Kuan Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Kuan Li has a total of 55 co-authors. Here are the top co-authors by number of shared publications:
- Jianping Yin
Dongguan University of Technology
7 shared publications Jianping Yin
Dongguan University of Technology
5 shared publications- Fuyong Zhang
Dongguan University of Technology
2 shared publications Haoyuan Chen
Dongguan University of Technology
2 shared publications- Haoyuan Chen
College of Intelligence Science and Technology, National University of Defense Technology, Changsha 410073, China
2 shared publications Pin Xu
Dongguan University of Technology
2 shared publications
Latest Publications
- 12 October 2026
- Published by Elsevier in Neural Networks
- Vol. 206 (Pt A) , 109754
Enhancing out-of-distribution detection via local-image set
- 12 October 2026
- Published by Elsevier in Neural Networks
- Vol. 206 (Pt A) , 109754
Enhancing out-of-distribution detection via local-image set
- 14 July 2026
- Published by Springer Nature
- p. 388-399
OCP-YOLO: Occlusion-Aware and Cross-modal Progressive Fusion for UAV Small Object Detection
- 14 July 2026
- Published by Springer Nature
- p. 388-399
OCP-YOLO: Occlusion-Aware and Cross-modal Progressive Fusion for UAV Small Object Detection
- Open Access
- 15 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 15-27
CausalTuner: Feature-Aware Causal Guidance for Compiler Auto-tuning
Modern compilers like LLVM and GCC provide hundreds of optimization options (e.g., flags or passes), yet their fixed, predefined sequences (e.g., -O3) often fail to exploit the full performance potential of specific programs. Search-based auto-tuning has emerged to address this pass selection and ordering problem — known as the phase ordering problem. Existing approaches typically reduce search complexity by identifying critical flags or employing localized group-based mutations. However, these methods often remain feature-agnostic or rely on manually designed static mappings during the online search. Consequently, they fail to adaptively link program features to optimization logic. Furthermore, they are prone to being misled by spurious correlations and noise inherent in sparse performance data. To address this limitation, we propose CausalTuner, a feature-aware framework that injects causal rules into a Two-Phase Rule-Injected Search Engine to guide the optimization search. By identifying performance-critical pass subsequences and mining the causal dependencies between program features and optimization effectiveness, CausalTuner effectively transforms the search process from heuristic-driven to causal-guided. We evaluate CausalTuner on a diverse range of benchmarks, including cBench, Polybench, SPEC CPU 2017, and llama.cpp. CausalTuner consistently outperforms existing autotuning methods, discovering superior optimization configurations more efficiently with fewer search iterations.
- 15 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 15-27
Open AccessCausalTuner: Feature-Aware Causal Guidance for Compiler Auto-tuning
Modern compilers like LLVM and GCC provide hundreds of optimization options (e.g., flags or passes), yet their fixed, predefined sequences (e.g., -O3) often fail to exploit the full performance potential of specific programs. Search-based auto-tuning has emerged to address this pass selection and ordering problem — known as the phase ordering problem. Existing approaches typically reduce search complexity by identifying critical flags or employing localized group-based mutations. However, these methods often remain feature-agnostic or rely on manually designed static mappings during the online search. Consequently, they fail to adaptively link program features to optimization logic. Furthermore, they are prone to being misled by spurious correlations and noise inherent in sparse performance data. To address this limitation, we propose CausalTuner, a feature-aware framework that injects causal rules into a Two-Phase Rule-Injected Search Engine to guide the optimization search. By identifying performance-critical pass subsequences and mining the causal dependencies between program features and optimization effectiveness, CausalTuner effectively transforms the search process from heuristic-driven to causal-guided. We evaluate CausalTuner on a diverse range of benchmarks, including cBench, Polybench, SPEC CPU 2017, and llama.cpp. CausalTuner consistently outperforms existing autotuning methods, discovering superior optimization configurations more efficiently with fewer search iterations.
