A5107274823- Topic Modeling
- Software Engineering Research
- Advanced Malware Detection Techniques
- Software Testing and Debugging Techniques
- Multimodal Machine Learning Applications
- Software Reliability and Analysis Research
- Natural Language Processing Techniques
- Stability and Control of Uncertain Systems
- Oil and Gas Production Techniques
- Educational Innovations and Challenges
- Structural Health Monitoring Techniques
- Software System Performance and Reliability
- Manufacturing Process and Optimization
- Online Learning and Analytics
- Advanced Technologies in Various Fields
- Adaptive Control of Nonlinear Systems
- Web Application Security Vulnerabilities
- Speech and dialogue systems
- Reliability and Maintenance Optimization
- Coastal wetland ecosystem dynamics
- Embedded Systems Design Techniques
- Expert finding and Q&A systems
- Fire effects on ecosystems
- Wireless Signal Modulation Classification
- Probabilistic and Robust Engineering Design
University of California, Riverside
2023-2024
Delft University of Technology
2024
The University of Melbourne
2023
Nanjing Institute of Technology
2023
Xi'an Technological University
2021-2022
Xi’an University of Posts and Telecommunications
2019
Nanjing University of Aeronautics and Astronautics
2014
Nanyang Technological University
2000
While static analysis is instrumental in uncovering software bugs, its precision analyzing large and intricate codebases remains challenging. The emerging prowess of Large Language Models (LLMs) offers a promising avenue to address these complexities. In this paper, we present LLift, pioneering framework that synergizes LLMs, with spotlight on identifying use-before-initialization (UBI) bugs within the Linux kernel. Drawing from our insights into variable usage conventions Linux, enhance...
Recent advances of Large Language Models (LLMs), e.g., ChatGPT, exhibited strong capabilities comprehending and responding to questions across a variety domains. Surprisingly, ChatGPT even possesses understanding program code. In this paper, we investigate where how LLMs can assist static analysis by asking appropriate questions. particular, target specific bug-finding tool, which produces many false positives from the analysis. our evaluation, find that these be effectively pruned carefully...
Oil and gas reserves are important resources for human survival, directly related to future oil production the sustainable development utilization of energy. It is crucial strengthen understanding judgment growth trend reserves. The prediction a forward-looking research work, its results will affect direction exploration investment. To explore new methods predicting reserves, promote energy, provide theoretical basis development. For this purpose, taking Llanos Basin in South America as an...
Static analysis is a widely used technique in software engineering for identifying and mitigating bugs. However, significant hurdle lies achieving delicate balance between precision scalability. Large Language Models (LLMs) offer promising alternative, as recent advances demonstrate remarkable capabilities comprehending, generating, even debugging code. Yet, the logic of bugs can be complex require sophisticated reasoning large scope spanning multiple functions. Therefore, at this point,...
Instruction tuning has shown great promise in improving the performance of large language models. However, research on multilingual instruction been limited due to scarcity high-quality instruction-response datasets across different languages. To bridge this gap, we present Bactrian-X, a comprehensive parallel dataset 3.4 million pairs 52 Leveraging dataset, train set adapters using low-rank adaptation (LoRA), which are lightweight components that seamlessly integrate with These have...
The recent surge in open-source Large Language Models (LLMs), such as LLaMA, Falcon, and Mistral, provides diverse options for AI practitioners researchers. However, most LLMs have only released partial artifacts, the final model weights or inference code, technical reports increasingly limit their scope to high-level design choices surface statistics. These hinder progress field by degrading transparency into training of forcing teams rediscover many details process. We present LLM360, an...
With the rapid growth of open-source software, code cloning has become increasingly prevalent. If there are security vulnerabilities in a cloned segment, those may spread related software to potentially lead incidents. The existing methods vulnerable detection performed on condition that source is converted into an intermediate representation. However, these do not fully consider rich semantic knowledge and patch information available for codes, which can induce high false positive rate...
