A5100458319- Software Engineering Research
- Topic Modeling
- Advanced Malware Detection Techniques
- Natural Language Processing Techniques
- Software System Performance and Reliability
- Web Data Mining and Analysis
- Software Testing and Debugging Techniques
- Digital and Cyber Forensics
- Software Engineering Techniques and Practices
- Hate Speech and Cyberbullying Detection
- Machine Learning in Materials Science
- Data Quality and Management
- Scientific Computing and Data Management
- Sentiment Analysis and Opinion Mining
- Computer Graphics and Visualization Techniques
- Cloud Data Security Solutions
- 3D Shape Modeling and Analysis
- Data Stream Mining Techniques
- Data Mining Algorithms and Applications
- Gastric Cancer Management and Outcomes
- Advanced Graph Neural Networks
- Open Source Software Innovations
- Technology and Data Analysis
- Technology and Security Systems
- Explainable Artificial Intelligence (XAI)
Singapore Management University
2020-2024
Microsoft Research (United Kingdom)
2023
Wuhan Technology and Business University
2022
Hangzhou Dianzi University
2022
PLA Information Engineering University
2022
University of Rochester
2020
Universität Hamburg
2016
First Affiliated Hospital of Jiangxi Medical College
2015
Nanjing University of Aeronautics and Astronautics
2015
Hangzhou Normal University
2014
In this work, we investigate the problem of creating high-fidelity 3D content from only a single image. This is inherently challenging: it essentially involves estimating underlying geometry while simultaneously hallucinating unseen textures. To address challenge, leverage prior knowledge well-trained 2D diffusion model to act as 3D-aware supervision for creation. Our approach, Make-It-3D, employs two-stage optimization pipeline: first stage optimizes neural radiance field by incorporating...
Previous learning-based vulnerability detection methods relied on either medium-sized pre-trained models or smaller neural networks from scratch. Recent advancements in Large Pre-Trained Language Models (LLMs) have showcased remarkable few-shot learning capabilities various tasks. However, the effectiveness of LLMs detecting software vulnerabilities is largely unexplored. This paper aims to bridge this gap by exploring how perform with prompts, particularly focusing two state-of-the-art...
Spectroscopic ellipsometry was used to characterize the complex refractive index of chemical vapor deposition (CVD) graphene grown on copper foils and transferred glass substrates. Two ellipsometers, with respective wavelength ranges extending into ultraviolet infrared (IR), have been CVD optical functions. The absorption follows same relation fine structure constant previously observed in IR region, displays exciton-dominated peak at ∼4.5 eV. functions show some differences when compared...
Extensive research has been conducted on sentiment analysis for software engineering (SA4SE). Researchers have invested much effort in developing customized tools (e.g., SentiStrength-SE, SentiCR) to classify the polarity Software Engineering (SE) specific contents discussions Stack Overflow and code review comments). Even so, there is still room improvement. Recently, pre-trained Transformer-based models BERT, XLNet) brought considerable breakthroughs field of natural language processing...
The tremendous success of Stack Overflow has accumulated an extensive corpus software engineering knowledge, thus motivating researchers to propose various solutions for analyzing its content. performance such hinges significantly on the selection representation models posts. As volume literature continues burgeon, it highlights need a powerful post model and drives researchers’ interest in developing specialized that can adeptly capture intricacies state-of-the-art (SOTA) are Post2Vec...
Large Language Models (LLMs) are widely utilized in software engineering (SE) tasks, such as code generation and automated program repair. However, their reliance on extensive often undisclosed pre-training datasets raises significant concerns about data leakage, where the evaluation benchmark is unintentionally ``seen'' by LLMs during model's construction phase. The leakage issue could largely undermine validity of LLM-based research evaluations. Despite increasing use SE community, there...
In both commercial and open-source software, bug reports or issues are used to track bugs feature requests. However, the quality of can differ a lot. Prior research has found that with good tend gain more attention than ones poor quality. As an essential component issue, title is important aspect issue Moreover, usually presented in list view, where only some metadata present. this case, concise accurate crucial for readers grasp general concept facilitate triaging. Previous work formulated...
Pull Requests (PRs) are a mechanism on modern collaborative coding platforms, such as GitHub. PRs allow developers to tell others that their code changes available for merging into another branch in repository. A PR needs be reviewed and approved by the core team of repository before merged branch. Usually, reviewers need identify is line with interests providing review. By default, arranged list view shows titles PRs. Therefore, it desirable have precise concise title, which beneficial both...
