检索结果-内蒙古大学图书馆

Comparing the Pretrained Models of source code by Re-pretraining Under a Unified Setup

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 2024年第12期35卷 17768-17778页

作者： Niu, Changan Li, Chuanyi Ng, Vincent Luo, Bin Nanjing Univ Software Inst State Key Lab Novel Software Technol Nanjing 210093 Peoples R China Univ Texas Dallas Human Language Technol Res Inst Richardson TX 75083 USA

Recent years have seen the successful application of large pretrained models of source code (codePTMs) to code representation learning, which have taken the field of software engineering (SE) from task-specific solutions to task-agnostic generic models. By the remarkable results, codePTMs are seen as a promising direction in both academia and industry. While a number of codePTMs have been proposed, they are often not directly comparable because they differ in experimental setups such as pretraining dataset, model size, evaluation tasks, and datasets. In this article, we first review the experimental setup used in previous work and propose a standardized setup to facilitate fair comparisons among codePTMs to explore the impacts of their pretraining tasks. Then, under the standardized setup, we re-pretrain codePTMs using the same model architecture, input modalities, and pretraining tasks, as they declared and fine-tune each model on each evaluation SE task for evaluating. Finally, we present the experimental results and make a comprehensive discussion on the relative strength and weakness of different pretraining tasks with respect to each SE task. We hope our view can inspire and advance the future study of more powerful codePTMs.

关键词： Pretraining task source code supervised learning

来源：评论

学校读者我要写书评

暂无评论

Reducing the Impact of Time Evolution on source code Authorship Attribution via Domain Adaptation

引用

ACM TRANSACTIONS ON SOFTWARE ENGINEERING AND METHODOLOGY 2024年第6期33卷 1-27页

作者： Li, Zhen Zhao, Shasha Chen, Chen Chen, Qian Hebei Univ Sch Cyber Secur & Comp Baoding 071002 Peoples R China Univ Cent Florida Ctr Res Comp Vis Orlando FL 32816 USA Univ Texas San Antonio San Antonio TX 78249 USA Huazhong Univ Sci & Technol Sch Cyber Sci & Engn Wuhan 430074 Peoples R China

source code authorship attribution is an important problem in practical applications such as plagiarism detection, software forensics, and copyright disputes. Recent studies show that existing methods for source code authorship attribution can be significantly affected by time evolution, leading to a decrease in attribution accuracy year by year. To alleviate the problem of Deep Learning (DL)-based source code authorship attribution degrading in accuracy due to time evolution, we propose a new framework called Time Domain Adaptation (TimeDA) by adding new feature extractors to the original DL-based code attribution framework that enhances the learning ability of the original model on source domain features without requiring new or more source data. Moreover, we employ a centroid-based pseudo-labeling strategy using neighborhood clustering entropy for adaptive learning to improve the robustness of DL-based code authorship attribution. Experimental results show that TimeDA can significantly enhance the robustness of DL-based source code authorship attribution to time evolution, with an average improvement of 8.7% on the Java dataset and 5.2% on the C++ dataset. In addition, our TimeDA benefits from employing the centroid-based pseudo-labeling strategy, which significantly reduced the model training time by 87.3% compared to traditional unsupervised domain adaptive methods.

关键词： Authorship attribution source code time evolution deep learning domain adaptation

来源：评论

学校读者我要写书评

暂无评论

Meta-Heuristic Guided Feature Optimization for Enhanced Authorship Attribution in Java source code