- 25 July 2025
- Published by Springer Nature
- p. 285-296
Using an Improved Lightweight YOLOv11 Model for Fuzzy Image Object Detection
- 25 July 2025
- Published by Springer Nature
- p. 285-296
Using an Improved Lightweight YOLOv11 Model for Fuzzy Image Object Detection
- Open Access
- 18 April 2025
- Published by Association for Computing Machinery (ACM)
- p. 852-859
Perci-AD: Automated Multi-Precision Optimization for High-Performance Computing
- 18 April 2025
- Published by Association for Computing Machinery (ACM)
- p. 852-859
Open AccessPerci-AD: Automated Multi-Precision Optimization for High-Performance Computing
- 2 December 2024
- Published by Springer Nature
- p. 344-358
Synthetic Images with Dense Annotations and Ensemble Learning for DFU Segmentation
- 2 December 2024
- Published by Springer Nature
- p. 344-358
Synthetic Images with Dense Annotations and Ensemble Learning for DFU Segmentation
- Open Access
- 18 November 2024
- Published by Springer Nature in Scientific Reports
- Vol. 14 (1) , 1-14
Exploring feature sparsity for out-of-distribution detection
Out-of-distribution (OOD) detection is a crucial problem in practice, especially, for the safe deployment of machine learning models in industrial settings. Previous work has used free energy as a score function and proposed a fine-tuning method that utilized OOD data in the training phase of the classification model, which achieves a higher performance on the OOD detection task compared with traditional methods. One key drawback, however, is that the loss function parameters are highly dependent on involved datasets, which means it cannot be dynamically adapted and implemented in others settings; in other words, the general ability of the energy score is considerably limited. In this work, our point of departure is to enlarge distinguishability between in-distribution features and OOD data. Consequently, we present a simple yet effective sparsity-regularized (SR) tuning framework for this purpose. Our framework has two types of workflows depending on if external OOD data is available, the complexity of the original training loss is sharply reduced by adopting this modification, meanwhile, the adapted ability and detection performance are enhanced. Also, we contribute a mini dataset as a light and efficient alternative of the previous large-scale one. In the experiments, we verify the effectiveness of our framework in a wide range of typical datasets along with common network architectures.
- 18 November 2024
- Published by Springer Nature in Scientific Reports
- Vol. 14 (1) , 1-14
Open AccessExploring feature sparsity for out-of-distribution detection
Out-of-distribution (OOD) detection is a crucial problem in practice, especially, for the safe deployment of machine learning models in industrial settings. Previous work has used free energy as a score function and proposed a fine-tuning method that utilized OOD data in the training phase of the classification model, which achieves a higher performance on the OOD detection task compared with traditional methods. One key drawback, however, is that the loss function parameters are highly dependent on involved datasets, which means it cannot be dynamically adapted and implemented in others settings; in other words, the general ability of the energy score is considerably limited. In this work, our point of departure is to enlarge distinguishability between in-distribution features and OOD data. Consequently, we present a simple yet effective sparsity-regularized (SR) tuning framework for this purpose. Our framework has two types of workflows depending on if external OOD data is available, the complexity of the original training loss is sharply reduced by adopting this modification, meanwhile, the adapted ability and detection performance are enhanced. Also, we contribute a mini dataset as a light and efficient alternative of the previous large-scale one. In the experiments, we verify the effectiveness of our framework in a wide range of typical datasets along with common network architectures.
- Open Access
- 30 September 2024
- Published by MDPI AG in Mathematics
- Vol. 12 (19) , 3061
A Hybrid Framework for Referring Image Segmentation: Dual-Decoder Model with SAM Complementation
In the realm of human–robot interaction, the integration of visual and verbal cues has become increasingly significant. This paper focuses on the challenges and advancements in referring image segmentation (RIS), a task that involves segmenting images based on textual descriptions. Traditional approaches to RIS have primarily focused on pixel-level classification. These methods, although effective, often overlook the interconnectedness of pixels, which can be crucial for interpreting complex visual scenes. Furthermore, while the PolyFormer model has shown impressive performance in RIS, its large number of parameters and high training data requirements pose significant challenges. These factors restrict its adaptability and optimization on standard consumer hardware, hindering further enhancements in subsequent research. Addressing these issues, our study introduces a novel two-branch decoder framework with SAM (segment anything model) for RIS. This framework incorporates an MLP decoder and a KAN decoder with a multi-scale feature fusion module, enhancing the model’s capacity to discern fine details within images. The framework’s robustness is further bolstered by an ensemble learning strategy that consolidates the insights from both the MLP and KAN decoder branches. More importantly, we collect the segmentation target edge coordinates and bounding box coordinates as input cues for the SAM model. This strategy leverages SAM’s zero-sample learning capabilities to refine and optimize the segmentation outcomes. Our experimental findings, based on the widely recognized RefCOCO, RefCOCO+, and RefCOCOg datasets, confirm the effectiveness of this method. The results not only achieve state-of-the-art (SOTA) performance in segmentation but are also supported by ablation studies that highlight the contributions of each component to the overall improvement in performance.
