Search

Large Language Models (LLMs) and pre-trained Language Models (LMs) have achieved impressive success on many software engineering tasks (e.g., code completion and code generation). By leveraging huge existing code corpora (e.g., GitHub), these models aim to understand the patterns in source code and use these patterns to predict code properties. However, fine-tuning LLMs is time-consuming and costly for end users and small organizations. Furthermore, fine-tuning LMs heavily depends on the amount and quality of datasets available. As a result, the current lack of data and the high cost of collecting it in real-world scenarios further limit the applicability of LMs. In this paper, we leverage the powerful generation capabilities of LLMs to enhance pre-trained LMs. Specifically, we use LLMs to generate domain-specific data, thereby improving the performance of pre-trained LMs on the target tasks. We conduct experiments by combining different LLMs in our generation phase and introducing various LMs to learn from the LLM-generated data. Then, we compare the performance of these LMs before and after learning the data. We find that LLM-generated data significantly enhances the performance of LMs. The improvement can reach up to 58.36% for fault localization and up to 6.09% for clone detection. Our study highlights that using LLMs to generate data for LMs can improve performance by a large margin.

Though many deep learning-based models have made great progress in vulnerability detection, we have no good understanding of these models, which limits the further advancement of model capability, understanding of the mechanism of model detection, and efficiency and safety of practical application of models. In this paper, we extensively and comprehensively investigate two types of state-of-the-art learning-based approaches (sequence-based and graph-based) by conducting experiments on a recently built large-scale dataset. We investigate seven research questions from five dimensions, namely model capabilities, model interpretation, model stability, ease of use of model, and model economy. We experimentally demonstrate the priority of sequence-based models and the limited abilities of both LLM (ChatGPT) and graph-based models. We explore the types of vulnerability that learning-based models skilled in and reveal the instability of the models though the input is subtlely semantical-equivalently changed. We empirically explain what the models have learned. We summarize the pre-processing as well as requirements for easily using the models. Finally, we initially induce the vital information for economically and safely practical usage of these models., Comment: 13 pages, 5 figures

Though many approaches have been proposed for Automated Program Repair (APR) and indeed achieved remarkable performance, they still have limitations in fixing bugs that require analyzing and reasoning about the logic of the buggy program. Recently, large language models (LLMs) instructed by prompt engineering have attracted much attention for their powerful ability to address many kinds of tasks including bug-fixing. However, the quality of the prompt will highly affect the ability of LLMs and manually constructing high-quality prompts is a costly endeavor. To address this limitation, we propose a self-directed LLM-based automated program repair, ThinkRepair, with two main phases: collection phase and fixing phase. The former phase automatically collects various chains of thoughts that constitute pre-fixed knowledge by instructing LLMs with the Chain-of-Thought (CoT) prompt. The latter phase targets fixing a bug by first selecting examples for few-shot learning and second automatically interacting with LLMs, optionally appending with feedback of testing information. Evaluations on two widely studied datasets (Defects4J and QuixBugs) by comparing ThinkRepair with 12 SOTA APRs indicate the priority of ThinkRepair in fixing bugs. Notably, ThinkRepair fixes 98 bugs and improves baselines by 27%-344.4% on Defects4J V1.2. On Defects4J V2.0, ThinkRepair fixes 12-65 more bugs than the SOTA APRs. Additionally, ThinkRepair also makes a considerable improvement on QuixBugs (31 for Java and 21 for Python at most)., Comment: Accepted By ISSTA'24

Software migration is garnering increasing attention with the evolution of software and society. Early studies mainly relied on handcrafted translation rules to translate between two languages, the translation process is error-prone and time-consuming. In recent years, researchers have begun to explore the use of pre-trained large language models (LLMs) in code translation. However, code translation is a complex task that LLMs would generate mistakes during code translation, they all produce certain types of errors when performing code translation tasks, which include (1) compilation error, (2) runtime error, (3) functional error, and (4) non-terminating execution. We found that the root causes of these errors are very similar (e.g. failure to import packages, errors in loop boundaries, operator errors, and more). In this paper, we propose a general corrector, namely Rectifier, which is a micro and universal model for repairing translation errors. It learns from errors generated by existing LLMs and can be widely applied to correct errors generated by any LLM. The experimental results on translation tasks between C++, Java, and Python show that our model has effective repair ability, and cross experiments also demonstrate the robustness of our method., Comment: arXiv admin note: text overlap with arXiv:2308.03109, arXiv:2302.03908 by other authors