引用

IEEE ACCESS 2023年 11卷 141657-141673页

作者： Al-Ahmad, Bilal Al-Madi, Nailah Alzaqebah, Abdullah Alkhawaldeh, Rami S. Aldebei, Khaled Kabir, Md. Faisal Altaharwa, Ismail Abu-Faraj, Mua'ad Aljarah, Ibrahim Univ Jordan Aqaba 77110 Jordan St Cloud State Univ Dept Comp Sci & Informat Technol St Cloud MN 56301 USA Princess Sumaya Univ Technol Amman 11941 Jordan Al Ahliyya Amman Univ Fac Informat Technol Comp Sci Dept Amman 19628 Jordan Penn State Univ Harrisburg Sch Sci Engn & Technol Middletown PA 17057 USA Univ Jordan Amman 11942 Jordan

source code authorship attribution is the task of identifying who develops the code based on learning based on the programmer style. It is one of the critical activities which used extensively in different aspects such as computer security, computer law, and plagiarism. This paper attempts to investigate source code authorship attribution by capturing natural language aspects of the code rather than only using minimal set of syntactic and stylistic code features as explored in the previous literature. It proposes an evolutionary feature selection model to improve the accuracy of authorship attribution by implementing two language models (uni-gram and bi-gram). The proposed approach uses K-Nearest Neighbor as a classifier and Genetic Algorithm as a feature selection technique. Two experiments have been demonstrated on a public Authorship Attribution dataset on GitHub, the experiments include various evolutionary feature selection models. Notably, the obtained results in both experiments were compared with the related studies, and show a significant improvement in terms of accuracy.

关键词： source coding Feature extraction Genetic algorithms codes Task analysis Java Classification algorithms Evolutionary computation data mining feature selection java source code authorship attribution

来源：评论

学校读者我要写书评

暂无评论

Sharing practices of software artefacts and source code for reproducible research

引用

INTERNATIONAL JOURNAL OF DATA SCIENCE AND ANALYTICS 2024年 1-12页

作者： Jean-Quartier, Claire Jeanquartier, Fleur Stryeck, Sarah Simon, Joerg Soeser, Birgit Hasani-Mavriqi, Ilire Graz Univ Technol Res Data Management Graz Austria Univ Nat Resources & Life Sci Dept Forest & Soil Sci Human Ctr AI Lab Vienna Austria Res Ctr Pharmaceut Engn GmbH Graz Austria Graz Univ Technol Inst Interact Syst & Data Sci Graz Austria

While source code of software and algorithms depicts an essential component in all fields of modern research involving data analysis and processing steps, it is uncommonly shared upon publication of results throughout disciplines. Simple guidelines to generate reproducible source code have been published. Still, code optimization supporting its repurposing to different settings is often neglected and even less thought of to be registered in catalogues for a public reuse. Though all research output should be reasonably curated in terms of reproducibility, it has been shown that researchers are frequently non-compliant with availability statements in their publications. These do not even include the use of persistent unique identifiers that would allow referencing archives of code artefacts at certain versions and time for long-lasting links to research articles. In this work, we provide an analysis on current practices of authors in open scientific journals in regard to code availability indications, FAIR principles applied to code and algorithms. We present common repositories of choice among authors. Results further show disciplinary differences of code availability in scholarly publications over the past years. We advocate proper description, archiving and referencing of source code and methods as part of the scientific knowledge, also appealing to editorial boards and reviewers for supervision.

关键词： source code Reproducibility FAIR principles Open science Software availability

来源：评论

学校读者我要写书评

暂无评论

A comparative study of adversarial training methods for neural models of source code

引用

FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE 2023年第1期142卷 165-181页

作者： Li, Zhen Huang, Xiang Li, Yangrui Chen, Guenevere Hebei Univ Sch Cyber Secur & Comp Baoding 071002 Hebei Peoples R China Huazhong Univ Sci & Technol Sch Cyber Sci & Engn Wuhan 430074 Hubei Peoples R China Univ Texas San Antonio Dept Elect & Comp Engn San Antonio TX 78249 USA

Adversarial training has been employed by researchers to protect AI models of source code. However, it is still unknown how adversarial training methods in this field compare to each other in effectiveness and robustness. This study surveys and investigates existing adversarial training methods, and conducts experiments to evaluate these neural models' performance in the domain of source code. First, we examine the process of adversarial training to identify four dimensions that could be used to classify different adversarial training methods into five categories, which are Mixing Directly, Composite Loss, Adversarial Fine-tuning, Min-max + Composite Loss, and Min-max. Second, we conduct empirical evaluations of these classified adversarial training methods under two tasks (i.e., code summarization and code authorship attribution) to determine their performance of effectiveness and robustness. Experimental results indicate that the performance of certain combinations of adversarial training techniques (i.e., min-max with composite loss, or directly-sample with ordinary loss) would be much better than other combinations or other techniques used alone. Our experiments also reveal that the model's robustness of defensive methods can be enhanced by using diverse input data for adversarial training, and that the number of fine-tuning epochs has little or no impact on model's performance.(c) 2022 Elsevier B.V. All rights reserved.