With the rapid evolution of large language models (LLMs), new and hard-to-predict harmful capabilities are emerging. This requires developers to be able identify risks through evaluation "dangerous capabilities" in order responsibly deploy LLMs. In this work, we collect first open-source dataset evaluate safeguards LLMs, safer LLMs at a low cost. Our is curated filtered consist only instructions that responsible should not follow. We annotate assess responses six popular these instructions....
Answering real-world tourism questions that seek Point-of-Interest (POI) recommendations is challenging, as it requires both spatial and non-spatial reasoning, over a large candidate pool. The traditional method of encoding each pair question POI becomes inefficient when the number candidates increases, making infeasible for applications. To overcome this, we propose treating QA task dense vector retrieval problem, where encode POIs separately retrieve most relevant by utilizing embedding...
Large language models (LLMs) are notorious for hallucinating, i.e., producing erroneous claims in their output. Such hallucinations can be dangerous, as occasional factual inaccuracies the generated text might obscured by rest of output being generally factual, making it extremely hard users to spot them. Current services that leverage LLMs usually do not provide any means detecting unreliable generations. Here, we aim bridge this gap. In particular, propose a novel fact-checking and...
Assigning aircraft to gates is one of the most important daily decision problems that airport professionals face. The solution this problem has raised a significant effort, with many researchers tackling different variants problem. However, existing studies on gate assignment contain only static perspective without considering possible future disruptions and uncertainties. We bridge gap by looking at assignments as dynamic decision-making process. This paper presents Real-time Gate...
In order to enhance the inheritance and innovation of Marxist theory in digital age, stimulate potential higher education promoting development new quality productivity, based on guidance theory, use consolidate foundation productivity momentum high-quality development, then promote Chinese-style modernization with development. This paper first discusses theoretical connotation significance its inherent logical relationship explores role practical problems aiming reveal influence clarify...
We introduce Loki, an open-source tool designed to address the growing problem of misinformation. Loki adopts a human-centered approach, striking balance between quality fact-checking and cost human involvement. It decomposes task into five-step pipeline: breaking down long texts individual claims, assessing their check-worthiness, generating queries, retrieving evidence, verifying claims. Instead fully automating claim verification process, provides essential information at each step assist...
Subset Simulation provides an efficient and robust method to estimate the failure probabilities at small probability levels for a single mode defined by system response concerned. However, estimating of multiple stochastic responses using run remains quite challenging task in structural reliability analysis. To address this problem, study develops generalized (GSS) approach, which unified intermediate event is drive simulation procedure progressively approaching regions, corresponding...
The traditional mobile phone signal blocker adopts high power broadband noise suppression technology, which has poor jamming effect and adverse effects on the health of people in shielded area. This paper proposes a synchronous coherent technology that jams with 5G signals. According to characteristics base station downlink signals, can conduct modulation process search destroy cell terminals. It is suitable for 5G, 4G, other simulation results show signal, advantages narrow band width utilization.
Answering real-world tourism questions that seek Point-of-Interest (POI) recommendations is challenging, as it requires both spatial and non-spatial reasoning, over a large candidate pool.The traditional method of encoding each pair question POI becomes inefficient when the number candidates increases, making infeasible for applications.To overcome this, we propose treating QA task dense vector retrieval problem, where encode POIs separately retrieve most relevant by utilizing embedding...
In this study, the training data of graduate students Xi'an University Posts and Telecommunications for past three years was used to construct a predictive model students' high-quality employment based on support vector machine through 13 feature attributes that may affect students.Six important characteristics high quality were studied, including gender, student origin, postgraduate score, innovation fund, book reading graduation thesis score. After many experiments, accuracy...
Abstract Vulnerabilities can have very serious consequences for information security, with huge implications economic, social, and even national security. Automated vulnerability detection has always been a keen topic researchers. From traditional manual mining to static dynamic detection, all rely on human experts define features. The rapid development of machine learning deep alleviated the tedious task manually defining features by while reducing lack objectivity caused subjective...