Wentao Wu

Wentao Wu

Auto-generated profile, based on publicly available information

Wentao Wu

China North Vehicle Research Institute, Beijing, China

Research Focus & Metrics

Academic Career Length

1Year (2026-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
10 July 2026 in ACM Transactions on Software Engineering and Methodology
Verified
Journal Article
Open access
HLHaoran Li
Haoran Li
SWShihai Wang
Shihai Wang
BLBin Liu
Bin Liu
TSTengfei Shi
Tengfei Shi
WWWentao Wu
Wentao Wu
YLYu Liu
Yu Liu
Although Deep Neural Networks (DNNs) have achieved significant success across various industries, their erroneous behaviors have raised widespread concerns. In safety-critical domains where DNNs are deployed, certain types of specific misclassifications (i.e., instances of one specific class being misclassified as another) can lead to catastrophic consequences and urgently need to be corrected. However, existing repair approaches often neglect the adverse impact of general feature extraction patterns on distinguishing specific classes, resulting in limited correction of specific misclassifications. Furthermore, most approaches correct errors by directly adjusting the global decision boundary, which inevitably disturbs the well-trained boundaries and leads to significant accuracy loss. In this article, we propose SMiR, an effective repair framework for addressing specific misclassifications with marginal accuracy loss. Rather than patching the original model, our method localizes and recovers the distorted feature dimensions to mitigate their negative impact on the distinction between specific classes. Moreover, SMiR narrows the scope of repair to the filtered suspect set, protecting other well-trained decision boundaries from disturbance. On average, our approach repairs 27% \(\sim\) 36.2% more specific misclassifications than the baselines, while introducing only 7.8% \(\sim\) 70.9% as many new errors. Besides, SMiR maintains state-of-the-art performance on safety-critical datasets.

8 May 2026 in IEEE Open Journal of the Computer Society
Verified
Journal Article
Open access
WWWentao Wu
Wentao Wu
XLXinan Liu
Xinan Liu
HDHong Du
Hong Du
YLYiding Liu
Yiding Liu
XLXingxin Li
Xingxin Li
RGRunzi Guo
Runzi Guo
In order to facilitate the timely discovery of potential defects by software testers, software defect prediction (SDP) activities are employed to allocate testing resources more efficiently. Among data mining techniques, association rule mining is a highly interpretable and high-performance algorithm that can effectively serve as a software defect predictor. On this basis, the association rule network analyzes the target item in the context environment, further enhancing the understanding of the knowledge reasoning process. In this study, we propose a SDP model using atomic association rule networks based on three distinct minimum support thresholds (AARN3). This model introduces two key innovations: (1) a cost-sensitive association rule mining framework with three minimum support thresholds to address class imbalance and enable embedded feature selection; and (2) a novel rule ranking and pruning method based on the Dijkstra algorithm within atomic association rule networks, which transforms conflicting rule path elimination into a shortest path problem. In the rule generation stage, the algorithm adopts a cost-sensitive strategy to ensure the generation of defective class rules under class imbalance conditions. In the rule ranking and pruning stage, we adopt atomic association rule networks to generate networks closely related to defect tendencies, identify conflicting paths in contextual relationships from the network structure, and transform the rule ranking into the shortest path problem, thereby eliminating potential erroneous rules in conflicting paths one by one. This not only increases the flexibility of rule pruning but also enhances the comprehensibility of knowledge mining. We conducted experimental comparisons between AARN3 and state-of-the-art baselines on 28 public datasets, and the results showed that AARN3 significantly outperformed other baseline models in $Balance$, $MCC$, and $Gmean$.

21 August 2026 in Big Data Research
Verified
Journal Article
Closed access
YSYuanxun Shao
Yuanxun Shao
WWWentao Wu
Wentao Wu
SWShihai Wang
Shihai Wang
BLBin Liu
Bin Liu
JYJingxiu Yao
Jingxiu Yao
WXWandong Xie
Wandong Xie

13 March 2026
Verified
Conference Paper
Open access
WWWentao Wu
Wentao Wu
XLXinan Liu
Xinan Liu
LYLihong Yin
Lihong Yin
XLXingxin Li
Xingxin Li
WXWandong Xie
Wandong Xie
BZBin Zhao
Bin Zhao