Boyu Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Boyu Zhang has a total of 27 co-authors. Here are the top co-authors by number of shared publications:
- Aleksandar Vakanski
Department of Computer Science, University of Idaho, Idaho Falls, USA
6 shared publications - Min Xian
University of Idaho
5 shared publications Aleksandar Vakanski
University of Idaho
2 shared publicationsFuchang Gao
Department of Mathematics and Statistical Science, University of Idaho, Moscow, USA
2 shared publicationsLawhori Chakrabarti
University of Idaho
2 shared publicationsMin Xian
University of Idaho
2 shared publications
Latest Publications
- 9 October 2026
- Published by Springer Nature
- p. 524-539
Post-Hoc Understanding of Metaphor Processing in Decoder-Only Language Models via Conditional Scale Entropy
- 9 October 2026
- Published by Springer Nature
- p. 524-539
Post-Hoc Understanding of Metaphor Processing in Decoder-Only Language Models via Conditional Scale Entropy
- Open Access
- 23 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 182661-182674
GCSAM: Gradient Centralized Sharpness Aware Minimization
The generalization performance of deep neural networks (DNNs) is a critical factor in achieving robust model behavior on unseen data. Recent studies have highlighted the importance of sharpness-based measures in promoting generalization by encouraging convergence to flatter minima. Among these approaches, Sharpness-Aware Minimization (SAM) has emerged as an effective optimization technique for reducing the sharpness of the loss landscape, thereby improving generalization. However, SAM’s computational overhead and sensitivity to noisy gradients limit its scalability and efficiency. To address these challenges, we propose Gradient-Centralized Sharpness-Aware Minimization (GCSAM), which incorporates Gradient Centralization (GC) to stabilize gradients and accelerate convergence. GCSAM normalizes gradients before the ascent step, reducing noise and variance, and improving stability during training. Our evaluations on both general vision benchmarks (CIFAR-10, CIFAR-100) and critical medical imaging datasets (breast ultrasound and COVID-19 chest X-rays) demonstrate that GCSAM consistently outperforms SAM and the Adam optimizer in terms of generalization and computational efficiency. These results highlight GCSAM’s potential for improving reliability in domains where robust generalization is essential, particularly in medical image analysis. Our code is available at https://github.com/mhassann22/GCSAM.
- 23 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 182661-182674
Open AccessGCSAM: Gradient Centralized Sharpness Aware Minimization
The generalization performance of deep neural networks (DNNs) is a critical factor in achieving robust model behavior on unseen data. Recent studies have highlighted the importance of sharpness-based measures in promoting generalization by encouraging convergence to flatter minima. Among these approaches, Sharpness-Aware Minimization (SAM) has emerged as an effective optimization technique for reducing the sharpness of the loss landscape, thereby improving generalization. However, SAM’s computational overhead and sensitivity to noisy gradients limit its scalability and efficiency. To address these challenges, we propose Gradient-Centralized Sharpness-Aware Minimization (GCSAM), which incorporates Gradient Centralization (GC) to stabilize gradients and accelerate convergence. GCSAM normalizes gradients before the ascent step, reducing noise and variance, and improving stability during training. Our evaluations on both general vision benchmarks (CIFAR-10, CIFAR-100) and critical medical imaging datasets (breast ultrasound and COVID-19 chest X-rays) demonstrate that GCSAM consistently outperforms SAM and the Adam optimizer in terms of generalization and computational efficiency. These results highlight GCSAM’s potential for improving reliability in domains where robust generalization is essential, particularly in medical image analysis. Our code is available at https://github.com/mhassann22/GCSAM.
