Scholarlink

<p>The growing significance of deep learning in software development has notably enhanced software quality by facilitating the efficient detection of defects, a major hurdle throughout the software development lifecycle. This study aims to identify the optimal model for uncovering defects in software projects. It introduces an intelligent model integrating antlion optimization (ALO) with a convolutional neural network (CNN). CNN is utilized for defect detection, while ALO is employed to fine-tune the network's weights. The paper proposes two intelligent models to tackle this research problem: (a) a basic convolutional neural network without parameter optimization and (b) a hybrid model that combines the CNN with ALO. The results indicate that the hybrid model significantly outperforms the basic CNN across several performance metrics, including area under the curve, sensitivity, specificity, accuracy, and error rate. Furthermore, the findings reveal that the hybrid model exceeds the performance of state-of-the-art methods from prior studies, including CNN, gated recurrent unit, and bidirectional long short-term memory, achieving accuracy improvements of 17.6%, 20.9%, and 24.4%, respectively. Additionally, the proposed model demonstrates superior performance compared to the deep forest method in terms of area under the curve across all datasets by 11.3%. These results affirm the proposed model’s effectiveness in accurately detecting defects across various software projects.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing software defect prediction with a hybrid convolutional neural network and antlion optimization model

  • 期刊论文

  • 发表日期：

    2025年6月18日

查看全文
  • Alia Nabil Mahmoud,
  • Vitor Santos,
  • Mario M. Freire,
  • Ahmed Abdelaziz

摘要

The growing significance of deep learning in software development has notably enhanced software quality by facilitating the efficient detection of defects, a major hurdle throughout the software development lifecycle. This study aims to identify the optimal model for uncovering defects in software projects. It introduces an intelligent model integrating antlion optimization (ALO) with a convolutional neural network (CNN). CNN is utilized for defect detection, while ALO is employed to fine-tune the network's weights. The paper proposes two intelligent models to tackle this research problem: (a) a basic convolutional neural network without parameter optimization and (b) a hybrid model that combines the CNN with ALO. The results indicate that the hybrid model significantly outperforms the basic CNN across several performance metrics, including area under the curve, sensitivity, specificity, accuracy, and error rate. Furthermore, the findings reveal that the hybrid model exceeds the performance of state-of-the-art methods from prior studies, including CNN, gated recurrent unit, and bidirectional long short-term memory, achieving accuracy improvements of 17.6%, 20.9%, and 24.4%, respectively. Additionally, the proposed model demonstrates superior performance compared to the deep forest method in terms of area under the curve across all datasets by 11.3%. These results affirm the proposed model’s effectiveness in accurately detecting defects across various software projects.