我们评估 ML 的不同源代码表示的有效性:我们评估传统使用的代码度量对代码嵌入(code2vec、code2seq 和 CuBERT)的有效性。 这项研究是第一个据我们所知评估预训练的神经源代码嵌入对代码气味检测的有效性的研究。 这种方法帮助我们利用了迁移学习的力量——我们的研究是第一个探索从代码理解模型中挖掘的知识是否可
1、扩展了《Deep Learning Based Feature Envy Detection》文中的特征嫉妒的检测方法,应用于多种代码坏味 2、聚合神经网络的分类算法,从给定的数据集中同时生成多个引导样本,并训练多个二元分类器,这些二元分类器一次投票确定最终分类。结果表明提高了分类性能。 2、相关工作 特征嫉妒的检测 特征嫉妒:一个类对另一个类...
Smell detection toolsDeep learningTransfer-learningContext An excessive number of code smells make a software system hard to evolve and maintain. Machine learning methods, in addition to metric-based and heuristic-based methods, have been recently applied to detect code smells; however, current ...
In this section, we present the background on code smell detection and elaborate on the related literature. Motivation This section presents the motivation to demonstrate the rationale. Building the dataset is important for deep-learning-based detection. During this procedure, it is important to ...
This paper aims to explore the application of deep learning in smart contract vulnerabilities detection. Smart contracts are an essential part of blockchain technology and are crucial for developing decentralized applications. However, smart contract vul
Deep learning based code smell detection (TSE, 2021) A deep method renaming prediction and refinement approach for Java projects (QRS, 2021) Graph neural network to dilute outliers for refactoring monolith application (AAAI, 2021) RefDiff 2.0: A multi-language refactoring detection tool (TSE, 2021...
[216] is the first attempt at deep learning-based code smell detection. They exploited traditional code metrics, e.g., coupling between code entities, by a CNN, and exploited the identifiers of code entities (i.e., names of the to-be-tested method and names of its enclosing class as ...
POS tagging: -In deep learning models, such as RNN, Bi-LSTM, and Transformer-based architecture like BERT, are significantly enhanced when combined with Part-of-Speech tagging for multi-label classification tasks like requirement smell detection. Part-of-Speech tagging provides valuable syntactic info...
We build a global fine-grained dumpsite detection dataset which can be used to train efficient deep learning networks. We design a novel deep convolutional model, BCA-Net, based on the characteristics of dumpsites. BCA-Net can detect more than 98% of dumpsites. ...
Andres solution to ieee-fraud-detection NODE: Neural Oblivious Decision Ensembles for Deep Learning on Tabular Data paper Continuous variables: Feed them directly to the network Categorical variable: Use embeddings Collaborative filtering: When you have users and items. Useful for recommendation systems....