Software development involves collaborative interactions where stakeholders express opinions across various platforms. Recognizing the sentiments conveyed in these is crucial for effective and ongoing maintenance of software systems. For products, analyzing sentiment user feedback, e.g., reviews, comments, forum posts can provide valuable insights into satisfaction areas improvement. This guide future updates features. However, accurately identifying engineering datasets remains challenging....
Many Duplicate Bug Report Detection (DBRD) techniques have been proposed in the research literature. The industry uses some other techniques. Unfortunately, there is insufficient comparison among them, and it unclear how far we been. This work fills this gap by comparing aforementioned To compare first need a benchmark that can estimate tool would perform if applied realistic setting today. Thus, investigated potential biases affect fair of accuracy DBRD Our experiments suggest data age...
Prior studies have demonstrated that approaches to generate an answer summary for a given technical query in Software Question and Answer (SQA) sites are desired. We find existing assessed solely through user studies. Hence, new study needs be performed every time approach is introduced; this time-consuming, slows down the development of approach, results from different may not comparable each other. There need benchmark with ground truth summaries as complement assessment Unfortunately,...
Duplicate bug report detection (DBRD) is a long-standing challenge in both academia and industry. Over the past decades, researchers have proposed various approaches to detect duplicate reports more accurately. With recent advancement of deep learning, also several that leverage learning models reports. A benchmarking study on DBRD reveals performance learning-based not always better than traditional approaches. However, limitations, e.g., they are usually based bag-of-words model, which...
Software developers often use social media (such as Twitter) to share programming knowledge such new tools, sample code snippets, and tips on programming. One of the topics they talk about is software library. The tweets may contain useful information a A good understanding this information, e.g., developer's views regarding library can be beneficial weigh pros cons using well general sentiments towards However, it not trivial recognize whether word actually refers or other meanings. For...
Recent years have seen the remarkable capabilities of large language models (LLMs) for code generation. Different from existing work that evaluate correctness generated by LLMs, we propose to further its efficiency. More efficient can lead higher performance and execution efficiency programs software completed LLM-assisted programming. First, LLMs on two benchmarks, HumanEval MBPP. Then, choose a set programming problems online judge platform LeetCode conduct more difficult evaluation....
Too many options can be a problem, which is the case for Application Programming Interfaces (APIs). As there are such APIs, with more being introduced periodically, it raises problem of choosing API to recommended. Furthermore, numerous APIs commonly used together other complementary third-party APIs. It challenging developers understand how use each and remember all they want use. Therefore, an accurate recommendation approach improve developers' efficiency in implementing certain...
While having options could be liberating, too many lead to the sub-optimal solution being chosen. This is not an exception in software engineering domain. Nowadays, API has become imperative making developers' life easier. APIs help developers implement a function faster and more efficiently. However, given large number of open-source libraries choose from, choosing right simple task. Previous studies on recommendation leverage natural language (query) identify which would suitable for these...
This paper presents a method to capture the computational intensity and computing resource requirements of data analysis in intelligent transportation systems (ITS). These can be transformed into generic methodological framework for Cy
With the fast development of Internet, size routing tables in backbone routers keeps a rapid growth recent years. An effective solution to control memory occupation ever-increased huge table is Forwarding Information Base (FIB) compression. Existing optimal FIB compression algorithm ORTC suffers from high computational complexity and poor update performance, due loss essential structure information during its process. To address this problem, we present two suboptimal algorithms — EAR-fast...
With the rise of pull request mechanism in software development, quality requests has gained more attention. Prior works focus on improving descriptions and several approaches have been proposed to automatically generate descriptions. As an essential component a request, titles not received similar level To further facilitate automation development help developers draft high-quality titles, we introduce AutoPRTitle. AutoPRTitle is specifically designed automatically. can precise succinct...
We construct a political connection index to capture variations in the strength of firm relations China. The incorporates channels through which firm’s executives and directors are politically connected with government officials bureaucrats. report hump-shaped relation between connections value stock returns. Firm Tobin’s Q cross-sectional returns increase at lower level connections, but decrease higher level. positive effect on is enhanced for firms headquartered regions strong...