Hefeng Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Hefeng Chen has a total of 44 co-authors. Here are the top co-authors by number of shared publications:
- Chin-Chen Chang
Department of Computer Science and Information Engineering, National Chin-Yi University of Technology, Taichung, Taiwan
4 shared publications - Yuanguo Lin
Jimei University
4 shared publications - Kang Chen
School of Computer Engineering, Jimei University, Xiamen, China
3 shared publications - Kangkang Li
School of Smart Education, Jiangsu Normal University, Xuzhou, China
3 shared publications - Yuanhui Yu
School of Computer Engineering, Jimei University, Xiamen, China
3 shared publications - Xingwang Huang
Computer Engineering College, Jimei University, Xiamen, China
2 shared publications
Latest Publications
- Open Access
- 25 August 2026
- Published by Springer Nature in Journal of King Saud University Computer and Information Sciences
- Vol. 38 (7) , 768
Data security in large language models: risks, defense, and directions
- 25 August 2026
- Published by Springer Nature in Journal of King Saud University Computer and Information Sciences
- Vol. 38 (7) , 768
Open AccessData security in large language models: risks, defense, and directions
- 29 November 2025
- Published by Elsevier in Neural Networks
- Vol. 196, 108388
Attack and defense techniques in large language models: A survey and new perspectives
- 29 November 2025
- Published by Elsevier in Neural Networks
- Vol. 196, 108388
Attack and defense techniques in large language models: A survey and new perspectives
- Open Access
- 14 October 2025
- Published by Springer Nature in Journal of King Saud University Computer and Information Sciences
- Vol. 37 (8) , 1-13
Attention-based federated learning for multi-dimensional personalized recommendation system
- 14 October 2025
- Published by Springer Nature in Journal of King Saud University Computer and Information Sciences
- Vol. 37 (8) , 1-13
Open AccessAttention-based federated learning for multi-dimensional personalized recommendation system
- Open Access
Attack and defense techniques in large language models: A survey and new perspectives
Large Language Models (LLMs) have become central to numerous natural language processing tasks, but their vulnerabilities present significant security and ethical challenges. This systematic survey explores the evolving landscape of attack and defense techniques in LLMs. We classify attacks into adversarial prompt attack, optimized attacks, model theft, as well as attacks on application of LLMs, detailing their mechanisms and implications. Consequently, we analyze defense strategies, including prevention-based and detection-based defense methods. Although advances have been made, challenges remain to adapt to the dynamic threat landscape, balance usability with robustness, and address resource constraints in defense implementation. We highlight open problems, including the need for adaptive scalable defenses, explainable security techniques, and standardized evaluation frameworks. This survey provides actionable insights and directions for developing secure and resilient LLMs, emphasizing the importance of interdisciplinary collaboration and ethical considerations to mitigate risks in real-world applications.
Open AccessAttack and defense techniques in large language models: A survey and new perspectives
Large Language Models (LLMs) have become central to numerous natural language processing tasks, but their vulnerabilities present significant security and ethical challenges. This systematic survey explores the evolving landscape of attack and defense techniques in LLMs. We classify attacks into adversarial prompt attack, optimized attacks, model theft, as well as attacks on application of LLMs, detailing their mechanisms and implications. Consequently, we analyze defense strategies, including prevention-based and detection-based defense methods. Although advances have been made, challenges remain to adapt to the dynamic threat landscape, balance usability with robustness, and address resource constraints in defense implementation. We highlight open problems, including the need for adaptive scalable defenses, explainable security techniques, and standardized evaluation frameworks. This survey provides actionable insights and directions for developing secure and resilient LLMs, emphasizing the importance of interdisciplinary collaboration and ethical considerations to mitigate risks in real-world applications.