关键词： Adversarial training Robustness source code Comparative study

来源：评论

学校读者我要写书评

暂无评论

BiAn: Smart Contract source code Obfuscation

引用

IEEE TRANSACTIONS ON SOFTWARE ENGINEERING 2023年第9期49卷 4456-4476页

作者： Zhang, Pengcheng Yu, Qifan Xiao, Yan Dong, Hai Luo, Xiapu Wang, Xiao Zhang, Meng Hohai Univ Coll Comp & Informat Nanjing 210000 Peoples R China Sun Yat Sen Univ Sch Cyber Sci & Technol Shenzhen 518000 Peoples R China RMIT Univ Sch Comp Technol Melbourne Vic 3000 Australia Hong Kong Polytech Univ Dept Comp Hong Kong 999077 Peoples R China

With the rising prominence of smart contracts, security attacks targeting them have increased, posing severe threats to their security and intellectual property rights. Existing simplistic datasets hinder effective vulnerability detection, raising security concerns. To address these challenges, we propose BiAn, a source code level smart contract obfuscation method that generates complex vulnerability test datasets. BiAn protects contracts by obfuscating data flows, control flows, and code layouts, increasing complexity and making it harder for attackers to discover vulnerabilities. Our experiments with buggy contracts showed an average complexity enhancement of approximately 174% after obfuscation. Decompilers Vandal and Gigahorse had total failure rate increments of 38.8% and 40.5% respectively. Obfuscated contracts also decreased vulnerability detection rates in more than 50% of cases for ten widely-used static analysis detection tools.

关键词： Smart contracts codes source coding Security Complexity theory Intellectual property Layout Blockchain Ethereum smart contract source code obfuscation

来源：评论

学校读者我要写书评

暂无评论

Detecting vulnerability in source code using CNN and LSTM network

引用

SOFT COMPUTING 2023年第2期27卷 1131-1141页

作者： Guo, Junjun Wang, Zhengyuan Li, Haonan Xue, Yang Xian Technol Univ Sch Comp Sci & Engn Xian Shanxi Peoples R China

Automated vulnerability detection has become a research hot spot because it is beneficial for improving software quality and security. The code metric (CM) is one class of important representations of vulnerability in source code. The implicit relationships among different metric attributes have not been sufficiently considered in traditional vulnerability detection based on CMs. In this paper, in view of the local perception capability of convolutional neural network (CNN) and the time-series prediction capability of long short-term memory (LSTM), we propose VulExplore, a compound neural network model for vulnerability detection that consists of a CNN for feature extraction and an LSTM network for deep representation. Moreover, to further indicate the vulnerability features in the source code, we reconstruct a CM dataset that includes two additional important attributes: maintainability index and average number of vulnerabilities committed per line. Our proposed numerical method can obtain both false-negative rate (FNR) and false-positive rate (FPR) under 20% and, meanwhile, achieve recall and precision over 80%, respectively.

关键词： Vulnerability detection code metrics (CMs) Convolutional neural network (CNN) Long short-term memory (LSTM) source code

来源：评论

学校读者我要写书评

暂无评论

source code Protection Against Unauthorised Copying and Analysis in IoT Devices 56

Source Code Protection Against Unauthorised Copying and Anal...

