Showing 1–11 of 11 results for author: Zhi, C

A Comprehensive Evaluation of Parameter-Efficient Fine-Tuning on Automated Program Repair

Authors: Guochang Li, Chen Zhi, Jialiang Chen, Junxiao Han, Shuiguang Deng

Abstract: Automated Program Repair (APR) aims to fix bugs by generating patches. And existing work has demonstrated that "pre-training and fine-tuning" paradigm enables Large Language Models (LLMs) improve fixing capabilities on APR. However, existing work mainly focuses on Full-Model Fine-Tuning (FMFT) for APR and limited research has been conducted on the execution-based evaluation of Parameter-Efficient… ▽ More Automated Program Repair (APR) aims to fix bugs by generating patches. And existing work has demonstrated that "pre-training and fine-tuning" paradigm enables Large Language Models (LLMs) improve fixing capabilities on APR. However, existing work mainly focuses on Full-Model Fine-Tuning (FMFT) for APR and limited research has been conducted on the execution-based evaluation of Parameter-Efficient Fine-Tuning (PEFT) for APR. Comparing to FMFT, PEFT can reduce computing resource consumption without compromising performance and has been widely adopted to other software engineering tasks. To fill this gap, we enhance the existing APR dataset by employing prompt engineering to create an instruction dataset, APR-INSTRUCTION, at first. Secondly, we fine-tune four pre-trained LLMs using four different PEFT methods with APR-INSTRUCTION. The best fine-tuned model fixes 58% more bugs than the state-of-the-art LLM-based APR techniques. The results also show that $(IA)^3$ improves the creativity of LLMs more effectively through fine-tuning and achieves the highest fixing capability compared to the other three PEFT methods. Thirdly, we explore the optimal configuration of PEFT hyperparameters, and assess the impact of instruction dataset size, showing that a larger number of parameters and a larger training dataset do not necessarily result in better performance for PEFT. Lastly, we analyze peak memory usage and trainable parameters to show the efficiency of PEFT. This work provides a comprehensive exploration of PEFT on APR and suggests potentially promising directions for extension to other software engineering downstream tasks. APR-INSTRUCTION, PEFT weights, and the fine-tuning code are publicly available as open-source resources. △ Less

Submitted 9 June, 2024; originally announced June 2024.

Confusing Pair Correction Based on Category Prototype for Domain Adaptation under Noisy Environments

Authors: Churan Zhi, Junbao Zhuo, Shuhui Wang

Abstract: In this paper, we address unsupervised domain adaptation under noisy environments, which is more challenging and practical than traditional domain adaptation. In this scenario, the model is prone to overfitting noisy labels, resulting in a more pronounced domain shift and a notable decline in the overall model performance. Previous methods employed prototype methods for domain adaptation on robust… ▽ More In this paper, we address unsupervised domain adaptation under noisy environments, which is more challenging and practical than traditional domain adaptation. In this scenario, the model is prone to overfitting noisy labels, resulting in a more pronounced domain shift and a notable decline in the overall model performance. Previous methods employed prototype methods for domain adaptation on robust feature spaces. However, these approaches struggle to effectively classify classes with similar features under noisy environments. To address this issue, we propose a new method to detect and correct confusing class pair. We first divide classes into easy and hard classes based on the small loss criterion. We then leverage the top-2 predictions for each sample after aligning the source and target domain to find the confusing pair in the hard classes. We apply label correction to the noisy samples within the confusing pair. With the proposed label correction method, we can train our model with more accurate labels. Extensive experiments confirm the effectiveness of our method and demonstrate its favorable performance compared with existing state-of-the-art methods. Our codes are publicly available at https://github.com/Hehxcf/CPC/. △ Less

Submitted 19 March, 2024; originally announced March 2024.

