Shengxin Zhu

Prof. Shengxin Zhu

orcid

Research Focus & Metrics

Academic Career Length

18Years (2009-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
8 May 2025 in Patterns
Verified
Journal Article
Open access
BCBanghao Chen
Banghao Chen
ZZZhaofeng Zhang
Zhaofeng Zhang
NLNicolas Langrené
Nicolas Langrené
SZShengxin Zhu
Shengxin Zhu
This review explores the role of prompt engineering in unleashing the capabilities of large language models (LLMs). Prompt engineering is the process of structuring inputs, and it has emerged as a crucial technique for maximizing the utility and accuracy of these models. Both foundational and advanced prompt engineering methodologies—including techniques such as self-consistency, chain of thought, and generated knowledge, which can significantly enhance the performance of models—are explored in this paper. Additionally, the prompt methods for vision language models (VLMs) are examined in detail. Prompt methods are evaluated with subjective and objective metrics, ensuring a robust analysis of their efficacy. Critical to this discussion is the role of prompt engineering in artificial intelligence (AI) security, particularly in terms of defending against adversarial attacks that exploit vulnerabilities in LLMs. Strategies for minimizing these risks and improving the robustness of models are thoroughly reviewed. Finally, we provide a perspective for future research and applications.

10 April 2015 in Numerical Algorithms
Verified
Journal Article
Closed access
PDF file
AWAndrew J. Wathen
Andrew J. Wathen
SZShengxin Zhu
Shengxin Zhu

11 February 2020 in IEEE Access
Verified
Journal Article
Open access
ZCZhiyi Chen
Zhiyi Chen
SZShengxin Zhu
Shengxin Zhu
Qiang NiuQiang Niu
Qiang Niu
TZTianyu Zuo
Tianyu Zuo
We give a concise tutorial on knowledge discovery with linear mixed model in movie recommendation. The versatility of mixed effects model is well explained. Commonly used methods for parameter estimation, confidence interval estimate and evaluation criteria for model selection are briefly reviewed. Mixed effects models produce sound inference based on a series of rigorous analysis. In particular, we analyze millions of movie rating data with LME4 R package and find solid evidences for a general social behavior: the young tend to be more censorious than senior people when evaluating the same object. Such a social behavior phenomenon can be used in recommender systems and business data analysis.

1 January 2014 in Computers & Mathematics with Applications
Verified
Journal Article
Closed access
SZShengxin Zhu
Shengxin Zhu
TGTong-Xiang Gu
Tong-Xiang Gu
XLXing-Ping Liu
Xing-Ping Liu

2 October 2014 in Journal of Scientific Computing
Verified
Journal Article
Closed access
PDF file
SZShengxin Zhu
Shengxin Zhu
AWAndrew J. Wathen
Andrew J. Wathen

1 October 2016
Verified
Conference Paper
Closed access
SZShengxin Zhu
Shengxin Zhu
TGTongxiang Gu
Tongxiang Gu
XXXiaowen Xu
Xiaowen Xu
ZMZeyao Mo
Zeyao Mo
Many statistical models require an estimation of unknown (co)-variance parameter(s). The estimation is usually obtained by maximizing a log-likelihood which involves log determinant terms. In principle, one requires the observed information-the negative Hessian matrix or the second derivative of the log-likelihood-to obtain an accurate maximum likelihood estimator according to the Newton method. When one uses the Fisher information, the expect value of the observed information, a simpler algorithm than the Newton method is obtained as the Fisher scoring algorithm. With the advance in high-throughput technologies in the biological sciences, recommendation systems and social networks, the sizes of data sets-and the corresponding statistical models-have suddenly increased by several orders of magnitude. Neither the observed information nor the Fisher information is easy to obtained for these big data sets. This paper introduces an information splitting technique to simplify the computation. After splitting the mean of the observed information and the Fisher information, an simpler approximate Hessian matrix for the log-likelihood can be obtained. This approximated Hessian matrix can significantly reduce computations, and makes the linear mixed model applicable for big data sets. Such a spitting and simpler formulas heavily depend on matrix algebra transforms, and applicable to large scale breeding model, genetics wide association analysis.