- 30 September 2024
- Published by MDPI AG in Mathematics
- Vol. 12 (19) , 3061
Open AccessA Hybrid Framework for Referring Image Segmentation: Dual-Decoder Model with SAM Complementation
In the realm of human–robot interaction, the integration of visual and verbal cues has become increasingly significant. This paper focuses on the challenges and advancements in referring image segmentation (RIS), a task that involves segmenting images based on textual descriptions. Traditional approaches to RIS have primarily focused on pixel-level classification. These methods, although effective, often overlook the interconnectedness of pixels, which can be crucial for interpreting complex visual scenes. Furthermore, while the PolyFormer model has shown impressive performance in RIS, its large number of parameters and high training data requirements pose significant challenges. These factors restrict its adaptability and optimization on standard consumer hardware, hindering further enhancements in subsequent research. Addressing these issues, our study introduces a novel two-branch decoder framework with SAM (segment anything model) for RIS. This framework incorporates an MLP decoder and a KAN decoder with a multi-scale feature fusion module, enhancing the model’s capacity to discern fine details within images. The framework’s robustness is further bolstered by an ensemble learning strategy that consolidates the insights from both the MLP and KAN decoder branches. More importantly, we collect the segmentation target edge coordinates and bounding box coordinates as input cues for the SAM model. This strategy leverages SAM’s zero-sample learning capabilities to refine and optimize the segmentation outcomes. Our experimental findings, based on the widely recognized RefCOCO, RefCOCO+, and RefCOCOg datasets, confirm the effectiveness of this method. The results not only achieve state-of-the-art (SOTA) performance in segmentation but are also supported by ablation studies that highlight the contributions of each component to the overall improvement in performance.
- Open Access
- 26 September 2024
- Published by MDPI AG in Mathematics
- Vol. 12 (19) , 2996
MS-UNet: Multi-Scale Nested UNet for Medical Image Segmentation with Few Training Data Based on an ELoss and Adaptive Denoising Method
Traditional U-shape segmentation models can achieve excellent performance with an elegant structure. However, the single-layer decoder structure of U-Net or SwinUnet is too “thin” to exploit enough information, resulting in large semantic differences between the encoder and decoder parts. Things get worse in the field of medical image processing, where annotated data are more difficult to obtain than other tasks. Based on this observation, we propose a U-like model named MS-UNet with a plug-and-play adaptive denoising module and ELoss for the medical image segmentation task in this study. Instead of the single-layer U-Net decoder structure used in Swin-UNet and TransUNet, we specifically designed a multi-scale nested decoder based on the Swin Transformer for U-Net. The proposed multi-scale nested decoder structure allows for the feature mapping between the decoder and encoder to be semantically closer, thus enabling the network to learn more detailed features. In addition, ELoss could improve the attention of the model to the segmentation edges, and the plug-and-play adaptive denoising module could prevent the model from learning the wrong features without losing detailed information. The experimental results show that MS-UNet could effectively improve network performance with more efficient feature learning capability and exhibit more advanced performance, especially in the extreme case with a small amount of training data. Furthermore, the proposed ELoss and denoising module not only significantly enhance the segmentation performance of MS-UNet but can also be applied individually to other models.