Comments: AAAI 2024

Resisting Backdoor Attacks in Federated Learning via Bidirectional Elections and Individual Perspective

Authors: Zhen Qin, Feiyi Chen, Chen Zhi, Xueqiang Yan, Shuiguang Deng

Abstract: Existing approaches defend against backdoor attacks in federated learning (FL) mainly through a) mitigating the impact of infected models, or b) excluding infected models. The former negatively impacts model accuracy, while the latter usually relies on globally clear boundaries between benign and infected model updates. However, model updates are easy to be mixed and scattered throughout in realit… ▽ More Existing approaches defend against backdoor attacks in federated learning (FL) mainly through a) mitigating the impact of infected models, or b) excluding infected models. The former negatively impacts model accuracy, while the latter usually relies on globally clear boundaries between benign and infected model updates. However, model updates are easy to be mixed and scattered throughout in reality due to the diverse distributions of local data. This work focuses on excluding infected models in FL. Unlike previous perspectives from a global view, we propose Snowball, a novel anti-backdoor FL framework through bidirectional elections from an individual perspective inspired by one principle deduced by us and two principles in FL and deep learning. It is characterized by a) bottom-up election, where each candidate model update votes to several peer ones such that a few model updates are elected as selectees for aggregation; and b) top-down election, where selectees progressively enlarge themselves through picking up from the candidates. We compare Snowball with state-of-the-art defenses to backdoor attacks in FL on five real-world datasets, demonstrating its superior resistance to backdoor attacks and slight impact on the accuracy of the global model. △ Less

Submitted 13 March, 2024; v1 submitted 28 September, 2023; originally announced September 2023.

Comments: Accepted by AAAI 2024. Codes are publicly available at https://github.com/zhenqincn/Snowball

Decoding the Mechanisms of Reversibility Loss in Rechargeable Zinc-Air Batteries

Authors: Zhibin Yi, Liangyu Li, Cheuk Kai Chan, Yaxin Tang, Zhouguang Lu, Chunyi Zhi, Qing Chen, Guangfu Luo

Abstract: Attaining high reversibility of electrodes and electrolyte is essential for the longevity of secondary batteries. Rechargeable zinc-air batteries (RZABs), however, encounter drastic irreversible changes in the zinc anodes and air cathodes during cycling. To uncover the mechanisms of reversibility loss in RZABs, we investigate the evolution of zinc anode, alkaline electrolyte, and air electrode thr… ▽ More Attaining high reversibility of electrodes and electrolyte is essential for the longevity of secondary batteries. Rechargeable zinc-air batteries (RZABs), however, encounter drastic irreversible changes in the zinc anodes and air cathodes during cycling. To uncover the mechanisms of reversibility loss in RZABs, we investigate the evolution of zinc anode, alkaline electrolyte, and air electrode through experiments and first-principles calculations. Morphology diagrams of zinc anodes under versatile operating conditions reveal that the nano-sized mossy zinc dominates the later cycling stage. Such anodic change is induced by the increased zincate concentration due to hydrogen evolution, which is catalyzed by the mossy structure and results in oxide passivation on electrodes, and eventually leads to low true Coulombic efficiencies and short lifespans of batteries. Inspired by these findings, we finally present a novel overcharge-cycling protocol to compensate the Coulombic efficiency loss caused by hydrogen evolution and significantly extend the battery life. △ Less

Submitted 8 August, 2023; originally announced August 2023.

ChatUniTest: A Framework for LLM-Based Test Generation

Authors: Yinghao Chen, Zehao Hu, Chen Zhi, Junxiao Han, Shuiguang Deng, Jianwei Yin

Abstract: Unit testing is an essential yet frequently arduous task. Various automated unit test generation tools have been introduced to mitigate this challenge. Notably, methods based on large language models (LLMs) have garnered considerable attention and exhibited promising results in recent years. Nevertheless, LLM-based tools encounter limitations in generating accurate unit tests. This paper presents… ▽ More Unit testing is an essential yet frequently arduous task. Various automated unit test generation tools have been introduced to mitigate this challenge. Notably, methods based on large language models (LLMs) have garnered considerable attention and exhibited promising results in recent years. Nevertheless, LLM-based tools encounter limitations in generating accurate unit tests. This paper presents ChatUniTest, an LLM-based automated unit test generation framework. ChatUniTest incorporates an adaptive focal context mechanism to encompass valuable context in prompts and adheres to a generation-validation-repair mechanism to rectify errors in generated unit tests. Subsequently, we have developed ChatUniTest Core, a common library that implements core workflow, complemented by the ChatUniTest Toolchain, a suite of seamlessly integrated tools enhancing the capabilities of ChatUniTest. Our effectiveness evaluation reveals that ChatUniTest outperforms TestSpark and EvoSuite in half of the evaluated projects, achieving the highest overall line coverage. Furthermore, insights from our user study affirm that ChatUniTest delivers substantial value to various stakeholders in the software testing domain. ChatUniTest is available at https://github.com/ZJU-ACES-ISE/ChatUniTest, and the demo video is available at https://www.youtube.com/watch?v=GmfxQUqm2ZQ. △ Less

Submitted 7 May, 2024; v1 submitted 8 May, 2023; originally announced May 2023.