- Open Access
- 20 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 269-276
Color Image Scrambling Technique Based on Chinese Remainder Theorem and Arnold Transformation
Image scrambling technology plays a vital role in the fields of digital watermarking and information hiding. Although traditional image scrambling technology can preprocess images to enhance the security of algorithms, more robust security measures are needed due to the rapid development of artificial intelligence techniques. In order to further improve the security of color image encryption, a color image scrambling technique based on the Chinese remainder theorem and the Arnold transformation is proposed, which combines traditional cryptography with image scrambling techniques. The method initially applies the Arnold transformation to color images, and then encrypts the pixel values of the three channels in the transformed pixel matrix using the Chinese remainder theorem to obtain the final scrambled image. The performance of the algorithm was tested using MATLAB, and the Structural Similarity Index (SSIM) between the original image and the attacked recovered image exceeded 0.9. The correlation between adjacent pixels of the scrambled image approached 0, and the running time was only 46.7s. The algorithm provides a significantly improved and extensive key space compared to using the Chinese remainder theorem or Arnold transform alone. The experimental results show that the algorithm has good resistance to attacks, good scrambling effect and high operational efficiency. The algorithm thoroughly considers both security and operational efficiency aspects, making it suitable for watermark image preprocessing in the field of digital watermarking research, so as to further realize the copyright protection of digital products.
- 20 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 269-276
Open AccessColor Image Scrambling Technique Based on Chinese Remainder Theorem and Arnold Transformation
Image scrambling technology plays a vital role in the fields of digital watermarking and information hiding. Although traditional image scrambling technology can preprocess images to enhance the security of algorithms, more robust security measures are needed due to the rapid development of artificial intelligence techniques. In order to further improve the security of color image encryption, a color image scrambling technique based on the Chinese remainder theorem and the Arnold transformation is proposed, which combines traditional cryptography with image scrambling techniques. The method initially applies the Arnold transformation to color images, and then encrypts the pixel values of the three channels in the transformed pixel matrix using the Chinese remainder theorem to obtain the final scrambled image. The performance of the algorithm was tested using MATLAB, and the Structural Similarity Index (SSIM) between the original image and the attacked recovered image exceeded 0.9. The correlation between adjacent pixels of the scrambled image approached 0, and the running time was only 46.7s. The algorithm provides a significantly improved and extensive key space compared to using the Chinese remainder theorem or Arnold transform alone. The experimental results show that the algorithm has good resistance to attacks, good scrambling effect and high operational efficiency. The algorithm thoroughly considers both security and operational efficiency aspects, making it suitable for watermark image preprocessing in the field of digital watermarking research, so as to further realize the copyright protection of digital products.
- 2 March 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 32 (9) , 6002-6016
Detecting Steganography in JPEG Images Recompressed With the Same Quantization Matrix
JPEG steganalysis aims to detect stego JPEG images. For some robust steganography methods, in order to enhance stego images’ robustness of resisting JPEG recompression from lossy channel such SNS or photo sharing websites, steganographer may intentionally recompress cover image several times with quantization matrix of targeted channel, which thereby make it possible to transmit stego data in such channel for better disguise. In addition, there are huge number of cover JPEG images may be recompressed for various reasons, such as processing by some tools. Thus a better steganalysis method for such images is needed. In this paper, we investigate the steganalysis method for images recompressed with the same quantization matrix, namely, discriminate recompressed JPEG cover images and its stego images. We present some observed phenomenon on recompressed JPEG images, and design methods to enhance the sensitivity of feature based and deep model based steganalysis methods for this task. To verify their effectiveness with different acquisition of recompression prior-knowledge, we conduct experiments in various settings including conventional setting and mixing samples of different recompressing times in training. Their results demonstrate that the proposed method can notably improve detection accuracy on recompressed JPEG images.
- 2 March 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 32 (9) , 6002-6016
Detecting Steganography in JPEG Images Recompressed With the Same Quantization Matrix
JPEG steganalysis aims to detect stego JPEG images. For some robust steganography methods, in order to enhance stego images’ robustness of resisting JPEG recompression from lossy channel such SNS or photo sharing websites, steganographer may intentionally recompress cover image several times with quantization matrix of targeted channel, which thereby make it possible to transmit stego data in such channel for better disguise. In addition, there are huge number of cover JPEG images may be recompressed for various reasons, such as processing by some tools. Thus a better steganalysis method for such images is needed. In this paper, we investigate the steganalysis method for images recompressed with the same quantization matrix, namely, discriminate recompressed JPEG cover images and its stego images. We present some observed phenomenon on recompressed JPEG images, and design methods to enhance the sensitivity of feature based and deep model based steganalysis methods for this task. To verify their effectiveness with different acquisition of recompression prior-knowledge, we conduct experiments in various settings including conventional setting and mixing samples of different recompressing times in training. Their results demonstrate that the proposed method can notably improve detection accuracy on recompressed JPEG images.
