making procedure and tools are proven to be available at the dental workshop. Following the model-making procedure, the dental bridge model shows a low residual stress level that the photoelasticity system cannot detect. Excellent reproducibility of the proposed procedure has been validated. These models exhibit stable maximum stress of 2.13 MPa around the natural tooth apex and apical implant region when a 300 N vertical loading is applied upon the dental bridge. Finally, a finite element model of the dental bridge, including the natural tooth and dental implant, is built and validated by the photoelastic measurement.
- Open Access
- 18 October 2021
- Published by MDPI AG in Electronics
- Vol. 10 (20) , 2540
Using a Flexible IoT Architecture and Sequential AI Model to Recognize and Predict the Production Activities in the Labor-Intensive Manufacturing Site
Under the pressures of global market uncertainty and rapid production changes, the labor-intensive industries demand instant manufacturing site information and accurate production forecasting. This research applies sensor modules with noise reduction, information abstracting, and wireless transmission functions to form a flexible internet of things (IoT) architecture for acquiring field information. Moreover, AI models are used to reveal human activities and predict the output of a group of workstations. The IoT architecture has been implemented in the actual shoe making site. Although there is a 5% missing data issue due to network transmission, neural network models can successfully convert the IoT data to machine utilization. By analyzing the field data, the actual collaboration among the worker team can be revealed. Furthermore, a sequential AI model is applied to learn to capture the characteristics of the team working. This AI model only requires training by 15 min of IoT data, then it can predict the current and next few days’ productions within 10% error. This research confirms that implementing the IoT architecture and applying the AI model enables instant manufacturing monitoring of labor-intensive manufacturing sites and accurate production forecasting.
- 18 October 2021
- Published by MDPI AG in Electronics
- Vol. 10 (20) , 2540
Open AccessUsing a Flexible IoT Architecture and Sequential AI Model to Recognize and Predict the Production Activities in the Labor-Intensive Manufacturing Site
Under the pressures of global market uncertainty and rapid production changes, the labor-intensive industries demand instant manufacturing site information and accurate production forecasting. This research applies sensor modules with noise reduction, information abstracting, and wireless transmission functions to form a flexible internet of things (IoT) architecture for acquiring field information. Moreover, AI models are used to reveal human activities and predict the output of a group of workstations. The IoT architecture has been implemented in the actual shoe making site. Although there is a 5% missing data issue due to network transmission, neural network models can successfully convert the IoT data to machine utilization. By analyzing the field data, the actual collaboration among the worker team can be revealed. Furthermore, a sequential AI model is applied to learn to capture the characteristics of the team working. This AI model only requires training by 15 min of IoT data, then it can predict the current and next few days’ productions within 10% error. This research confirms that implementing the IoT architecture and applying the AI model enables instant manufacturing monitoring of labor-intensive manufacturing sites and accurate production forecasting.