Comments: This shorter version is accepted by the FSE 2024 Demonstrations Track, and the previous longer version titled "ChatUniTest: a ChatGPT-based automated unit test generation tool" can be found at arXiv:2305.04764v1

The prolific thermonuclear X-ray bursts from the outburst of the newly discovered millisecond pulsar MAXI~J1816--195 observed by Insight-HXMT and NICER

Authors: Chen Yu-Peng, Zhang Shu, Ji Long, Zhang Shuang-Nan, Wang Peng-Ju, Kong Ling-Da, Chang Zhi, Peng **g-Qiang, Shui Qing-Cang, Li Jian, Tao Lian, Ge Ming-Yu, Qu **-Lu

Abstract: MAXI J1816-195 is a newly discovered accreting millisecond pulsar with prolific thermonuclear bursts, detected during its outburst in 2022 June by Insight-HXMT and NICER. During the outburst, Insight-HXMT detected 73 bursts in its peak and decay phase, serving as a prolific burst system found in the accreting millisecond pulsars. By analyzing one burst which was simultaneously detected by Insight-… ▽ More MAXI J1816-195 is a newly discovered accreting millisecond pulsar with prolific thermonuclear bursts, detected during its outburst in 2022 June by Insight-HXMT and NICER. During the outburst, Insight-HXMT detected 73 bursts in its peak and decay phase, serving as a prolific burst system found in the accreting millisecond pulsars. By analyzing one burst which was simultaneously detected by Insight-HXMT and NICER, we find a mild deviation from the conventional blackbody model. By stacking the Insight-HXMT lightcurves of 66 bursts which have similar profiles and intensities, a hard X-ray shortage is detected with a significance of 15.7 sigma in 30-100 keV. The shortage is about 30% of the persistent flux, which is low compared with other bursters. The shortage fraction is energy-dependent: larger in a higher energy band. These findings make the newly discovered millisecond MAXI J1816-195 a rather peculiar system compared with other millisecond pulsars and atoll bursters. In addition, based on the brightest burst, we derive an upper limit of the distance as 6.3 kpc, and therefore estimate the upper limit of the inner disc radius of the accretion disc to be~ 40 km. Assuming the radius as the magnetospheric radius, the derived magnetic field strength is about 7.1*10^8 G. △ Less

Submitted 25 August, 2022; originally announced August 2022.

Comments: ApJL in press

An Empirical Study of the Landscape of Open Source Projects in Baidu, Alibaba, and Tencent

Authors: Junxiao Han, Shuiguang Deng, David Lo, Chen Zhi, Jianwei Yin, Xin Xia

Abstract: Open source software has drawn more and more attention from researchers, developers and companies nowadays. Meanwhile, many Chinese technology companies are embracing open source and choosing to open source their projects. Nevertheless, most previous studies are concentrated on international companies such as Microsoft or Google, while the practical values of open source projects of Chinese techno… ▽ More Open source software has drawn more and more attention from researchers, developers and companies nowadays. Meanwhile, many Chinese technology companies are embracing open source and choosing to open source their projects. Nevertheless, most previous studies are concentrated on international companies such as Microsoft or Google, while the practical values of open source projects of Chinese technology companies remain unclear. To address this issue, we conduct a mixed-method study to investigate the landscape of projects open sourced by three large Chinese technology companies, namely Baidu, Alibaba, and Tencent (BAT). We study the categories and characteristics of open source projects, the developer's perceptions towards open sourcing effort for these companies, and the internationalization effort of their open source projects. We collected 1,000 open source projects that were open sourced by BAT in GitHub and performed an online survey that received 101 responses from developers of these projects. Some key findings include: 1) BAT prefer to open source frontend development projects, 2) 88\% of the respondents are positive towards open sourcing software projects in their respective companies, 3) 64\% of the respondents reveal that the most common motivations for BAT to open source their projects are the desire to gain fame, expand their influence and gain recruitment advantage, 4) respondents believe that the most common internationalization effort is "providing an English version of readme files", 5) projects with more internationalization effort (i.e., include an English readme file) are more popular. Our findings provide directions for software engineering researchers and provide practical suggestions to software developers and Chinese technology companies. △ Less