Distribution system state estimation (DSSE) plays a crucial role in the real-time monitoring, control, and operation of distribution networks. Besides intensive computational requirements, conventional DSSE methods need high-quality measurements to obtain accurate states, whereas missing values often occur due to sensor failures or communication delays. To address these challenging issues, a forecast-then-estimate framework of edge learning is proposed for DSSE, leveraging large language models (LLMs) to forecast missing measurements and provide pseudo-measurements. Firstly, natural language-based prompts and measurement sequences are integrated by the proposed LLM to learn patterns from historical data and provide accurate forecasting results. Secondly, a convolutional layer-based neural network model is introduced to improve the robustness of state estimation under missing measurement. Thirdly, to alleviate the overfitting of the deep learning-based DSSE, it is reformulated as a multi-task learning framework containing shared and task-specific layers. The uncertainty weighting algorithm is applied to find the optimal weights to balance different tasks. The numerical simulation on the Simbench case is used to demonstrate the effectiveness of the proposed forecast-then-estimate framework.

Two-dimensional (2D) intrinsic superconductors with nontrivial topological band and vertical ferroelectricity exhibit fascinating characteristics to achieving electrostatic control of quantum phases. While, only a few such 2D materials have been theoretically predicted. In this work, based on first principles calculations, we explore the superconductivity and ferroelectric properties in fluorinated 2D Nb2N. In the stable Nb2NF2, H3-Nb2NF2 breaks the spatial inversion symmetry, exhibiting vertical ferroelectric. More interestingly, it not only possesses intrinsic superconductivity with superconducting transition temperatures (Tc) of 10 K, but also exhibits nontrivial band topology. While, H1-Nb2NF2 shows topological band and superconductivity with Tc of 32 K, surpassing most of 2D conventional topological superconductors' candidates. Our research has enriched 2D superconducting materials with nontrivial band topology and ferroelectric properties, and provided a theoretical basis for the preparation of devices switching between superconducting and ferroelectric states with external electric field.

This paper proposes a pipeline for quantitatively evaluating interactive LLMs such as ChatGPT using publicly available dataset. We carry out an extensive technical evaluation of ChatGPT using Big-Vul covering five different common software vulnerability tasks. We evaluate the multitask and multilingual aspects of ChatGPT based on this dataset. We found that the existing state-of-the-art methods are generally superior to ChatGPT in software vulnerability detection. Although ChatGPT improves accuracy when providing context information, it still has limitations in accurately predicting severity ratings for certain CWE types. In addition, ChatGPT demonstrates some ability in locating vulnerabilities for certain CWE types, but its performance varies among different CWE types. ChatGPT exhibits limited vulnerability repair capabilities in both providing and not providing context information. Finally, ChatGPT shows uneven performance in generating CVE descriptions for various CWE types, with limited accuracy in detailed information. Overall, though ChatGPT performs well in some aspects, it still needs improvement in understanding the subtle differences in code vulnerabilities and the ability to describe vulnerabilities in order to fully realize its potential. Our evaluation framework provides valuable insights for further enhancing ChatGPT' s software vulnerability handling capabilities., Comment: arXiv admin note: substantial text overlap with arXiv:2404.02056

This paper proposes a pipeline for quantitatively evaluating interactive Large Language Models (LLMs) using publicly available datasets. We carry out an extensive technical evaluation of LLMs using Big-Vul covering four different common software vulnerability tasks. This evaluation assesses the multi-tasking capabilities of LLMs based on this dataset. We find that the existing state-of-the-art approaches and pre-trained Language Models (LMs) are generally superior to LLMs in software vulnerability detection. However, in software vulnerability assessment and location, certain LLMs (e.g., CodeLlama and WizardCoder) have demonstrated superior performance compared to pre-trained LMs, and providing more contextual information can enhance the vulnerability assessment capabilities of LLMs. Moreover, LLMs exhibit strong vulnerability description capabilities, but their tendency to produce excessive output significantly weakens their performance compared to pre-trained LMs. Overall, though LLMs perform well in some aspects, they still need improvement in understanding the subtle differences in code vulnerabilities and the ability to describe vulnerabilities to fully realize their potential. Our evaluation pipeline provides valuable insights into the capabilities of LLMs in handling software vulnerabilities.

Colloidal self-assembly allows rational design of structures on the micrometer and submicrometer scale. One architecture that can generate complete 3D photonic band gaps is the diamond cubic lattice, which has remained difficult to realize at length scales comparable to the wavelength of visible or ultraviolet light. Here, we demonstrate three-dimensional photonic crystals self-assembled from DNA origami that act as precisely programmable patchy colloids. Our DNA-based nanoscale tetrapods crystallize into a rod-connected diamond cubic lattice with a periodicity of 170 nm. This structure serves as a scaffold for atomic layer deposition of high refractive index materials such as TiO$_2$, yielding a tunable photonic band gap in the near-ultraviolet.