- 1 October 2025
- Published by Springer Nature
- p. 99-115
Data-Aware Customization of Activation Functions Reduces Neural Network Error
- 1 October 2025
- Published by Springer Nature
- p. 99-115
Data-Aware Customization of Activation Functions Reduces Neural Network Error
- Open Access
- 23 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 136320-136335
Automated Skin Cancer Report Generation via a Knowledge-Distilled Vision-Language Model
Artificial Intelligence (AI)’s capacity to analyze dermoscopic images promises a groundbreaking leap in skin cancer diagnostics, offering exceptional accuracy and an effortlessly non-invasive image acquisition process. However, this immense potential, which has ignited widespread research enthusiasm, is critically undermined due to the lack of transparency and interpretability. The automated generation of articulate and comprehensive diagnostic reports will bridge this critical gap by not only illuminate the AI’s diagnostic rational but also substantially reduce the demanding workload of the medical professionals. This study presents a multimodal vision-language model (VLM) trained using a two-stage knowledge distillation (KD) framework to generate structured medical reports from dermoscopic images, with descriptive features based on the 7-point melanoma checklist. The reports are organized into clinically relevant sections—Findings, Impression, and Differential Diagnosis—aligned with dermatological standards. Experimental evaluation demonstrates the system’s ability to produce accurate and interpretable reports. Human feedback from a medical professional, assessing clinical relevance, completeness, and interpretability, supports the utility of the generated reports, while computational metrics validate their accuracy and alignment with reference pseudo-reports, achieving a SacreBLEU score of 55.59, a ROUGE-1 score of 0.5438, a ROUGE-L score of 0.3828, and a BERTScore F1 of 0.9025. These findings underscore the model’s ability to generalize effectively to unseen data, enabled by its multimodal design, clinical alignment, and explainability.
- 23 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 136320-136335
Open AccessAutomated Skin Cancer Report Generation via a Knowledge-Distilled Vision-Language Model
Artificial Intelligence (AI)’s capacity to analyze dermoscopic images promises a groundbreaking leap in skin cancer diagnostics, offering exceptional accuracy and an effortlessly non-invasive image acquisition process. However, this immense potential, which has ignited widespread research enthusiasm, is critically undermined due to the lack of transparency and interpretability. The automated generation of articulate and comprehensive diagnostic reports will bridge this critical gap by not only illuminate the AI’s diagnostic rational but also substantially reduce the demanding workload of the medical professionals. This study presents a multimodal vision-language model (VLM) trained using a two-stage knowledge distillation (KD) framework to generate structured medical reports from dermoscopic images, with descriptive features based on the 7-point melanoma checklist. The reports are organized into clinically relevant sections—Findings, Impression, and Differential Diagnosis—aligned with dermatological standards. Experimental evaluation demonstrates the system’s ability to produce accurate and interpretable reports. Human feedback from a medical professional, assessing clinical relevance, completeness, and interpretability, supports the utility of the generated reports, while computational metrics validate their accuracy and alignment with reference pseudo-reports, achieving a SacreBLEU score of 55.59, a ROUGE-1 score of 0.5438, a ROUGE-L score of 0.3828, and a BERTScore F1 of 0.9025. These findings underscore the model’s ability to generalize effectively to unseen data, enabled by its multimodal design, clinical alignment, and explainability.
- Open Access
- 9 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 82972-82985
Do Sharpness-Based Optimizers Improve Generalization in Medical Image Analysis?
Effective clinical deployment of deep learning models in healthcare demands high generalization performance to ensure accurate diagnosis and treatment planning. In recent years, significant research has focused on improving the generalization of deep learning models by regularizing the sharpness of the loss landscape. Among the optimization approaches that explicitly minimize sharpness, Sharpness-Aware Minimization (SAM) has shown potential in enhancing generalization performance on general domain image datasets. This success has led to the development of several advanced sharpness-based algorithms aimed at addressing the limitations of SAM, such as Adaptive SAM, Surrogate-Gap SAM, Weighted SAM, and Curvature Regularized SAM. These sharpness-based optimizers have shown improvements in model generalization compared to conventional stochastic gradient descent optimizers and their variants on general domain image datasets, but they have not been thoroughly evaluated on medical images. This work provides a review of recent sharpness-based methods for improving the generalization of deep learning networks and evaluates the methods’ performance on three medical image datasets, including breast ultrasound, chest X-ray, and colon histopathology images. Our findings indicate that the initial SAM method successfully enhances the generalization of various deep learning models. While Adaptive SAM improves generalization of convolutional neural networks, it fails to do so for vision transformers. Other sharpness-based optimizers, however, do not demonstrate consistent results. The results reveal that contrary to findings in the non-medical domain, SAM is the only recommended sharpness-based optimizer that consistently improves generalization in medical image analysis, and further research is necessary to refine the variants of SAM to enhance generalization performance in this field.
- 9 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 82972-82985
Open AccessDo Sharpness-Based Optimizers Improve Generalization in Medical Image Analysis?