Submitted 2 March, 2021; originally announced March 2021.

QPOs and Orbital elements of X-ray binary 4U 0115+63 during the 2017 outburst observed by Insight-HXMT

Authors: Y. Z. Ding, W. Wang, P. Zhang, Q. C. Bu, C. Cai, X. L. Cao, C. Zhi, L. Chen, T. X. Chen, Y. B. Chen, Y. Chen, Y. P. Chen, W. W. Cui, Y. Y. Du, G. H. Gao, H. Gao, M. Y. Ge, Y. D. Gu, J. Guan, C. C. Guo, D. W. Han, Y. Huang, J. Huo, S. M. Jia, W. C. Jiang , et al. (69 additional authors not shown)

Abstract: In this paper, we presented a detailed timing analysis of a prominent outburst of 4U 0115+63 detected by \textit{Insight}-HXMT in 2017 August. The spin period of the neutron star was determined to be $3.61398\pm 0.00002$ s at MJD 57978. We measured the period variability and extract the orbital elements of the binary system. The angle of periastron evolved with a rate of $0.048\pm0.003$ $yr^{-1}$.… ▽ More In this paper, we presented a detailed timing analysis of a prominent outburst of 4U 0115+63 detected by \textit{Insight}-HXMT in 2017 August. The spin period of the neutron star was determined to be $3.61398\pm 0.00002$ s at MJD 57978. We measured the period variability and extract the orbital elements of the binary system. The angle of periastron evolved with a rate of $0.048\pm0.003$ $yr^{-1}$. The light curves are folded to sketch the pulse profiles in different energy ranges. A multi-peak structure in 1-10 keV is clearly illustrated. We introduced wavelet analysis into our data analysis procedures to study QPO signals and perform a detailed wavelet analysis in many different energy ranges. Through the wavelet spectra, we report the discovery of a QPO at the frequency $\sim 10$ mHz. In addition, the X-ray light curves showed multiple QPOs in the period of $\sim 16-32 $ s and $\sim 67- 200 $ s. We found that the $\sim100$ s QPO was significant in most of the observations and energies. There exist positive relations between X-ray luminosity and their Q-factors and S-factors, while the QPO periods have no correlation with X-ray luminosity. In wavelet phase maps, we found that the pulse phase of $\sim 67- 200 $ s QPO drifting frequently while the $\sim 16-32 $ s QPO scarcely drifting. The dissipation of oscillations from high energy to low energy was also observed. These features of QPOs in 4U 0115+63 provide new challenge to our understanding of their physical origins. △ Less

Submitted 18 February, 2021; originally announced February 2021.

Comments: 14 pages, 9 figures, and 6 tables. This work has been submitted to MNRAS after the referee's report

Halogenated MXenes with Electrochemically Active Terminals for High Performance Zinc Ion Batteries

Authors: Mian Li, XinLiang Li, Guifang Qin, Kan Luo, Jun Lu, Youbing Li, Guo** Liang, Zhaodong Huang, Lars Hultman, Per Eklund, Per O. A. Persson, Shiyu Du, Zhifang Chai, Chunyi Zhi, Qing Huang