Though many deep learning (DL)-based vulnerability detection approaches have been proposed and indeed achieved remarkable performance, they still have limitations in the generalization as well as the practical usage. More precisely, existing DL-based approaches (1) perform negatively on prediction tasks among functions that are lexically similar but have contrary semantics; (2) provide no intuitive developer-oriented explanations to the detected results. In this paper, we propose a novel approach named SVulD, a function-level Subtle semantic embedding for Vulnerability Detection along with intuitive explanations, to alleviate the above limitations. Specifically, SVulD firstly trains a model to learn distinguishing semantic representations of functions regardless of their lexical similarity. Then, for the detected vulnerable functions, SVulD provides natural language explanations (e.g., root cause) of results to help developers intuitively understand the vulnerabilities. To evaluate the effectiveness of SVulD, we conduct large-scale experiments on a widely used practical vulnerability dataset and compare it with four state-of-the-art (SOTA) approaches by considering five performance measures. The experimental results indicate that SVulD outperforms all SOTAs with a substantial improvement (i.e., 23.5%-68.0% in terms of F1-score, 15.9%-134.8% in terms of PR-AUC and 7.4%-64.4% in terms of Accuracy). Besides, we conduct a user-case study to evaluate the usefulness of SVulD for developers on understanding the vulnerable code and the participants' feedback demonstrates that SVulD is helpful for development practice., Comment: Accepted By FSE'23

Quantum key distribution (QKD) is a well-known application of quantum information theory that guarantees information-theoretically secure key exchange. As QKD becomes more and more commercially viable, challenges such as scalability, network integration, and high production costs need to be addressed. Photonic and electronic integrated circuits that can be produced in large volumes at low cost hold the key to large-scale deployment of next-generation QKD systems. Here, we present a continuous-variable (CV) QKD system using an integrated photonic-electronic receiver that combines a silicon photonic integrated circuit implementing a phase-diverse receiver with custom-designed GaAs pHEMT transimpedance amplifiers. The QKD system operates at a classical telecom symbol rate of 10 GBaud, generating high secret key rates exceeding 0.7 Gb/s over a distance of 5 km and 0.3 Gb/s over a distance of 10 km. The secret keys are secure against collective attacks with finite-size effects taken into account. Well-designed digital signal processing enabled the high-speed operation. Our experiment sets a new record for secure quantum communication and paves the way for the next generation of CV-QKD systems.

The digitization of displaced archives is of great historical and cultural significance. Through the construction of digital humanistic platforms represented by MISS platform, and the comprehensive application of IIIF technology, knowledge graph technology, ontology technology, and other popular information technologies. We can find that the digital framework of displaced archives built through the MISS platform can promote the establishment of a standardized cooperation and dialogue mechanism between the archives authoritiess and other government departments. At the same time, it can embed the works o fichives ction of digital government and the economy, promote the exploration of the integration of archives management, data management, and information resource management, and ultimately promote the construction of a digital society. By fostering a new partnership between archives departments and enterprises, think tanks, research institutes, and industry associations, the role of multiple social subjects in the modernization process of the archives governance system and governance capacity will be brought into play. The National Archives Administration has launched a special operation to recover scattered archives overseas, drawing up a list and a recovery action plan for archives lost to overseas institutions and individuals due to war and other reasons. Through the National Archives Administration, the State Administration of Cultural Heritage, the Ministry of Foreign Affairs, the Supreme People's Court, the Supreme People's Procuratorate, and the Ministry of Justice, specific recovery work is carried out by studying and working on international laws.

Emerging communication and cryptography applications call for reliable, fast, unpredictable random number generators. Quantum random number generation (QRNG) allows for the creation of truly unpredictable numbers thanks to the inherent randomness available in quantum mechanics. A popular approach is using the quantum vacuum state to generate random numbers. While convenient, this approach was generally limited in speed compared to other schemes. Here, through custom co-design of opto-electronic integrated circuits and side-information reduction by digital filtering, we experimentally demonstrated an ultrafast generation rate of 100 Gbps, setting a new record for vacuum-based quantum random number generation by one order of magnitude. Furthermore, our experimental demonstrations are well supported by an upgraded device-dependent framework that is secure against both classical and quantum side-information and that also properly considers the non-linearity in the digitization process. This ultrafast secure random number generator in the chip-scale platform holds promise for next generation communication and cryptography applications.