HWHuiyan Wang

Huiyan Wang

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Huiyan Wang has a total of 50 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    VGPO-MCTS: Distilling Step-Level Supervision from Value-Guided Tree Search for Mathematical Reasoning

    by Pin Wu, Yufei Zhu, Huiyan Wang

    Large language models (LLMs) are increasingly used in applied intelligent systems, but mid-sized models still lag on mathematical reasoning, partly because reliable step-level supervision is scarce. Many existing remedies rely on costly human annotation, stronger teacher models, or heavy training pipelines, which limits practical adoption. We propose VGPO-MCTS (Value-Guided Group-wise Policy Optimization over Monte Carlo Tree Search), a search-and-distillation framework that constructs reusable step-level supervision from datasets that provide only problems and final answers. VGPO-MCTS augments a frozen backbone with (i) a lightweight value model that scores candidate reasoning states formed by a reasoning prefix and its candidate next step, and (ii) a policy updated with parameter-efficient adaptation. During search, the value model guides tree expansion and selection, while verified outcomes are propagated backward to correct node utilities. The corrected search trees are then distilled into two complementary datasets: a value regression dataset for value learning and group-wise sibling candidate sets for GRPO-style policy optimization. Experiments on GSM8K and the MATH dataset with ChatGLM3-6B and SciGLM-6B show stable round-wise improvements in final-answer exact match under a lightweight adaptation setting. After three rounds of self-training, the proposed framework improves performance by about 6.3 percentage points on GSM8K and about 3.9 percentage points on MATH across the two backbones.

  • Cracking the Code of Backdoor Attacks With Confidence Consistency

    by Zhen Zhuang, Yijian Ding, Shuyu Zheng, Jun Shao, Donghai Zhu, Huiyan Wang

    It is widely believed that increasing the amount of training data enhances the intelligence of deep learning models, which in turn heightens dependence on external datasets. However, these datasets are susceptible to adversarial poisoning, allowing attackers to insert backdoors that trigger misclassifications. Although various defense strategies have been suggested, training-phase defenses (TPDs) appear most promising, as they can significantly lower the attack success rate (ASR) without greatly impacting model performance. Nevertheless, creating TPDs that achieve both high accuracy and low ASR is challenging due to two main issues: 1) Many solutions require additional clean samples that match the distribution of the poisoned dataset, which is not always practical in real-world scenarios; 2) most existing solutions have high computational costs, sometimes requiring five to ten times the expense of standard training, which severely limits their practical use. To tackle these challenges, we introduce Confidence Consistency Detection (CCD), an efficient and lightweight training-phase backdoor detection method. CCD is particularly advantageous in situations where clean data is scarce or unavailable, as it completely eliminates the need for external clean samples. Moreover, CCD significantly reduces computational costs to just 25% to 50% of existing solutions (1.7 times the standard training time), providing a notable improvement over current TPD methods. The core innovation of CCD lies in its ability to utilize the high confidence shown by backdoor samples during the early stages of model training for precise detection. Specifically, we initially train a model on a poisoned dataset for a few epochs, followed by intra-class loss fine-tuning to increase sensitivity to poisoned samples. We then create preliminary sets of poisoned and clean samples by assessing the consistency of confidence variations before and after fine-tuning. These sets guide the model training, enabling the detection of high-confidence poisoned samples. Extensive experiments demonstrate that CCD effectively reduces the attack success rate (ASR) to 1.43%, while having a negligible impact on the model’s clean accuracy. In detecting poisoned samples, CCD achieves a 99% true positive rate (TPR) and a 0.033% false positive rate (FPR), setting a new benchmark in the field.


  • MFF-SDD: A Bidirectional Guidance and Multiscale Multimodal Fusion Model for Small Defect Detection in Industrial Films

    by Huiyan Wang, Ruihao Peng, Ming Ying, Fashuai Li, Jiuyi Zhang, Xiaolan Li, Yan Tian, Guofeng Zhang

    Detecting small defects in industrial biaxially oriented polypropylene (BOPP) films is challenging due to their limited visual cues, low contrast, and high similarity to background textures. Existing deep learning-based methods, which rely solely on visual data, often suffer from high miss rates and poor classification accuracy in such complex scenarios. To address these limitations, we propose the multimodal fusion-based film small defect detection (MFF-SDD) model, specifically designed for small defect detection in BOPP film production. Unlike conventional language-image models that neglect the bidirectional interaction between modalities, MFF-SDD introduces the bidirectional guidance multimodal cross-fusion module, which enhances visual–textual integration through mutual guidance and attention mechanisms, enabling more effective foreground focus and background suppression. Furthermore, the text and image cross-modality neighbors multiscale fusion module employs multiscale cross-modal fusion to preserve fine-grained details and exploit complementary features from both modalities, improving detection accuracy and reducing misclassification of small defects. We also present the film small defect (FSD) dataset, comprising 10 385 annotated bounding boxes across seven defect categories. Experimental results show that MFF-SDD surpasses state-of-the-art methods by 2.31% in average precision and 3.75% in recall on the FSD dataset and achieves leading performance on public benchmarks, such as PASCAL VOC and TinyPerson. These findings demonstrate the effectiveness and robustness of our approach in multimodal small defect detection. Our dataset and code will be made publicly available upon acceptance of this article for publication.