Effective clinical deployment of deep learning models in healthcare demands high generalization performance to ensure accurate diagnosis and treatment planning. In recent years, significant research has focused on improving the generalization of deep learning models by regularizing the sharpness of the loss landscape. Among the optimization approaches that explicitly minimize sharpness, Sharpness-Aware Minimization (SAM) has shown potential in enhancing generalization performance on general domain image datasets. This success has led to the development of several advanced sharpness-based algorithms aimed at addressing the limitations of SAM, such as Adaptive SAM, Surrogate-Gap SAM, Weighted SAM, and Curvature Regularized SAM. These sharpness-based optimizers have shown improvements in model generalization compared to conventional stochastic gradient descent optimizers and their variants on general domain image datasets, but they have not been thoroughly evaluated on medical images. This work provides a review of recent sharpness-based methods for improving the generalization of deep learning networks and evaluates the methods’ performance on three medical image datasets, including breast ultrasound, chest X-ray, and colon histopathology images. Our findings indicate that the initial SAM method successfully enhances the generalization of various deep learning models. While Adaptive SAM improves generalization of convolutional neural networks, it fails to do so for vision transformers. Other sharpness-based optimizers, however, do not demonstrate consistent results. The results reveal that contrary to findings in the non-medical domain, SAM is the only recommended sharpness-based optimizer that consistently improves generalization in medical image analysis, and further research is necessary to refine the variants of SAM to enhance generalization performance in this field.
- Open Access
- 11 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 37724-37736
Comparative Analysis of Multi-Omics Integration Using Graph Neural Networks for Cancer Classification
Recent studies on integrating multiple omics data highlighted the potential to advance our understanding of the cancer disease process. Computational models based on graph neural networks and attention-based architectures have demonstrated promising results for cancer classification due to their ability to model complex relationships among biological entities. However, challenges related to addressing the high dimensionality and complexity in integrating multi-omics data, as well as in constructing graph structures that effectively capture the interactions between nodes, remain active areas of research. This study evaluates graph neural network architectures for multi-omics (MO) data integration based on graph-convolutional networks (GCN), graph-attention networks (GAT), and graph-transformer networks (GTN). Differential gene expression and LASSO (Least Absolute Shrinkage and Selection Operator) regression are employed for reducing the omics data dimensionality and feature selection; hence, the developed models are referred to as LASSO-MOGCN, LASSO-MOGAT, and LASSO-MOGTN. Graph structures constructed using sample correlation matrices and protein-protein interaction networks are investigated. Experimental validation is performed with a dataset of 8,464 samples from 31 cancer types and normal tissue, comprising messenger-RNA, micro-RNA, and DNA methylation data. The results show that the models integrating multi-omics data outperformed the models trained on single omics data, where LASSO-MOGAT achieved the best overall performance, with an accuracy of 95.9%. The findings also suggest that correlation-based graph structures enhance the models’ ability to identify shared cancer-specific signatures across patients in comparison to protein-protein interaction networks-based graph structures. The code and data used in this study are available in the link (https://github.com/FadiAlharbi2024/Graph_Based_Architecture.git).
- 11 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 37724-37736
Open AccessComparative Analysis of Multi-Omics Integration Using Graph Neural Networks for Cancer Classification
Recent studies on integrating multiple omics data highlighted the potential to advance our understanding of the cancer disease process. Computational models based on graph neural networks and attention-based architectures have demonstrated promising results for cancer classification due to their ability to model complex relationships among biological entities. However, challenges related to addressing the high dimensionality and complexity in integrating multi-omics data, as well as in constructing graph structures that effectively capture the interactions between nodes, remain active areas of research. This study evaluates graph neural network architectures for multi-omics (MO) data integration based on graph-convolutional networks (GCN), graph-attention networks (GAT), and graph-transformer networks (GTN). Differential gene expression and LASSO (Least Absolute Shrinkage and Selection Operator) regression are employed for reducing the omics data dimensionality and feature selection; hence, the developed models are referred to as LASSO-MOGCN, LASSO-MOGAT, and LASSO-MOGTN. Graph structures constructed using sample correlation matrices and protein-protein interaction networks are investigated. Experimental validation is performed with a dataset of 8,464 samples from 31 cancer types and normal tissue, comprising messenger-RNA, micro-RNA, and DNA methylation data. The results show that the models integrating multi-omics data outperformed the models trained on single omics data, where LASSO-MOGAT achieved the best overall performance, with an accuracy of 95.9%. The findings also suggest that correlation-based graph structures enhance the models’ ability to identify shared cancer-specific signatures across patients in comparison to protein-protein interaction networks-based graph structures. The code and data used in this study are available in the link (https://github.com/FadiAlharbi2024/Graph_Based_Architecture.git).