Abstract: The class of two-dimensional metal carbides and nitrides known as MXenes offer a distinct manner of property tailoring for a wide range of applications. The ability to tune the surface chemistry for expanding the property space of MXenes is thus an important topic, although experimental exploration of new surface terminals remains a challenge. Here, we synthesized Ti3C2 MXene with unitary, binary… ▽ More The class of two-dimensional metal carbides and nitrides known as MXenes offer a distinct manner of property tailoring for a wide range of applications. The ability to tune the surface chemistry for expanding the property space of MXenes is thus an important topic, although experimental exploration of new surface terminals remains a challenge. Here, we synthesized Ti3C2 MXene with unitary, binary and ternary halogen terminals, e.g. -Cl, -Br, -I, -BrI and -ClBrI, to investigate the effect of surface chemistry on the properties of MXenes. The electrochemical activity of Br and I element result in the extraordinary electrochemical performance of the MXenes as cathodes for aqueous zinc ion batteries. The -Br and -I containing MXenes, e.g. Ti3C2Br2 and Ti3C2I2, exhibit distinct discharge platforms with considerable capacities of 97.6 mAh g-1 and 135 mAh g-1. Ti3C2(BrI) and Ti3C2(ClBrI) exhibit dual discharge platforms with capacities of 117.2 mAh g-1 and 106.7 mAh g-1. In contrast, the previously discovered MXenes Ti3C2Cl2 and Ti3C2(OF) exhibit no discharge platforms, and only ~50% of capacities and energy densities of Ti3C2Br2. These results emphasize the effectiveness of the Lewis-acidic-melt etching route for tuning the surface chemistry of MXenes, and also show promise for expanding the MXene family towards various applications. △ Less

Submitted 16 June, 2020; originally announced June 2020.

Noncovalent Functionalization of Boron Nitride Nanotubes in Aqueous Media

Authors: Zhenghong Gao, Chunyi Zhi, Dimtri Golberg, Yoshio Bando, Takeshi Serizawa

Abstract: Boron nitride nanotubes (BNNTs) are of intense scientific interests due to their unique physiochemical properties and prospective applications in various nanotechnologies. A critical problem hampering the application processing of BNNTs is the outer sidewall functionalization, which is primarily acquired to lead BNNTs dispersible in various solvents. Furthermore, the surface of BNNTs should be int… ▽ More Boron nitride nanotubes (BNNTs) are of intense scientific interests due to their unique physiochemical properties and prospective applications in various nanotechnologies. A critical problem hampering the application processing of BNNTs is the outer sidewall functionalization, which is primarily acquired to lead BNNTs dispersible in various solvents. Furthermore, the surface of BNNTs should be intelligently designed and precisely controlled to satisfy the specific demands of different applications. For these purposes, covalent and noncovalent approaches have been factually developed for opening up the key door of applications. Importantly, wrap** the outermost sidewall of BNNTs with either water-soluble polymers or biomolecules through weak noncovalent interactions has been proved to be efficient for giving BNNTs considerable dispersity in aqueous media, and endowing novel chemical functions to the BNNTs with almost no change in their pristine physiochemical properties. This contribution is made to summarize recent progresses, and further addresses the future perspectives on the noncovalent functionalization of BNNTs for promoting their application processing. △ Less

Submitted 23 January, 2014; originally announced January 2014.

Comments: Review, 26 pages, 17 figures

Comparative High Pressure Raman Study of Boron Nitride Nanotubes and Hexagonal Boron Nitride

Authors: Surajit Saha, D. V. S. Muthu, D. Golberg, C. Tang, C. Zhi, Y. Bando, A. K. Sood

Abstract: High pressure Raman experiments on Boron Nitride multi-walled nanotubes show that the intensity of the vibrational mode at ~ 1367 cm-1 vanishes at ~ 12 GPa and it does not recover under decompression. In comparison, the high pressure Raman experiments on hexagonal Boron Nitride show a clear signature of a phase transition from hexagonal to wurtzite at ~ 13 GPa which is reversible on decompressio… ▽ More High pressure Raman experiments on Boron Nitride multi-walled nanotubes show that the intensity of the vibrational mode at ~ 1367 cm-1 vanishes at ~ 12 GPa and it does not recover under decompression. In comparison, the high pressure Raman experiments on hexagonal Boron Nitride show a clear signature of a phase transition from hexagonal to wurtzite at ~ 13 GPa which is reversible on decompression. These results are contrasted with the pressure behavior of carbon nanotubes and graphite. △ Less

Submitted 8 March, 2006; originally announced March 2006.

Comments: 13 pages and 5 figures

Journal ref: Chemical Physics Letters 421 (2006) 86-90