  • Few-Shot Photovoltaic Film Defect Detection With Contextual Ensemble Language-Image Multimodal Network

    by Huiyan Wang, Ruihao Peng, Yiheng Zhu, Jiachen Li, Mengchu Zhou, Ming Ying, Shouguang Wang

    The automatic detection of defects in industrial photovoltaic film is crucial for ensuring the reliability of photovoltaic modules. Key challenges include limited defect samples, interclass feature similarities, and interference from complex backgrounds. Existing deep learning-based methods require large-scale datasets and focus solely on visual data, which limits their effectiveness in few-shot defect detection (FSDD). To address these challenges, we propose Contextual Ensemble Language-Image multimodal Network (CELIN), which enhances FSDD in photovoltaic films by incorporating textual information through prompt tuning. Unlike traditional language-image models that rely on single fixed text prompts, CELIN employs a position-aware context ensemble strategy to integrate position-specific prompt vectors, enabling the model to capture global contextual information and reduce background interference. In addition, a cross class mask method is introduced to differentiate between similar defect categories by blocking interclass interactions during attention computation, thereby minimizing misclassification. Extensive experiments on our own few-shot photovoltaic film defect dataset and various public benchmarks demonstrates that CELIN significantly outperforms existing methods.


  • Learning Dynamic Cloth Deformation for Virtual Try-on

    by Xianghui Chen, Huiyan Wang, Zhaopeng Cui, Guofeng Zhang

  • Clothes-Changing Person Re-Identification Using Color and Texture Invariant Representation

    by Huiyan Wang, Ming Ying, Huan Jiang, Qiufang Shen, Fashuai Li, Xiaolan Li

    This paper tackles the challenge of Clothes-Changing person Re-Identification (CC-ReID) through the dual lenses of model development and dataset creation. Unlike conventional person ReID, which presumes that individuals maintain consistent clothing, CC-ReID acknowledges the frequent changes in attire encountered in real-world scenarios. This task is particularly difficult due to the substantial variations in visual cues such as colors and textures that accompany changes in clothing. We propose a Multi-modal Assisted feature Learning Framework (MAL-F) designed to learn representations that are invariant to color and texture by utilizing RGB, grayscale, and contour images. MAL-F is a versatile framework that can be seamlessly integrated with existing CC-ReID models, significantly enhancing their accuracy. To further reduce the impact of clothing variations, we introduce a novel CC-ReID backbone named ResTNet, which combines ResNet with a Transformer. ResTNet features a Non-Clothing Salient Region (NC-SR) reinforcement transformer module that employs random clothing block drop and spatial activation mapping. These methods direct the model to emphasize key regions unrelated to clothing, enhancing these features with minimal computational overhead. This strategy substantially enhances CC-ReID performance and supports real-time video processing. Furthermore, recognizing the limitations of current small-scale CC-ReID datasets, we have developed two larger datasets: FBCC and Market1501-CC. Extensive experimental results demonstrate that our proposed methodology surpasses existing state-of-the-art methods for CC-ReID in terms of accuracy, robustness, and efficiency.


  • Open Access

    Variation rs9929218 and risk of the colorectal Cancer and adenomas: A meta-analysis

    by Huiyan Wang, Dongying Gu, Miao Yu, Yanjun Hu, Zhe Chen, Xinying Huo, Tao Yu, Jinfei Chen, Yang Zheng

    Backgrounds: Genome-wide association studies (GWAS) have identified multiple common CRC-related (colorectal cancer) SNPs (single nucleotide polymorphisms) including the Cadherin 1(CDH1) rs9929218 may act by increasing the risk of colorectal cancer, colorectal adenoma, or both. These studies, however, reported inconsistent associations. Methods: To derive a more accurate approximation of the connection, we carried out a meta-analysis of 12 published pieces of research including 11,590 controls and 8192 cases. We used odds ratios (ORs) and 95% confidence intervals (CIs) to evaluate the associations’ strength. Results: Meta-analysis implied considerable association between CRC and rs9929218 (OR = 1.21, 95%CI 1.04–1.42 for GG versus AA; OR = 1.22, 95%CI 1.05–1.42 for GG/AG versus AA). In the subgroup analyses, significantly increased risks were found among Europeans. Conclusions: In summary, our meta-analysis studies in different populations confirmed that SNP rs9929218 is significantly associated with CRC risk and that this variant may have a greater impact on Europeans.


  • Cow, yak, and camel milk diets differentially modulated the systemic immunity and fecal microbiota of rats

    by Yongping Wen, Qiuwen He, Jia Ding, Huiyan Wang, Qiangchuan Hou, Yi Zheng, Changkun Li, Yuzhu Ma, Heping Zhang, Lai-Yu Kwok