- Open Access
Comparative Analysis of Multi-Omics Integration Using Advanced Graph Neural Networks for Cancer Classification
Multi-omics data is increasingly being utilized to advance computational methods for cancer classification. However, multi-omics data integration poses significant challenges due to the high dimensionality, data complexity, and distinct characteristics of various omics types. This study addresses these challenges and evaluates three graph neural network architectures for multi-omics (MO) integration based on graph-convolutional networks (GCN), graph-attention networks (GAT), and graph-transformer networks (GTN) for classifying 31 cancer types and normal tissues. To address the high-dimensionality of multi-omics data, we employed LASSO (Least Absolute Shrinkage and Selection Operator) regression for feature selection, leading to the creation of LASSO-MOGCN, LASSO-MOGAT, and LASSO-MOTGN models. Graph structures for the networks were constructed using gene correlation matrices and protein-protein interaction networks for multi-omics integration of messenger-RNA, micro-RNA, and DNA methylation data. Such data integration enables the networks to dynamically focus on important relationships between biological entities, improving both model performance and interpretability. Among the models, LASSO-MOGAT with a correlation-based graph structure achieved state-of-the-art accuracy (95.9%) and outperformed the LASSO-MOGCN and LASSO-MOTGN models in terms of precision, recall, and F1-score. Our findings demonstrate that integrating multi-omics data in graph-based architectures enhances cancer classification performance by uncovering distinct molecular patterns that contribute to a better understanding of cancer biology and potential biomarkers for disease progression.
Open AccessComparative Analysis of Multi-Omics Integration Using Advanced Graph Neural Networks for Cancer Classification
Multi-omics data is increasingly being utilized to advance computational methods for cancer classification. However, multi-omics data integration poses significant challenges due to the high dimensionality, data complexity, and distinct characteristics of various omics types. This study addresses these challenges and evaluates three graph neural network architectures for multi-omics (MO) integration based on graph-convolutional networks (GCN), graph-attention networks (GAT), and graph-transformer networks (GTN) for classifying 31 cancer types and normal tissues. To address the high-dimensionality of multi-omics data, we employed LASSO (Least Absolute Shrinkage and Selection Operator) regression for feature selection, leading to the creation of LASSO-MOGCN, LASSO-MOGAT, and LASSO-MOTGN models. Graph structures for the networks were constructed using gene correlation matrices and protein-protein interaction networks for multi-omics integration of messenger-RNA, micro-RNA, and DNA methylation data. Such data integration enables the networks to dynamically focus on important relationships between biological entities, improving both model performance and interpretability. Among the models, LASSO-MOGAT with a correlation-based graph structure achieved state-of-the-art accuracy (95.9%) and outperformed the LASSO-MOGCN and LASSO-MOTGN models in terms of precision, recall, and F1-score. Our findings demonstrate that integrating multi-omics data in graph-based architectures enhances cancer classification performance by uncovering distinct molecular patterns that contribute to a better understanding of cancer biology and potential biomarkers for disease progression.
- Open Access
- 25 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 79480-79494
BI-RADS-NET-V2: A Composite Multi-Task Neural Network for Computer-Aided Diagnosis of Breast Cancer in Ultrasound Images With Semantic and Quantitative Explanations
Computer-aided Diagnosis (CADx) based on explainable artificial intelligence (XAI) can gain the trust of radiologists and effectively improve diagnosis accuracy and consultation efficiency. This paper proposes BI-RADS-Net-V2, a novel machine learning approach for fully automatic breast cancer diagnosis in ultrasound images. The BI-RADS-Net-V2 can accurately distinguish malignant tumors from benign ones and provides both semantic and quantitative explanations. The explanations are provided in terms of clinically proven morphological features used by clinicians for diagnosis and reporting mass findings, i.e., Breast Imaging Reporting and Data System (BI-RADS). The experiments on 1,192 Breast Ultrasound (BUS) images indicate that the proposed method improves the diagnosis accuracy by taking full advantage of the medical knowledge in BI-RADS while providing both semantic and quantitative explanations for the decision.