引用

56th Annual Hawaii International Conference on System Sciences (HICSS)

作者： Hyla, Tomasz Byczyk, Sebastian West Pomeranian Univ Technol Szczecin Poland

ISBN: (纸本)9780998133164

One problem for manufacturers of IoT devices is protecting intellectual rights to the software. Unprotected software can easily be copied or analysed and used on other devices. Proving another party that the source code has been illegally copied is difficult. One of the solutions is code obfuscation, i.e., modifying the code, so it works the same way, but its structure is complicated to understand and analyse. The paper presents a solution dedicated to IoT devices that combines code obfuscation techniques and uses the trusted platform module to decrypt part of the data during execution. A dedicated obfuscation method is described. Experiments show that the method increases the time needed to change the code at least several times, and some junior programmers cannot understand an obfuscated code. Test results show almost no similarity between the code in clear form and obfuscated form. The obfuscated code is more complicated but takes slightly longer to execute.

关键词： IoT device source code obfuscation reverse engineering TPM

来源：评论

学校读者我要写书评

暂无评论

The lifecycle of Technical Debt that manifests in both source code and issue trackers

引用

INFORMATION AND SOFTWARE TECHNOLOGY 2023年第1期159卷

作者： Tan, Jie Feitosa, Daniel Avgeriou, Paris Acad Mil Sci Beijing Peoples R China Univ Groningen Bernoulli Inst Math Comp Sci & Artificial Intellig Groningen Netherlands

Context: Although Technical Debt (TD) has increasingly gained attention in recent years, most studies exploring TD are based on a single source (e.g., source code, code comments or issue trackers). Objective: Investigating information combined from different sources may yield insight that is more than the sum of its parts. In particular, we argue that exploring how TD items are managed in both issue trackers and software repositories (including source code and commit messages) can shed some light on what happens between the commits that incur TD and those that pay it back. Method: To this end, we randomly selected 3,000 issues from the trackers of five projects, manually analyzed 300 issues that contained TD information, and identified and investigated the lifecycle of 312 TD ***: The results indicate that most of the TD items marked as resolved in issue trackers are also paid back in source code, although many are not discussed after being identified in the issue tracker. Test Debt items are the least likely to be paid back in source code. We also learned that although TD items may be resolved a few days after being identified, it often takes a long time to be identified (around one year). In general, time is reduced if the same developer is involved in consecutive moments (i.e., introduction, identification, repayment decision-making and remediation), but whether the developer who paid back the item is involved in discussing the TD item does not seem to affect how quickly it is ***: Investigating how developers manage TD across both source code repositories and issue trackers can lead to a more comprehensive oversight of this activity and support efforts to shorten the lifecycle of undesirable debt.

关键词： Technical Debt source code Issue tracker

来源：评论

学校读者我要写书评

暂无评论

CognitIDE: An IDE Plugin for Mapping Physiological Measurements to source code 32

CognitIDE: An IDE Plugin for Mapping Physiological Measureme...

引用

32nd ACM International Conference on the Foundations of Software Engineering (FSE)

作者： Stolp, Fabian Stellmacher, Malte Arnrich, Bert Hasso Plattner Inst Potsdam Germany Univ Potsdam Potsdam Germany

ISBN: (纸本)9798400706585

We present CognitIDE, a tool for collecting physiological measurements, mapping them to source code, and visualizing them directly within IntelliJ-based Integrated Development Environments (IDE)s. CognitIDE facilitates the setup and conduction of empirical studies evaluating the relationships between software artifacts and physiological parameters. Corresponding measurements enable researchers to evaluate, for example, the cognitive load software developers are experiencing. Our tool lets study participants use IDEs in a natural way while eye gaze, further body sensor data, and interactions with the IDE are collected. Furthermore, CognitIDE enables highlighting code positions according to the physiological values collected while corresponding positions were looked at. This facilitates the identification of poorly maintainable code and provides a direct way for study participants to reflect on whether the measurements mirror their perception. Moreover, the plugin has additional features for facilitating studies, such as interrupting participants and letting them answer predefined questions. Our tool supports recording measurements with the wide variety of devices supported by the Lab Streaming Layer. Video: https://***/9yLV5AdTiJw

关键词： physiological measurements source code integrated development environment multimodality cognition

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：