- 1 November 2021
- Published by Elsevier in Journal of Visual Communication and Image Representation
- Vol. 81
Improving UNIWARD distortion function via isotropic construction and hierarchical merging
- 1 November 2021
- Published by Elsevier in Journal of Visual Communication and Image Representation
- Vol. 81
Improving UNIWARD distortion function via isotropic construction and hierarchical merging
- 1 September 2020
- Published by Elsevier in Information Sciences
- Vol. 533, 181-199
Polynomial-complexity robust deadlock controllers for a class of automated manufacturing systems with unreliable resources using Petri nets
- 1 September 2020
- Published by Elsevier in Information Sciences
- Vol. 533, 181-199
Polynomial-complexity robust deadlock controllers for a class of automated manufacturing systems with unreliable resources using Petri nets
- 1 January 2020
- Published by Inderscience Publishers in International Journal of Computational Science and Engineering
- Vol. 22 (2/3) , 270
Plaintext aware encryption in the standard model under the linear Diffie-Hellman knowledge assumption
- 1 January 2020
- Published by Inderscience Publishers in International Journal of Computational Science and Engineering
- Vol. 22 (2/3) , 270
Plaintext aware encryption in the standard model under the linear Diffie-Hellman knowledge assumption
- Open Access
- 26 December 2019
- Published by Hindawi Limited in Security and Communication Networks
- Vol. 2019, 1-17
An Improved Bidirectional Shift-Based Reversible Data Hiding Scheme Using Double-Way Prediction Strategy
Reversible data hiding (RDH) is a method that allows a cover image to be completely recovered from its corresponding stego image without distortion after the embedded secret messages have been extracted. Prediction-error expansion (PEE), as a classic RDH scheme, has been studied extensively due to its high quality of stego images. Based on prediction errors, threshold values, and the relative distances between each bin and zero bin, we present a bidirectional shift and double-way prediction strategy to solve the multiple embedding problem. Compared with the original algorithm, this scheme only takes a little more time and reduces the PSNR slightly, but it improves the embedding capacity significantly and allows for reversible data hiding. When both threshold values of and are equal to 2, the average ER value of 108 test images is 1.2 bpp which is ideal for medium data payload. At the same time, the PSNR is above 30 dB, making embedded information visually imperceptible. These data, together with other experimental results, show that the method proposed in this paper has obvious advantages in image quality and embedding capacity.- 26 December 2019
- Published by Hindawi Limited in Security and Communication Networks
- Vol. 2019, 1-17
Open AccessAn Improved Bidirectional Shift-Based Reversible Data Hiding Scheme Using Double-Way Prediction Strategy
Reversible data hiding (RDH) is a method that allows a cover image to be completely recovered from its corresponding stego image without distortion after the embedded secret messages have been extracted. Prediction-error expansion (PEE), as a classic RDH scheme, has been studied extensively due to its high quality of stego images. Based on prediction errors, threshold values, and the relative distances between each bin and zero bin, we present a bidirectional shift and double-way prediction strategy to solve the multiple embedding problem. Compared with the original algorithm, this scheme only takes a little more time and reduces the PSNR slightly, but it improves the embedding capacity significantly and allows for reversible data hiding. When both threshold values of and are equal to 2, the average ER value of 108 test images is 1.2 bpp which is ideal for medium data payload. At the same time, the PSNR is above 30 dB, making embedded information visually imperceptible. These data, together with other experimental results, show that the method proposed in this paper has obvious advantages in image quality and embedding capacity.