This dissertation aims to introduce interpretability techniques to comprehensively evaluate the performance of Large Language Models (LLMs) in softwareengineering tasks, beyond canonical metrics. In software engineer...
详细信息
This study examines the potential of the LEGO® Serious Play® (LSP) method, augmented with generative artificial intelligence (AI), to facilitate introspection and its application across various settings. Emp...
详细信息
This paper introduces a new network model - the Image Guidance Encoder-Decoder Model (IG-ED), designed to enhance the efficiency of image captioning and improve predictive accuracy. IG-ED, a fusion of the convolutiona...
详细信息
Methodologies serve as the foundation for managing and organizing the software development process, encompassing diverse approaches like feature-driven development, waterfall, and extreme programming. In addressing ch...
详细信息
This study addresses the prevalent issues with new energy vehicle batteries, including failure and other complications. It focuses on lithium-ion batteries in pure electric vehicles and proposes a diagnostic approach ...
详细信息
In recent years, the emerging open source RISC-V architecture has gradually attracted wide attention. In order to support the compilation of multiple Linux operating system distribution images and packages, developers...
详细信息
ISBN:
(纸本)9783031664557;9783031664564
In recent years, the emerging open source RISC-V architecture has gradually attracted wide attention. In order to support the compilation of multiple Linux operating system distribution images and packages, developers need to build and adapt the packages. Due to the complexity of software packages and the diversity of developer experience levels, the success of software package construction is uncertain. Existing research lacks automatic classification methods for the reasons of RISC-V architecture software package construction failures. Therefore, an automatic classification model Word2Vec-BERT-bmu is proposed to effectively and automatically locate software package construction failures. Firstly, two popular Linux distribution building platforms, OpenSuse and OpenEuler, were selected as the sources of build failure log data, and 10 types of build failures were manually analyzed and summarized. Secondly, the Word2Vec-BERT-bmu model is proposed to construct the failure classification using an automated software package with multi-feature concatenation. Experimental results show that the Macro F1 value is improved by 2-4% compared with other models. In addition, for realworld software packages, the effectiveness and accuracy of our modelwe proposed are further verified by manual repair of software packages.
Incorporating ethics into engineering education is crucial to address the intricate challenges of societal technologization, artificial intelligence, and automation. Across-disciplinary and ongoing ethical education, ...
详细信息
ISBN:
(纸本)9783031613043;9783031613050
Incorporating ethics into engineering education is crucial to address the intricate challenges of societal technologization, artificial intelligence, and automation. Across-disciplinary and ongoing ethical education, engaging students with practical scenarios relevant to their field, has been identified as a superior approach for ethical training of engineers. Realizing this advanced ethical training necessitates dedicated support for both instructors and students. This paper introduces a major enhanced version 2.0 of EthicApp, a case-based collaborative learning platform that facilitates ethical education. The platform enables individual and collective examination of ethical cases across various disciplines. EthicApp 2.0 promotes analysis and collaborative decision-making in ethical contexts. A formative study of EthicApp 2.0 was conducted with softwareengineering students at a South American university (N= 109). The study utilized tasks involving semantic differential scales and the ranking of case stakeholders. Students performed these tasks on personal computers and smartphones, with random assignment. The study confirmed that EthicApp 2.0 could be used for various task types without prior training. Instructors can repurpose learning designs and monitor activities in real-time effectively. However, it was observed that the mobile interface posed usability challenges, and responses submitted via smartphones tended to be shorter. Despite these limitations, EthicApp 2.0 shows promise for scalability to larger samples and integration into diverse courses, aiding in developing ethical competencies as a cross-disciplinary skill.
Skin melanoma, a dangerous kind of skin cancer, necessitates accurate identification and diagnosis to improve patient outcomes. The use of deep learning algorithms into medical imaging promises significant advancement...
详细信息
software vulnerabilities damage the functionality of software systems. Recently, many deep learning-based approaches have been proposed to detect vulnerabilities at the function level by using one or a few different m...
详细信息
ISBN:
(纸本)9798350329964
software vulnerabilities damage the functionality of software systems. Recently, many deep learning-based approaches have been proposed to detect vulnerabilities at the function level by using one or a few different modalities (e.g., text representation, graph-based representation) of the function and have achieved promising performance. However, some of these existing studies have not completely leveraged these diverse modalities, particularly the underutilized image modality, and the others using images to represent functions for vulnerability detection have not made adequate use of the significant graph structure underlying the images. In this paper, we propose MVulD, a multi-modal-based function-level vulnerability detection approach, which utilizes multi-modal features of the function (i.e., text representation, graph representation, and image representation) to detect vulnerabilities. Specifically, MVulD utilizes a pre-trained model (i.e., UniXcoder) to learn the semantic information of the textual source code, employs the graph neural network to distill graph-based representation, and makes use of computer vision techniques to obtain the image representation while retaining the graph structure of the function. We conducted a large-scale experiment on 25,816 functions. The experimental results show that MVulD improves four state-of-the-art baselines by 30.8%-81.3%, 12.8%-27.4%, 48.8%-115%, and 22.9%-141% in terms of F1-score, Accuracy, Precision, and PR-AUC respectively.
At the moment, the number of developed programmes is growing rapidly all over the world. There is a huge number of development methodologies, but the problem of architecture of complex systems remains relevant. This p...
详细信息
暂无评论