- 25 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 79480-79494
Open AccessBI-RADS-NET-V2: A Composite Multi-Task Neural Network for Computer-Aided Diagnosis of Breast Cancer in Ultrasound Images With Semantic and Quantitative Explanations
Computer-aided Diagnosis (CADx) based on explainable artificial intelligence (XAI) can gain the trust of radiologists and effectively improve diagnosis accuracy and consultation efficiency. This paper proposes BI-RADS-Net-V2, a novel machine learning approach for fully automatic breast cancer diagnosis in ultrasound images. The BI-RADS-Net-V2 can accurately distinguish malignant tumors from benign ones and provides both semantic and quantitative explanations. The explanations are provided in terms of clinically proven morphological features used by clinicians for diagnosis and reporting mass findings, i.e., Breast Imaging Reporting and Data System (BI-RADS). The experiments on 1,192 Breast Ultrasound (BUS) images indicate that the proposed method improves the diagnosis accuracy by taking full advantage of the medical knowledge in BI-RADS while providing both semantic and quantitative explanations for the decision.
- 3 September 2022
- Published by Springer Nature in SN Computer Science
- Vol. 3 (6) , 1-12
Contrastive Metric Learning for Lithium Super-ionic Conductor Screening
High-performance Li-ion battery significantly impacts modern society, and materials with high conductivity play critical roles in battery development. Machine learning (ML) technologies have rapidly changed the field in recent years. However, it is still challenging to predict the high conductors directly due to the lack of validated conductor samples. This paper presents a succinct but effective metric-learning framework for high conductor screening. The material structures are mapped to an optimized feature space using a Siamese network, and an instance-based method is used to classify the input sample. The experiments demonstrate that the proposed method could effectively extract knowledge from imbalanced data and has good performance and generalization ability.
- 3 September 2022
- Published by Springer Nature in SN Computer Science
- Vol. 3 (6) , 1-12
Contrastive Metric Learning for Lithium Super-ionic Conductor Screening
High-performance Li-ion battery significantly impacts modern society, and materials with high conductivity play critical roles in battery development. Machine learning (ML) technologies have rapidly changed the field in recent years. However, it is still challenging to predict the high conductors directly due to the lack of validated conductor samples. This paper presents a succinct but effective metric-learning framework for high conductor screening. The material structures are mapped to an optimized feature space using a Siamese network, and an instance-based method is used to classify the input sample. The experiments demonstrate that the proposed method could effectively extract knowledge from imbalanced data and has good performance and generalization ability.
- Open Access
- 9 June 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 62440-62449
Predicting the Materials Properties Using a 3D Graph Neural Network With Invariant Representation
Accurate prediction of physical properties is critical for discovering and designing novel materials. Machine learning technologies have attracted significant attention in the materials science community for their potential for large-scale screening. Graph Convolution Neural Network (GCNN) is one of the most successful machine learning methods because of its flexibility and effectiveness in describing 3D structural data. Most existing GCNN models focus on the topological structure but overly simplify the three-dimensional geometric structure. However, in materials science, the 3D-spatial distribution of atoms is crucial for determining the atomic states and interatomic forces. This paper proposes an adaptive GCNN with a novel convolution mechanism that simultaneously models atomic interactions among all neighboring atoms in three-dimensional space. We apply the proposed model to two distinctly challenging materials properties prediction problems. The first is Henry’s constant for gas adsorption in Metal-Organic Frameworks (MOFs), which is notoriously difficult because of its high sensitivity to atomic configurations. The second is the ion conductivity in solid-state crystal materials, which is difficult because of the few labeled data available for training. The new model outperforms existing graph-based models on both data sets, suggesting that the critical three-dimensional geometric information is indeed captured.
- 9 June 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 62440-62449
Open AccessPredicting the Materials Properties Using a 3D Graph Neural Network With Invariant Representation
Accurate prediction of physical properties is critical for discovering and designing novel materials. Machine learning technologies have attracted significant attention in the materials science community for their potential for large-scale screening. Graph Convolution Neural Network (GCNN) is one of the most successful machine learning methods because of its flexibility and effectiveness in describing 3D structural data. Most existing GCNN models focus on the topological structure but overly simplify the three-dimensional geometric structure. However, in materials science, the 3D-spatial distribution of atoms is crucial for determining the atomic states and interatomic forces. This paper proposes an adaptive GCNN with a novel convolution mechanism that simultaneously models atomic interactions among all neighboring atoms in three-dimensional space. We apply the proposed model to two distinctly challenging materials properties prediction problems. The first is Henry’s constant for gas adsorption in Metal-Organic Frameworks (MOFs), which is notoriously difficult because of its high sensitivity to atomic configurations. The second is the ion conductivity in solid-state crystal materials, which is difficult because of the few labeled data available for training. The new model outperforms existing graph-based models on both data sets, suggesting that the critical three-dimensional geometric information is indeed captured.