- 26 September 2024
- Published by MDPI AG in Mathematics
- Vol. 12 (19) , 2996
Open AccessMS-UNet: Multi-Scale Nested UNet for Medical Image Segmentation with Few Training Data Based on an ELoss and Adaptive Denoising Method
Traditional U-shape segmentation models can achieve excellent performance with an elegant structure. However, the single-layer decoder structure of U-Net or SwinUnet is too “thin” to exploit enough information, resulting in large semantic differences between the encoder and decoder parts. Things get worse in the field of medical image processing, where annotated data are more difficult to obtain than other tasks. Based on this observation, we propose a U-like model named MS-UNet with a plug-and-play adaptive denoising module and ELoss for the medical image segmentation task in this study. Instead of the single-layer U-Net decoder structure used in Swin-UNet and TransUNet, we specifically designed a multi-scale nested decoder based on the Swin Transformer for U-Net. The proposed multi-scale nested decoder structure allows for the feature mapping between the decoder and encoder to be semantically closer, thus enabling the network to learn more detailed features. In addition, ELoss could improve the attention of the model to the segmentation edges, and the plug-and-play adaptive denoising module could prevent the model from learning the wrong features without losing detailed information. The experimental results show that MS-UNet could effectively improve network performance with more efficient feature learning capability and exhibit more advanced performance, especially in the extreme case with a small amount of training data. Furthermore, the proposed ELoss and denoising module not only significantly enhance the segmentation performance of MS-UNet but can also be applied individually to other models.
- Open Access
- 12 March 2024
- Published by MDPI AG in Mathematics
- Vol. 12 (6) , 834
Improving Adversarial Robustness of Ensemble Classifiers by Diversified Feature Selection and Stochastic Aggregation
Learning-based classifiers are found to be vulnerable to attacks by adversarial samples. Some works suggested that ensemble classifiers tend to be more robust than single classifiers against evasion attacks. However, recent studies have shown that this is not necessarily the case under more realistic settings of black-box attacks. In this paper, we propose a novel ensemble approach to improve the robustness of classifiers against evasion attacks by using diversified feature selection and a stochastic aggregation strategy. Our proposed scheme includes three stages. Firstly, the adversarial feature selection algorithm is used to select a feature each time that can trade-offbetween classification accuracy and robustness, and add it to the feature vector bank. Secondly, each feature vector in the bank is used to train a base classifier and is added to the base classifier bank. Finally, m classifiers from the classifier bank are randomly selected for decision-making. In this way, it can cause each classifier in the base classifier bank to have good performance in terms of classification accuracy and robustness, and it also makes it difficult to estimate the gradients of the ensemble accurately. Thus, the robustness of classifiers can be improved without reducing the classification accuracy. Experiments performed using both Linear and Kernel SVMs on genuine datasets for spam filtering, malware detection, and handwritten digit recognition demonstrate that our proposed approach significantly improves the classifiers’ robustness against evasion attacks.
- 12 March 2024
- Published by MDPI AG in Mathematics
- Vol. 12 (6) , 834
Open AccessImproving Adversarial Robustness of Ensemble Classifiers by Diversified Feature Selection and Stochastic Aggregation
Learning-based classifiers are found to be vulnerable to attacks by adversarial samples. Some works suggested that ensemble classifiers tend to be more robust than single classifiers against evasion attacks. However, recent studies have shown that this is not necessarily the case under more realistic settings of black-box attacks. In this paper, we propose a novel ensemble approach to improve the robustness of classifiers against evasion attacks by using diversified feature selection and a stochastic aggregation strategy. Our proposed scheme includes three stages. Firstly, the adversarial feature selection algorithm is used to select a feature each time that can trade-offbetween classification accuracy and robustness, and add it to the feature vector bank. Secondly, each feature vector in the bank is used to train a base classifier and is added to the base classifier bank. Finally, m classifiers from the classifier bank are randomly selected for decision-making. In this way, it can cause each classifier in the base classifier bank to have good performance in terms of classification accuracy and robustness, and it also makes it difficult to estimate the gradients of the ensemble accurately. Thus, the robustness of classifiers can be improved without reducing the classification accuracy. Experiments performed using both Linear and Kernel SVMs on genuine datasets for spam filtering, malware detection, and handwritten digit recognition demonstrate that our proposed approach significantly improves the classifiers’ robustness against evasion attacks.