Shuyu Jiang

Shuyu Jiang

Auto-generated profile, based on publicly available information

Shuyu Jiang

School of Cyber Science and Engineering, Sichuan University, Chengdu, China

Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

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

Representative Publications

View All Publications
1 April 2020 in Knowledge-Based Systems
Verified
Journal Article
Closed access
RTRui Tang
Rui Tang
SJShuyu Jiang
Shuyu Jiang
XCXingshu Chen
Xingshu Chen
HWHaizhou Wang
Haizhou Wang
WWWenxian Wang
Wenxian Wang
WWWei Wang
Wei Wang

1 August 2018
Verified
Conference Paper
Closed access
SJShuyu Jiang
Shuyu Jiang
Nowadays, wireless sensor networks (WSN) technology has become a research hotspot for information gathering in the Internet of Things. In view of the limited energy of WSN and the large amount of data transmission, it is essentially important to optimize LEACH protocol for network data transmission of WSN. This paper uses ant colony algorithm (AC) and particle swarm algorithm (PSO) to optimize the LEACH protocol most commonly used in WSN routing protocol, in order to reduce the energy consumption of WSN data transmission and optimize the data transmission routing. In the clustering process of LEACH protocol, the cluster heads are randomly chosen without considering the energy consumption caused by factors such as location and data analysis, resulting in that the selected cluster heads are not the optimal one. Therefore, PSO is used to optimize LEACH protocol so as to get the most global optimal communication representative-cluster head. In the data transmission phase of LEACH protocol, a single-hop route is easy to cause a cluster head to consume all the energy too earlier, thus the network cycle is shortened. In order to solve this problem, the inter-cluster communication route is established with the help of the AC, turning the single-hop route to multi-hop route algorithm. Considering that the AC algorithm is prone to fall into a local optimal solution, PSO is used to interfere with the updated pheromone to get rid of the local optimal solution thus accelerating its acquisition of the global optimal communication path under the LEACH protocol. The simulation results show the performance of artificial intelligence (AI) of LEACH routing protocol is enhanced and its imbalance between the data transmission and network energy consumption is well solved through jointly integrated innovation method of AI algorithm.

26 September 2024 in Neural Networks
Verified
Journal Article
Closed access
XMXian Mo
Xian Mo
Jun PangJun Pang
Jun Pang
BWBinyuan Wan
Binyuan Wan
RTRui Tang
Rui Tang
Hao LiuHao Liu
Hao Liu
SJShuyu Jiang
Shuyu Jiang

20 December 2024 in Neurocomputing
Verified
Journal Article
Closed access
SJShuyu Jiang
Shuyu Jiang
WTWenyi Tang
Wenyi Tang
XCXinyu Chen
Xinyu Chen
RTRui Tang
Rui Tang
HWHaizhou Wang
Haizhou Wang
WWWenxian Wang
Wenxian Wang

13 February 2025 in ACM Transactions on Autonomous and Adaptive Systems
Verified
Journal Article
Open access
SJShuyu Jiang
Shuyu Jiang
XCXinyu Chen
Xinyu Chen
RTRui Tang
Rui Tang
Recently, large language models (LLMs) have demonstrated promising applications in the autonomous driving (AD) domain, including language-based interactions and decision-making. Ensuring they safely handle harmful inputs is crucial before formal deployment. However, research reveals emerging hand-crafted jailbreak attacks, which pack harmful prompts into harmless instructions, can bypass LLMs’ security mechanisms and elicit harmful responses. To deeply understand such jailbreaks, this paper introduces a Compositional Instruction Attack (CIA) framework to generalize them, and develop two CIA jailbreaking methods to automatically generate tailored jailbreak prompts for each harmful prompt. Then, this paper builds the first CIA question-answering (CIAQA) dataset with 2.7K multiple-choice questions of 900 successful jailbreaks, for assessing LLMs’ ability to identify underlying harmful intents, harmfulness, and task priority in CIA jailbreaks. Combined with experimental analysis on CIAQA and other datasets, this paper concludes three possible reasons for the failure of LLM defenses against CIAs. Finally, we propose an intent-based defense paradigm (IBD), enabling LLMs to defend against CIA by leveraging its capability to identify intents. Experimental results show CIA can achieve attack success rates (ASR) of 95%+ and 85%+ in AD and common harmful scenarios for three well-known LLMs (GPT-4, GPT-3.5, and Llama2-70b-chat), and IBD reduces ASR by 74%+.

3 January 2025 in IEEE Internet of Things Journal
Verified
Journal Article
Closed access
SJShuyu Jiang
Shuyu Jiang
XCXinyu Chen
Xinyu Chen
KXKaiyu Xu
Kaiyu Xu
LCLiangguo Chen
Liangguo Chen
HRHao Ren
Hao Ren
RTRui Tang
Rui Tang
Large language models (LLMs) can transform natural language instructions into executable commands for IoT devices like unmanned aerial vehicles (UAVs), creating new development opportunities. However, safety concerns about LLMs translating commands into machine or program control instructions cannot be overlooked. Currently, jailbreak instructions used to test the LLM security are often restricted to specific modes or tasks, resulting in a lack of diversity and leaving some tasks unexplored. To address this issue, we introduce a Multi-Instruction Fusion (MIF) method that can automatically fuse harmful prompts and various task instructions into jailbreaks. Firstly, we adopt a reverse decomposition strategy to acquire sufficient supervised data for fusing harmful prompts and instructions into jailbreaks and construct a task instruction synthesizer based on it. Then, to determine the optimal instruction combinations in the vast combination space, we propose a representative-node-based selection strategy, ReNB, to rank and filter the instruction combinations on a few representative samples, thereby accelerating the identification of the valid ones. Experimental results demonstrate that MIF significantly improves the attack success rate, achieving over 90% on GPT-4o-mini, LLaMa2-70B and Qwen2-7B models, outperforming the state-of-the-art baselines.